Machine Learning Interview Questions &amp; Answers

Show answer

Machine learning is a way of building programs that learn patterns from data instead of following hand-written rules. In traditional programming you give rules and data to get answers; in ML you give data and answers to get the rules, stored as model parameters. For example, a spam filter learns which word patterns indicate spam from labelled emails rather than from a fixed keyword list.

Explain supervised, unsupervised and reinforcement learning with one example each.

Show answer

Supervised learning trains on labelled data, such as predicting house prices from past sales. Unsupervised learning finds structure in unlabelled data, such as grouping customers into segments with k-means. Reinforcement learning trains an agent through rewards and penalties from an environment, such as a program learning to play a game. Semi-supervised and self-supervised learning sit between these when labels are scarce.

What is the difference between a regression and a classification problem?

Show answer

Regression predicts a continuous number, such as sales next month or delivery time in minutes. Classification predicts a discrete category, such as fraud or not fraud, or one of five product types. The choice decides the loss function and metrics: regression uses MSE, MAE or R-squared, while classification uses log loss, accuracy, precision, recall or AUC.

What are overfitting and underfitting, and how do you detect them?

Show answer

Overfitting is when a model memorises training noise and performs much worse on unseen data. Underfitting is when the model is too simple to capture the pattern, so both training and validation scores are poor. Detect them by comparing training and validation metrics or plotting learning curves. A large gap means overfitting; two low, close scores mean underfitting.

Why do we split data into training, validation and test sets?

Show answer

The training set fits model parameters, the validation set is used to choose hyperparameters and compare models, and the test set gives a final unbiased estimate of performance. If you tune on the test set, its score becomes optimistic because information has leaked into model choices. A common split is 70/15/15, or train/test plus cross-validation on the training part.

What is the difference between parameters and hyperparameters?

Show answer

Parameters are learned from data during training, such as the weights of a linear regression or split thresholds of a tree. Hyperparameters are set before training and control the learning process, such as learning rate, max_depth, number of trees or regularisation strength C. Hyperparameters are tuned with validation data using grid search, random search or Bayesian optimisation.

How do you set up a basic train-test split with scikit-learn?

Show answer

Use train_test_split with a fixed random_state for reproducibility and stratify for classification so class ratios match. from sklearn.model_selection import train_test_split X_train, X_test, y_train, y_test = train_test_split( X, y, test_size=0.2, random_state=42, stratify=y ) For time series, avoid random splits and use the most recent period as the test set.

Write a minimal scikit-learn workflow to train and evaluate a model.

Show answer

The fit, predict, score pattern is the same for almost every estimator. from sklearn.linear_model import LogisticRegression from sklearn.metrics import accuracy_score model = LogisticRegression(max_iter=1000) model.fit(X_train, y_train) preds = model.predict(X_test) print(accuracy_score(y_test, preds)) Use predict_proba when you need probabilities for thresholds or ranking.

How would you build a quick baseline before trying complex models?

Show answer

Use a DummyClassifier or DummyRegressor to see the score of a trivial guess, then a simple model. from sklearn.dummy import DummyClassifier base = DummyClassifier(strategy='most_frequent') base.fit(X_train, y_train) print(base.score(X_test, y_test)) Any real model must clearly beat this number; otherwise the features carry little signal.

Why do you want to work as a machine learning engineer?

Show answer

I enjoy turning data into decisions that people actually use. During my projects I liked the full cycle: framing the problem, cleaning messy data, comparing models and explaining results. ML engineering lets me combine coding with statistics and see measurable business impact, such as reducing manual review time. I also want to keep growing into deployment and monitoring, which I am already practising with FastAPI and Docker.

What is the bias-variance trade-off?

Show answer

Bias is error from overly simple assumptions, which causes underfitting. Variance is error from sensitivity to the particular training sample, which causes overfitting. Increasing model complexity usually lowers bias but raises variance. The goal is the complexity that minimises total test error. For example, a linear model on curved data has high bias, while a very deep unpruned tree has high variance.

What is a loss function? Name common ones for regression and classification.

Show answer

A loss function measures how wrong a prediction is for one example, and training minimises its average over the data. Regression commonly uses mean squared error, mean absolute error or Huber loss, which is robust to outliers. Classification uses log loss, also called cross-entropy, and hinge loss for SVMs. The choice should reflect the business cost of errors.

Explain gradient descent and the role of the learning rate.

Show answer

Gradient descent minimises a loss by repeatedly moving parameters in the direction of the negative gradient: w = w - learning_rate * gradient. The learning rate controls step size. Too high and the loss oscillates or diverges; too low and training is very slow or gets stuck. Variants include batch, stochastic and mini-batch gradient descent, plus adaptive optimisers like Adam.

What is the difference between batch, stochastic and mini-batch gradient descent?

Show answer

Batch gradient descent uses the full dataset for each update, giving stable but slow and memory-heavy steps. Stochastic gradient descent updates after every single example, which is fast and noisy and can escape shallow minima. Mini-batch uses small groups such as 32 to 512 rows, balancing stability with speed and allowing vectorised computation. Mini-batch is the default in practice.

What is the curse of dimensionality?

Show answer

As the number of features grows, data becomes sparse in the feature space, distances between points become similar, and the amount of data needed to generalise grows rapidly. Distance-based methods like k-NN and k-means suffer most. Remedies include feature selection, dimensionality reduction such as PCA, regularisation and collecting more data. Text and genomics data often face this problem.

Your model shows 99 percent training accuracy but 72 percent validation accuracy. What will you do?

Show answer

This is overfitting. I would first check for leakage or a split mismatch, then simplify the model: add regularisation, reduce depth or features, use early stopping, or prune trees. I would gather more data or use augmentation if possible and use cross-validation to confirm the fix. Plotting learning curves shows whether more data would actually help.

A business stakeholder asks you to use deep learning for a 5,000-row tabular dataset. How do you respond?

Show answer

I would explain that for small tabular data, gradient-boosted trees or regularised linear models usually match or beat neural networks, train faster and are easier to explain. I would propose building a simple baseline, then a boosting model, and only test a neural network if there is a clear reason. I would show the comparison on cross-validated metrics so the decision is data-driven.

You are asked to predict customer churn. How would you frame the ML problem from scratch?

Show answer

First define churn precisely, for example no purchase in 60 days, and the prediction window. It becomes binary classification. Build features from data available before the cut-off date only, choose a time-based split, and pick metrics like recall at a fixed precision or lift in the top decile. Start with logistic regression as a baseline, then try boosting, and agree on how retention teams will use the scores.

Leadership wants a model in one week, but data quality is poor. How do you handle it?

Show answer

I would agree on a narrow first version: a simple, explainable baseline built on the cleanest available fields, with clear notes on data gaps and expected accuracy. In parallel I would log specific data issues and their impact. Delivering a working baseline with an honest limitations section builds trust and gives a reference point for improvement in later iterations.

How do you implement simple gradient descent for linear regression in NumPy?

Show answer

Compute predictions, the gradient of MSE, and update weights in a loop. import numpy as np w = np.zeros(X.shape[1]); b = 0.0; lr = 0.01 for _ in range(1000): err = X @ w + b - y w -= lr * (2 / len(y)) * (X.T @ err) b -= lr * (2 / len(y)) * err.sum() Scale features first or the loop converges slowly.

How do you make an ML experiment reproducible?

Show answer

Fix random seeds in NumPy, Python and the model, pin library versions in requirements.txt, version the data or record its hash, and save the exact code commit and hyperparameters. Tools like MLflow log parameters, metrics and artifacts per run. Use pipelines so preprocessing is identical each time, and keep the train-test split stored rather than recreated randomly.

Tell me about a time your model did not work as expected. What did you do?

Show answer

Situation: my first churn model scored 0.95 AUC in validation but failed in a pilot. Task: find out why. Action: I traced features and found one field updated after customers left, which leaked the label. I removed it, rebuilt features with a strict cut-off date and moved to a time-based split. Result: AUC dropped to 0.78 but held steady in production, and I added a leakage checklist to my workflow.

What is the no free lunch theorem and what does it mean in practice?

ML FundamentalsHard

Show answer

It states that no single algorithm is best for every problem when averaged over all possible problems. In practice it means you should try several model families, start with a simple baseline, and choose based on validation results for your data. For tabular data gradient boosting often wins, but a logistic regression may be preferred if interpretability matters more.

What is the difference between a generative and a discriminative model?

ML FundamentalsHard

Show answer

A discriminative model learns the decision boundary or P(y|x) directly, such as logistic regression or SVM. A generative model learns how data is produced, P(x|y) and P(y), and can generate new samples, such as Naive Bayes or Gaussian mixture models. Discriminative models usually give better accuracy with enough data, while generative models handle missing data and small samples better.

Your training loss is not decreasing at all. What could be wrong?

ML FundamentalsHard

Show answer

Common causes are a learning rate that is too high or too low, unscaled features, a bug in labels or feature alignment, wrong loss for the task, or data that has been shuffled away from its labels. I would try overfitting a tiny subset first; if the model cannot even fit 50 rows, there is a bug. Then check scaling, learning rate and gradient values.

Why is feature scaling needed and which algorithms need it?

Show answer

Scaling puts features on comparable ranges so that one large-valued feature does not dominate distances or gradients. Distance and gradient based models need it: k-NN, k-means, SVM, PCA, linear and logistic regression with regularisation, and neural networks. Tree-based models like random forest and XGBoost split on thresholds, so they are largely unaffected by scaling.

What is the difference between standardisation and normalisation?

Show answer

Standardisation (StandardScaler) transforms a feature to mean 0 and standard deviation 1 using (x - mean) / std; it keeps outliers but does not bound values. Min-max normalisation (MinMaxScaler) rescales to a fixed range such as 0 to 1 and is sensitive to outliers. RobustScaler uses median and IQR and is better when outliers are present. Always fit the scaler on training data only.

How do you handle missing values in a dataset?

Show answer

First understand why values are missing: completely at random, at random, or not at random. Options are dropping rows or columns with heavy missingness, simple imputation with mean, median or mode, model-based imputation like KNNImputer or IterativeImputer, or a separate 'missing' category. Adding a missing-indicator column often helps because missingness itself can be predictive, such as income left blank.

What is one-hot encoding and when can it cause problems?

Show answer

One-hot encoding turns a categorical column into binary columns, one per category, so models do not assume order. It causes problems with high-cardinality features like city or product ID, creating thousands of sparse columns and higher memory and overfitting risk. Alternatives are target encoding, frequency encoding, hashing, or grouping rare categories into 'Other'. Use handle_unknown='ignore' for unseen categories.

What is feature engineering and why does it matter?

Show answer

Feature engineering is creating input variables that make patterns easier for a model to learn, using domain knowledge. Examples are extracting day of week from a date, ratios like debt-to-income, rolling averages of past purchases, or text length. On tabular problems good features often improve results more than switching algorithms. Features must use only information available at prediction time.

How do you extract useful features from a datetime column in pandas?

Show answer

Convert to datetime and pull out calendar parts. import pandas as pd df['ts'] = pd.to_datetime(df['ts']) df['hour'] = df['ts'].dt.hour df['dow'] = df['ts'].dt.dayofweek df['month'] = df['ts'].dt.month df['is_weekend'] = df['dow'].isin([5, 6]).astype(int) For cyclical features like hour, sine and cosine encoding keeps 23 and 0 close together.

How do you find and remove duplicate rows and check missing values in pandas?

Show answer

Quick data-quality checks before modelling: print(df.isna().sum().sort_values(ascending=False)) print(df.duplicated().sum()) df = df.drop_duplicates() df['age'] = df['age'].fillna(df['age'].median()) For modelling, do the median imputation inside a pipeline instead so the value is learned only from training data.

How do you explain data cleaning decisions to a non-technical manager?

Show answer

I avoid jargon and tie each decision to its effect on business results. For example: 'About 8 percent of orders had missing delivery times. Instead of deleting them, which would hide slow regions, I filled them using similar orders and flagged them, so the model still learns from those regions.' I keep a short log of decisions so they can be reviewed later and repeated.

What is the difference between label encoding and ordinal encoding?

Show answer

Both map categories to integers. In scikit-learn, LabelEncoder is meant for the target variable, while OrdinalEncoder is for input features. Integer encoding implies an order, so it is fine for truly ordered values like low, medium, high, or for tree models, but it can mislead linear models on nominal data like colours. For nominal features with linear models, prefer one-hot encoding.

What is data leakage? Give two examples.

Show answer

Data leakage is when information not available at prediction time gets into training, giving falsely high validation scores. Example one: fitting a scaler or imputer on the full dataset before splitting. Example two: using a feature created after the event, such as 'refund issued' when predicting fraud. Prevent it with pipelines, time-aware splits and checking each feature's timestamp.

How do you detect and treat outliers?

Show answer

Detect with box plots, the IQR rule (below Q1 - 1.5 IQR or above Q3 + 1.5 IQR), z-scores above 3, or models like Isolation Forest. Treatment depends on cause: fix data entry errors, cap values (winsorise), apply log transforms for skewed data, or use robust models and losses. Genuine extreme values, such as large fraud amounts, may be the most important signal and should not be removed blindly.

Compare filter, wrapper and embedded feature selection methods.

Show answer

Filter methods score features independently of a model using statistics such as correlation, chi-square or mutual information; they are fast but ignore interactions. Wrapper methods like Recursive Feature Elimination train models on feature subsets; they are accurate but slow. Embedded methods select during training, such as L1 regularisation or tree feature importances. Embedded methods are a common practical middle ground.

What is multicollinearity and how do you detect it?

Show answer

Multicollinearity is high correlation between input features, which makes linear model coefficients unstable and hard to interpret, though predictions may remain fine. Detect it with a correlation matrix or the Variance Inflation Factor, where VIF above 5 to 10 signals a problem. Fix it by dropping one of the correlated features, combining them, using PCA, or applying Ridge regularisation.

Why do skewed features often need a log or power transform?

Show answer

Heavily right-skewed features like income or transaction amount have long tails that can dominate linear models and distance metrics. A log1p, Box-Cox or Yeo-Johnson transform compresses large values and makes the distribution closer to normal, which often improves linear and distance-based models. Yeo-Johnson works with zero and negative values. Tree models usually do not need these transforms.

Your dataset has a city column with 2,000 unique values. How will you encode it?

Show answer

One-hot would create 2,000 sparse columns, so I would first group rare cities below a frequency threshold into 'Other'. Then options are frequency encoding, out-of-fold target encoding, or mapping cities to useful attributes such as tier, state or population. For boosting models like CatBoost or LightGBM, native categorical handling also works well. I would compare options with cross-validation.

Validation AUC is 0.99 on a loan default model. What do you suspect?

Show answer

Such a high score on a hard problem usually signals leakage. I would check feature importances for a suspiciously dominant feature, look for fields recorded after the loan outcome such as collection status, verify preprocessing was fit only on training data, and check for duplicate customers across train and test. Then re-run with a time-based split to confirm a realistic score.

40 percent of values in an important income column are missing. What do you do?

Show answer

I would study the missingness pattern first, for example whether self-employed applicants skip it. I would add a binary was_missing feature, impute with the median within relevant groups or use IterativeImputer, and compare against models that handle missing values natively, like HistGradientBoosting or XGBoost. I would also raise the issue with the data owner so collection can improve.

How do you build a scikit-learn pipeline for mixed numeric and categorical data?

Show answer

Use ColumnTransformer inside a Pipeline so preprocessing is fit only on training folds. from sklearn.compose import ColumnTransformer from sklearn.pipeline import Pipeline from sklearn.impute import SimpleImputer from sklearn.preprocessing import StandardScaler, OneHotEncoder num = Pipeline([('imp', SimpleImputer(strategy='median')), ('sc', StandardScaler())]) pre = ColumnTransformer([ ('num', num, num_cols), ('cat', OneHotEncoder(handle_unknown='ignore'), cat_cols) ]) model = Pipeline([('pre', pre), ('clf', LogisticRegression(max_iter=1000))])

How do you calculate VIF for features in Python?

Show answer

Use statsmodels on the numeric feature matrix with a constant added. import pandas as pd from statsmodels.stats.outliers_influence import variance_inflation_factor from statsmodels.tools import add_constant Xc = add_constant(X_num) vif = pd.Series([variance_inflation_factor(Xc.values, i) for i in range(Xc.shape[1])], index=Xc.columns) print(vif.drop('const')) Values above 5 to 10 indicate strong multicollinearity.

How do you select the top k features using mutual information in scikit-learn?

Show answer

SelectKBest with mutual_info_classif captures non-linear relationships, unlike plain correlation. from sklearn.feature_selection import SelectKBest, mutual_info_classif sel = SelectKBest(score_func=mutual_info_classif, k=20) X_new = sel.fit_transform(X_train, y_train) print(X_train.columns[sel.get_support()]) Put the selector inside a pipeline so selection is redone in each CV fold.

Describe a time you found a serious data quality issue late in a project.

Show answer

Situation: two days before a demo, I noticed sales for one region were doubled. Task: fix it without missing the deadline. Action: I traced it to a join that duplicated rows on a non-unique key, fixed the join, added row-count assertions to the pipeline and retrained. I informed my lead immediately with the impact. Result: the demo went ahead with correct numbers, and the assertions caught a similar issue later.

How would you encode cyclical features like hour of day or month?

Data Preprocessing and Feature EngineeringHard

Show answer

Map the value onto a circle with sine and cosine so the end connects to the start. import numpy as np df['hour_sin'] = np.sin(2 * np.pi * df['hour'] / 24) df['hour_cos'] = np.cos(2 * np.pi * df['hour'] / 24) Now 23:00 and 00:00 are close, which linear and distance-based models cannot see with raw integers. Tree models often manage with the raw value.

What is target encoding and how do you avoid leakage with it?

Show answer

Target encoding replaces each category with the mean target value for that category, which works well for high-cardinality features. It leaks if computed on the full data because each row sees its own label. Avoid this by computing encodings out-of-fold within cross-validation and adding smoothing toward the global mean for rare categories. scikit-learn 1.3+ has TargetEncoder with built-in cross-fitting.

New categories appear in production that were not in training data. How do you handle it?

Data Preprocessing and Feature EngineeringHard

Show answer

Make the encoder tolerant: OneHotEncoder with handle_unknown='ignore' or infrequent_if_exist, or map unseen values to an 'Other' bucket. Monitor the share of unseen categories as a drift signal. If new categories become common, such as a new city launch, retrain with fresh data. Tests in the inference pipeline should include unseen values so the service never crashes.

Your preprocessing works in notebooks but gives different results in production. Why?

Data Preprocessing and Feature EngineeringHard

Show answer

Common causes are preprocessing written separately in two places, statistics like means recomputed on production data instead of using training values, different library versions, or different column order and data types. The fix is to save one fitted Pipeline with preprocessing and model together, pin versions, validate input schema, and test that the saved pipeline gives identical outputs on sample rows.

What are the assumptions of linear regression?

Show answer

Linear regression assumes a linear relationship between features and target, independent errors, constant error variance (homoscedasticity), normally distributed errors for valid inference, and little multicollinearity among features. Violations mainly hurt confidence intervals and coefficient interpretation more than raw predictions. Check them with residual plots, Q-Q plots, the Durbin-Watson test and VIF.

What is R-squared and what is adjusted R-squared?

Show answer

R-squared is the share of variance in the target explained by the model, 1 - SS_res / SS_tot. It never decreases when you add features, even useless ones. Adjusted R-squared penalises extra features: 1 - (1 - R2)(n - 1)/(n - p - 1), so it only rises if a new feature improves fit more than expected by chance. R-squared can be negative on test data if the model is worse than predicting the mean.

Compare MAE, MSE and RMSE as regression metrics.

Show answer

MAE is the average absolute error, easy to explain and robust to outliers. MSE averages squared errors and punishes large errors heavily, which suits cases where big misses are costly. RMSE is the square root of MSE, so it is in the same unit as the target. If RMSE is much higher than MAE, a few large errors are present. MAPE is useful for percentage errors but fails near zero.

How do you interpret a coefficient in a multiple linear regression?

Show answer

A coefficient is the expected change in the target for a one-unit increase in that feature, holding other features constant. For example, a coefficient of 2,500 on 'area in square metres' means each extra square metre adds about 2,500 to price. Interpretation breaks down with multicollinearity or when features are scaled, so report units clearly and check stability.

Fit a linear regression in scikit-learn and print RMSE and R-squared.

Show answer

from sklearn.linear_model import LinearRegression from sklearn.metrics import mean_squared_error, r2_score import numpy as np lr = LinearRegression().fit(X_train, y_train) pred = lr.predict(X_test) print('RMSE', np.sqrt(mean_squared_error(y_test, pred))) print('R2', r2_score(y_test, pred)) print(dict(zip(X_train.columns, lr.coef_))) Newer scikit-learn versions also provide root_mean_squared_error directly.

How do you check regression residuals visually in Python?

Show answer

import matplotlib.pyplot as plt res = y_test - pred fig, ax = plt.subplots(1, 2, figsize=(10, 4)) ax[0].scatter(pred, res, s=8); ax[0].axhline(0, color='red') ax[0].set_xlabel('Predicted'); ax[0].set_ylabel('Residual') ax[1].hist(res, bins=40) plt.show() A funnel shape suggests heteroscedasticity; a curve suggests a missing non-linear term.

How does ordinary least squares find the regression coefficients?

Show answer

OLS chooses coefficients that minimise the sum of squared residuals between actual and predicted values. It has a closed-form solution, beta = (X^T X)^-1 X^T y, when X^T X is invertible. For very large or ill-conditioned data, gradient descent or a pseudo-inverse is used instead. Squaring errors penalises large mistakes heavily, which makes OLS sensitive to outliers.

What is the difference between Ridge, Lasso and Elastic Net regression?

Show answer

All add a penalty on coefficient size to reduce overfitting. Ridge uses an L2 penalty (sum of squares), shrinking coefficients toward zero but rarely to zero, and handles multicollinearity well. Lasso uses L1 (sum of absolute values) and can set coefficients exactly to zero, performing feature selection. Elastic Net mixes both, which is useful when there are groups of correlated features.

What is polynomial regression and what risk does it carry?

Show answer

Polynomial regression adds powers and interaction terms of features, such as x squared and x1 times x2, so a linear model can fit curves. It is still linear in its coefficients. The main risk is overfitting at high degrees, plus wild behaviour outside the training range. Use cross-validation to choose the degree, scale features, and combine it with Ridge regularisation.

What is heteroscedasticity and why does it matter?

Show answer

Heteroscedasticity means the variance of residuals changes with the level of the prediction, such as errors growing for expensive houses. Predictions remain unbiased, but standard errors and p-values become unreliable. Detect it with a residuals versus fitted plot or the Breusch-Pagan test. Fixes include log-transforming the target, weighted least squares, or robust standard errors.

How do you approach a time series forecasting problem differently from normal regression?

Show answer

Time series data is ordered and autocorrelated, so random splits leak future information. I use time-based splits or TimeSeriesSplit, create lag and rolling-window features from past values only, and handle trend and seasonality. Baselines like last value or seasonal naive come first. Models range from ARIMA and Prophet to gradient boosting on lag features.

Your house price model's residuals grow larger for expensive houses. What do you do?

Show answer

That is heteroscedasticity, typical for price data that is right-skewed. I would model log(price) instead of price, which turns absolute errors into roughly percentage errors and stabilises variance. Then I would convert predictions back with exp and check residual plots again. Evaluating with RMSLE or MAPE also fits better when relative error matters more than absolute error.

A demand forecast is consistently under-predicting during festivals. How would you fix it?

Show answer

The model lacks festival information. I would add features for festival dates such as Diwali or regional holidays, days-to-festival, and past festival uplift for the same item. If the business cost of stock-outs is higher than over-stock, I would use a quantile loss for a higher percentile. Then validate specifically on past festival periods, not just the average error.

Two models have RMSE 120 and 125, but the second has much lower MAE. Which do you choose?

Show answer

It depends on error cost. The first model makes fewer big errors, while the second is better on typical cases but has some large misses. If large errors are expensive, such as stock-outs of costly items, choose the first. If typical accuracy matters more, choose the second. I would also check error distributions by segment and discuss with stakeholders before deciding.

How do you tune the alpha of Ridge regression using cross-validation?

Show answer

RidgeCV tests many alphas efficiently; scale features first. from sklearn.linear_model import RidgeCV from sklearn.pipeline import make_pipeline from sklearn.preprocessing import StandardScaler import numpy as np m = make_pipeline(StandardScaler(), RidgeCV(alphas=np.logspace(-3, 3, 30))) m.fit(X_train, y_train) print(m[-1].alpha_) LassoCV and ElasticNetCV work the same way.

How do you create polynomial features with regularisation in a pipeline?

Show answer

from sklearn.preprocessing import PolynomialFeatures, StandardScaler from sklearn.linear_model import Ridge from sklearn.pipeline import make_pipeline model = make_pipeline( PolynomialFeatures(degree=2, include_bias=False), StandardScaler(), Ridge(alpha=1.0) ) model.fit(X_train, y_train) Tune degree and alpha together with GridSearchCV to avoid overfitting.

Use statsmodels to get p-values for regression coefficients.

Show answer

import statsmodels.api as sm Xc = sm.add_constant(X_train) ols = sm.OLS(y_train, Xc).fit() print(ols.summary()) The summary shows coefficients, standard errors, p-values, confidence intervals, R-squared and diagnostics. scikit-learn focuses on prediction and does not report p-values, so statsmodels is preferred for inference.

Tell me about a time you had to explain a model's error to a business team.

Show answer

Situation: a sales forecast missed by 18 percent in one month. Task: explain why to the category team. Action: I broke the error down by region and found most came from a sudden promotion not in our data. I showed a simple chart and proposed adding a promotion calendar feed. Result: the team shared promotion plans in advance, and the next quarter's error fell below 8 percent.

How do you handle disagreement with a senior colleague about model choice?

Show answer

I keep it about evidence. In one project a senior preferred a complex model while I proposed regularised regression. I suggested we compare both on the same cross-validation folds and also measure inference time and explainability. The complex model was only 1 percent better, so we chose the simpler one. I made sure to credit his idea of the extra features, which we kept.

What is regularisation in general and why does it reduce overfitting?

Show answer

Regularisation adds a constraint or penalty that discourages overly complex models, such as large weights in linear models, deep trees or too many boosting rounds. It trades a little extra bias for a larger drop in variance, so test error falls. Examples include L1 and L2 penalties, dropout in neural networks, early stopping, pruning and limiting tree depth.

Why does L1 regularisation produce sparse models while L2 does not?

Show answer

The L1 constraint region is a diamond with corners on the axes, so the optimal point often lands on a corner where some coefficients are exactly zero. The L2 region is a circle with no corners, so coefficients shrink smoothly but stay non-zero. In gradient terms, L1 applies a constant push toward zero regardless of size, while L2's push becomes tiny as weights get small.

When would you use Huber or quantile regression instead of OLS?

Show answer

Huber regression uses squared loss for small errors and absolute loss for large ones, so it resists outliers while staying smooth. Quantile regression predicts a chosen percentile, such as the 90th percentile of delivery time, which is useful for setting safe promises or prediction intervals. Gradient boosting libraries support both losses for non-linear versions.

Lasso dropped a feature the business considers very important. How do you handle it?

Show answer

Lasso often keeps only one of a group of correlated features, so the dropped feature may be redundant rather than useless. I would check correlations, try Elastic Net which keeps correlated groups, and compare cross-validated performance with the feature forced in. I would explain the trade-off clearly; if the accuracy difference is small, keeping the feature for trust and interpretability is reasonable.

You need prediction intervals, not just point estimates, for delivery time. How?

Show answer

Options are quantile regression, training separate models for the 10th and 90th percentiles with GradientBoostingRegressor(loss='quantile') or LightGBM; conformal prediction, which wraps any model and gives intervals with guaranteed coverage; or bootstrapping. I would check calibration: about 80 percent of actual values should fall within an 80 percent interval on a holdout set.

How do you create lag and rolling features for a sales forecasting model in pandas?

Show answer

Sort by item and date, then shift so features use only past values. df = df.sort_values(['item', 'date']) g = df.groupby('item')['sales'] df['lag_7'] = g.shift(7) df['roll_mean_28'] = g.transform(lambda s: s.shift(1).rolling(28).mean()) The shift before rolling prevents the current day's sales from leaking into its own feature.

How does logistic regression work?

Show answer

Logistic regression computes a weighted sum of features and passes it through the sigmoid function, 1 / (1 + e^-z), to get a probability between 0 and 1. It is trained by minimising log loss. Coefficients represent change in log-odds per unit of a feature, and exp(coefficient) gives the odds ratio. A threshold, 0.5 by default, converts probabilities into class labels.

Why is logistic regression called regression if it is used for classification?

Show answer

It models a continuous quantity, the log-odds of the positive class, as a linear function of features, which is regression in the statistical sense. The classification comes only when you apply a threshold to the predicted probability. This is why it outputs well-calibrated probabilities more naturally than many other classifiers.

Explain the k-nearest neighbours algorithm and its weaknesses.

Show answer

k-NN predicts by finding the k closest training points and taking a majority vote for classification or the average for regression. It has no training phase but is slow at prediction on large data, needs feature scaling, suffers from the curse of dimensionality and is sensitive to irrelevant features. Small k overfits, large k underfits; choose k with cross-validation.

What is Naive Bayes and why is it called naive?

Show answer

Naive Bayes applies Bayes' theorem to compute the probability of each class given the features, assuming features are conditionally independent given the class. That independence assumption is rarely true, hence 'naive', but the model still works well for text classification like spam detection. Variants include Gaussian, Multinomial for word counts and Bernoulli for binary features. It is fast and needs little data.

What is a confusion matrix?

Show answer

A confusion matrix is a table of actual versus predicted classes. For binary problems it has true positives, false positives, true negatives and false negatives. From it you derive precision, recall, specificity and accuracy. For example, in fraud detection a false negative is a missed fraud and a false positive is a genuine customer blocked, and these usually have very different costs.

Explain precision, recall and F1 score.

Show answer

Precision is TP / (TP + FP): of the cases flagged positive, how many were right. Recall is TP / (TP + FN): of the actual positives, how many were found. F1 is the harmonic mean of the two and is low if either is low. Use precision when false alarms are costly, such as spam filters; use recall when misses are costly, such as cancer screening or fraud.

A fraud model has 99.6 percent accuracy, but the business says it is useless. Why?

Show answer

If only 0.4 percent of transactions are fraud, predicting 'not fraud' for everything gives 99.6 percent accuracy and catches nothing. I would switch to recall, precision and PR-AUC, check the confusion matrix, then retrain with class weights or resampling and tune the threshold so the fraud team gets a manageable number of alerts with high recall.

You must build a spam classifier quickly with limited compute. What do you choose?

Show answer

I would use TF-IDF features with a Multinomial Naive Bayes or linear model like logistic regression or LinearSVC. They train in seconds, handle high-dimensional sparse text well and give strong baselines. I would evaluate precision carefully because blocking genuine mail is costly. Only if this is not good enough would I try fine-tuning a small transformer model.

Print a confusion matrix and classification report in scikit-learn.

Show answer

from sklearn.metrics import confusion_matrix, classification_report pred = clf.predict(X_test) print(confusion_matrix(y_test, pred)) print(classification_report(y_test, pred, digits=3)) The report shows precision, recall, F1 and support per class plus macro and weighted averages. ConfusionMatrixDisplay.from_estimator draws it as a plot.

How do you plot an ROC curve and get AUC in scikit-learn?

Show answer

from sklearn.metrics import roc_auc_score, RocCurveDisplay import matplotlib.pyplot as plt proba = clf.predict_proba(X_test)[:, 1] print('AUC', roc_auc_score(y_test, proba)) RocCurveDisplay.from_predictions(y_test, proba) plt.show() Pass probabilities or decision scores, not hard labels, otherwise the curve has only one point.

How do you keep learning new ML techniques while working full time?

Show answer

I set a small weekly routine: one paper summary or library release note, and one hands-on experiment on a public dataset. I follow scikit-learn and LightGBM release notes, and I rebuild ideas on a small problem before suggesting them at work. Recently I tried TunedThresholdClassifierCV and later used it to replace a manual threshold script in our project.

How does a support vector machine work?

Show answer

An SVM finds the hyperplane that separates classes with the maximum margin, and the points on the margin are the support vectors. The soft-margin version uses C to trade off margin width against misclassification. Kernels like RBF or polynomial let SVMs learn non-linear boundaries by implicitly mapping data to higher dimensions. SVMs need scaled features and become slow beyond roughly a hundred thousand rows.

What is the ROC curve and AUC? When is PR-AUC better?

Show answer

The ROC curve plots true positive rate against false positive rate at every threshold, and AUC is the probability that a random positive is ranked above a random negative. It is threshold-independent. On heavily imbalanced data, like 0.5 percent fraud, ROC-AUC can look high even when precision is poor, so the precision-recall curve and PR-AUC give a more honest picture.

How do you handle class imbalance?

Show answer

Start with the right metrics, such as PR-AUC, recall at fixed precision or F1, not accuracy. Then use class_weight='balanced', resampling like random undersampling or SMOTE applied only on training folds, and threshold tuning based on business cost. Boosting models with scale_pos_weight also help. Stratified splits keep class ratios consistent across folds.

How do one-vs-rest and multinomial approaches differ for multiclass classification?

Show answer

One-vs-rest trains one binary classifier per class against all others and picks the highest score; it is simple but probabilities may not sum to one. Multinomial (softmax) logistic regression models all classes jointly with a single loss, giving consistent probabilities. For metrics, use macro averaging to treat classes equally or weighted averaging to account for class frequency.

The fraud team can only review 500 alerts a day. How do you set the threshold?

Show answer

Instead of 0.5, rank transactions by predicted probability and pick the threshold that yields about 500 alerts per day on recent data. Then report precision and recall at that cut-off, or precision at k. I would monitor daily alert volume because drift can change it, and recompute the threshold periodically. If cost data exists, I would choose the threshold that maximises expected savings.

How do you choose a classification threshold that maximises F1?

Show answer

from sklearn.metrics import precision_recall_curve import numpy as np proba = clf.predict_proba(X_val)[:, 1] p, r, t = precision_recall_curve(y_val, proba) f1 = 2 * p * r / (p + r + 1e-9) best = t[np.argmax(f1[:-1])] print(best) Choose it on validation data, not the test set. scikit-learn 1.5+ also offers TunedThresholdClassifierCV.

How do you build a text classifier with TF-IDF and logistic regression?

Show answer

from sklearn.feature_extraction.text import TfidfVectorizer from sklearn.linear_model import LogisticRegression from sklearn.pipeline import make_pipeline clf = make_pipeline( TfidfVectorizer(ngram_range=(1, 2), min_df=2, sublinear_tf=True), LogisticRegression(max_iter=1000, class_weight='balanced') ) clf.fit(train_texts, train_labels) print(clf.score(test_texts, test_labels))

Tell me about a time you balanced model accuracy against a business constraint.

Show answer

Situation: our lead-scoring model had the best AUC with a large ensemble, but the CRM needed scores within 50 ms. Task: meet latency without losing much quality. Action: I compared a smaller LightGBM model with fewer features and measured both AUC and latency. Result: we lost 0.01 AUC but met the 50 ms limit, and sales teams adopted it because scores appeared instantly.

What is the softmax function?

Show answer

Softmax turns a vector of raw scores into probabilities that sum to one: p_i = exp(z_i) / sum of exp(z_j). It is used in multinomial logistic regression and the output layer of neural networks for multiclass problems. Larger scores get exponentially larger shares. In practice the maximum score is subtracted before exponentiation to avoid numerical overflow.

What is the kernel trick?

Show answer

The kernel trick computes dot products of data in a high-dimensional feature space without explicitly transforming the data, using a kernel function such as RBF, exp(-gamma times squared distance). This lets linear algorithms like SVM learn non-linear boundaries efficiently. In RBF SVMs, larger gamma makes each point's influence more local, which can lead to overfitting.

What is probability calibration and why does it matter?

Show answer

A calibrated model's predicted probabilities match real frequencies: of cases scored 0.8, about 80 percent should be positive. It matters when scores drive decisions like credit limits or expected loss. Random forests and SVMs are often poorly calibrated. Fix it with CalibratedClassifierCV using Platt scaling (sigmoid) or isotonic regression, and check with a reliability curve or Brier score.

Your classifier works well overall but poorly for one customer segment. What do you do?

Show answer

I would measure metrics per segment to confirm, then check if that segment is under-represented or has different feature distributions. Fixes include collecting more data, sample weighting, segment-specific features, or a separate model if the behaviour is truly different. If the segment relates to a protected attribute, I would also review fairness metrics and involve the compliance team.

A credit model's probability of 0.3 is used to set interest rates, but defaults in that band are 12 percent. What is wrong?

Show answer

The model is poorly calibrated: it ranks customers but its probabilities are not reliable. Since pricing depends on actual probabilities, I would calibrate it with isotonic regression or Platt scaling on a held-out set and validate with a reliability curve and Brier score. I would also check whether the default rate has drifted since training, which would require retraining.

100

How do you apply SMOTE correctly without leakage?

Show answer

Use an imbalanced-learn pipeline so oversampling happens only on training folds. from imblearn.pipeline import Pipeline from imblearn.over_sampling import SMOTE from sklearn.model_selection import cross_val_score pipe = Pipeline([('pre', pre), ('smote', SMOTE(random_state=42)), ('clf', LogisticRegression(max_iter=1000))]) print(cross_val_score(pipe, X, y, cv=5, scoring='average_precision').mean()) Applying SMOTE before splitting puts synthetic copies of test points into training.

101

How does a decision tree decide where to split?

Show answer

At each node the tree tries candidate features and thresholds and picks the split that most reduces impurity in the child nodes. For classification, impurity is measured by Gini or entropy; for regression, by variance or MSE. The process repeats recursively until a stopping rule such as max_depth, min_samples_leaf or pure leaves is reached. This greedy approach is fast but not globally optimal.

102

What is the difference between Gini impurity and entropy?

Show answer

Gini impurity is 1 - sum of p squared and measures the chance of misclassifying a random sample. Entropy is - sum of p log p and comes from information theory; the reduction is called information gain. Both usually produce very similar trees. Gini is slightly faster because it avoids logarithms, which is why it is the scikit-learn default.

103

Why do single decision trees overfit and how do you control it?

Show answer

An unrestricted tree keeps splitting until leaves are pure, memorising noise. Control it with pre-pruning hyperparameters like max_depth, min_samples_split, min_samples_leaf and max_leaf_nodes, or with post-pruning through cost-complexity pruning using ccp_alpha. In practice ensembles such as random forests or boosting are used to reduce the variance of single trees.

104

Explain bagging and how a random forest uses it.

Show answer

Bagging trains many models on bootstrap samples, random samples with replacement, and averages their predictions to reduce variance. A random forest adds feature randomness: each split considers only a random subset of features, controlled by max_features. This de-correlates the trees, so averaging helps more. Out-of-bag samples, about 37 percent left out per tree, give a free validation estimate.

105

Train a random forest and show feature importances in scikit-learn.

Show answer

from sklearn.ensemble import RandomForestClassifier import pandas as pd rf = RandomForestClassifier(n_estimators=300, min_samples_leaf=2, n_jobs=-1, random_state=42, oob_score=True) rf.fit(X_train, y_train) print('OOB', rf.oob_score_) imp = pd.Series(rf.feature_importances_, index=X_train.columns) print(imp.sort_values(ascending=False).head(10))

106

Visualise a small decision tree and its rules.

Show answer

from sklearn.tree import DecisionTreeClassifier, export_text, plot_tree import matplotlib.pyplot as plt dt = DecisionTreeClassifier(max_depth=3, random_state=0).fit(X_train, y_train) print(export_text(dt, feature_names=list(X_train.columns))) plot_tree(dt, feature_names=X_train.columns, filled=True) plt.show() Shallow trees are a good way to explain rules to business teams.

107

What is boosting and how is it different from bagging?

Show answer

Boosting trains models sequentially, each one focusing on the errors of the combined previous models, which mainly reduces bias. Bagging trains models independently in parallel and averages them, which mainly reduces variance. Boosting uses shallow trees and a learning rate, and can overfit if run too long, so early stopping is important. Examples are AdaBoost, Gradient Boosting, XGBoost, LightGBM and CatBoost.

108

How does gradient boosting work?

Show answer

Gradient boosting starts with a simple prediction, like the mean, then repeatedly fits a small tree to the negative gradient of the loss, which for MSE is just the residuals. Each tree's output is scaled by a learning rate and added to the ensemble. A lower learning rate with more trees usually generalises better. Any differentiable loss can be used, which makes it flexible.

109

What makes XGBoost, LightGBM and CatBoost different?

Show answer

XGBoost adds regularisation to the objective, uses second-order gradients and handles missing values natively. LightGBM grows trees leaf-wise and uses histogram binning, making it very fast on large data, though it can overfit small data. CatBoost handles categorical features with ordered target statistics and uses symmetric trees, giving strong defaults with less tuning. All three support GPU training.

110

Your random forest model file is 2 GB and too slow for real-time scoring. What can you do?

Show answer

Reduce size by limiting max_depth and min_samples_leaf, using fewer trees since gains flatten after a few hundred, and dropping weak features. Alternatively switch to a gradient boosting model like LightGBM, which usually gives similar accuracy with far smaller, faster models. Converting to ONNX or using a compiled predictor can also speed up inference. Measure accuracy and latency after each change.

111

XGBoost validation loss starts rising after 300 rounds while training loss keeps falling. What do you do?

Show answer

That is overfitting beyond about 300 rounds. I would use early stopping with a patience such as 50 rounds on a validation set so training stops at the best iteration, then lower the learning rate, reduce max_depth, add subsample and colsample_bytree below 1, and increase min_child_weight or lambda. The final model is refit using the best iteration count found.

112

Random forest and LightGBM give the same AUC. Which would you deploy?

Show answer

I would compare beyond AUC: model size, inference latency, training time, calibration and stability across time periods. LightGBM is usually smaller and faster to score, while random forest needs less tuning and is less prone to overfitting. If both are equal, I would choose the one that is cheaper to run and easier for the team to maintain and retrain.

113

Train LightGBM with early stopping.

Show answer

import lightgbm as lgb m = lgb.LGBMClassifier(n_estimators=5000, learning_rate=0.05, num_leaves=31, subsample=0.8, subsample_freq=1, colsample_bytree=0.8) m.fit(X_train, y_train, eval_set=[(X_val, y_val)], eval_metric='auc', callbacks=[lgb.early_stopping(100)]) print(m.best_iteration_) The large n_estimators is only an upper limit; early stopping picks the real number.

114

Compute permutation importance on validation data.

Show answer

from sklearn.inspection import permutation_importance import pandas as pd r = permutation_importance(model, X_val, y_val, n_repeats=10, scoring='roc_auc', random_state=42, n_jobs=-1) imp = pd.Series(r.importances_mean, index=X_val.columns) print(imp.sort_values(ascending=False)) Near-zero or negative values mean the feature adds little on unseen data.

115

Describe a project where you improved a model's performance significantly.

Show answer

Situation: a product-return prediction model had 0.68 AUC with logistic regression. Task: improve it for a warehouse pilot. Action: I engineered features like past return rate per customer and category, moved to LightGBM with early stopping, and tuned it with Optuna on time-based folds. Result: AUC rose to 0.81, and the pilot reduced avoidable returns in two categories. I documented each step's gain.

116

How do you decide when a model is good enough to stop tuning?

Show answer

I agree on a target metric and business threshold before starting, such as recall of 70 percent at 90 percent precision. I stop tuning when gains fall below the noise between CV folds or when further work costs more than the expected value. In one project I stopped at a 0.3 percent gain and spent the time on monitoring instead, which proved more useful.

117

When would you use AdaBoost and how does it re-weight samples?

Show answer

AdaBoost trains weak learners, usually decision stumps, in sequence. After each round it increases the weights of misclassified samples so the next learner focuses on them, and gives each learner a vote weight based on its accuracy. It works well on clean data but is sensitive to noisy labels and outliers, which get ever larger weights. Gradient boosting is a generalisation that has largely replaced it.

118

Use HistGradientBoosting with native categorical support in scikit-learn.

Show answer

from sklearn.ensemble import HistGradientBoostingClassifier clf = HistGradientBoostingClassifier( categorical_features='from_dtype', max_iter=500, learning_rate=0.05, early_stopping=True, random_state=42) X_train[cat_cols] = X_train[cat_cols].astype('category') clf.fit(X_train, y_train) It handles missing values natively and is much faster than GradientBoostingClassifier on large data.

119

Which hyperparameters matter most when tuning a gradient boosting model?

Show answer

The key ones are learning_rate and number of trees, tuned together with early stopping; tree complexity through max_depth or num_leaves and min_child_samples; and randomness through subsample and colsample_bytree. Regularisation terms like lambda and alpha help on noisy data. A common approach is to fix a small learning rate such as 0.05, use early stopping for trees, then tune complexity.

120

Why can impurity-based feature importance be misleading?

Show answer

Impurity-based importance is computed on training data and favours high-cardinality or continuous features, which offer more split points, even if they are noise. It also splits credit between correlated features. Permutation importance on validation data, which measures the score drop when a feature is shuffled, is more reliable, and SHAP values give consistent per-prediction attributions.

121

What is stacking and when is it worth it?

Show answer

Stacking trains several diverse base models and then a meta-model on their out-of-fold predictions to combine them. It often gives a small accuracy gain, common in competitions. In production it adds complexity, latency and maintenance, so it is worth it only when the gain has clear business value. scikit-learn provides StackingClassifier and StackingRegressor.

122

Business users do not trust a gradient boosting model because it is a black box. How do you help?

Show answer

I would use SHAP to show global drivers with a summary plot and explain individual decisions with waterfall plots, such as why one applicant got a high risk score. I would also check that the top drivers make business sense, add monotonic constraints where logic requires it, and share a few real case examples. If trust is still an issue, a scorecard model can serve as a transparent benchmark.

123

Your tree model predicts higher risk for customers with higher income, which is illogical. What do you do?

Show answer

First check whether data supports it, perhaps due to confounding or a small segment. If the relationship should be one-directional for business or regulatory reasons, I would apply monotonic constraints, available in XGBoost, LightGBM and HistGradientBoosting, so risk can only decrease with income. Then compare validation metrics; usually the loss is small and the model becomes more trustworthy.

124

Explain a single prediction with SHAP for a tree model.

Show answer

import shap explainer = shap.TreeExplainer(model) sv = explainer(X_val) shap.plots.waterfall(sv[0]) shap.plots.beeswarm(sv) The waterfall shows how each feature pushes one prediction up or down from the base value; the beeswarm shows global importance and direction. TreeExplainer is fast and exact for tree ensembles.

125

Your gradient boosting model performs worse on recent months than older ones. What do you check?

Clustering and Dimensionality ReductionEasy

Show answer

I would check for drift between periods in feature distributions and target rate, and confirm that the evaluation uses a time-based split. If behaviour has changed, I would train on a more recent window or weight recent data more, add features that capture the change, and set up a regular retraining schedule. Comparing SHAP importance across periods shows which relationships have shifted.

126

How does k-means clustering work?

Show answer

k-means picks k initial centroids, assigns each point to its nearest centroid, recomputes each centroid as the mean of its points, and repeats until assignments stop changing. It minimises within-cluster sum of squares (inertia). It needs k in advance, scaled features, and assumes roughly spherical, similar-sized clusters. k-means++ initialisation and multiple runs with n_init reduce bad starting points.

127

How do you choose the number of clusters k?

Clustering and Dimensionality ReductionEasy

Show answer

Common methods are the elbow method, plotting inertia against k and looking for the bend; the silhouette score, which measures how close points are to their own cluster versus the next one; and the gap statistic. Business usability matters too: four clear segments a marketing team can act on are better than nine statistically optimal but confusing ones.

128

Run k-means and plot the elbow and silhouette scores.

Clustering and Dimensionality ReductionEasy

Show answer

from sklearn.cluster import KMeans from sklearn.metrics import silhouette_score Xs = StandardScaler().fit_transform(X) for k in range(2, 10): km = KMeans(n_clusters=k, n_init=10, random_state=42).fit(Xs) print(k, round(km.inertia_), round(silhouette_score(Xs, km.labels_), 3)) Look for the bend in inertia and a high silhouette value.

129

What is DBSCAN and when is it better than k-means?

Show answer

DBSCAN groups points that lie in dense regions, defined by eps (neighbourhood radius) and min_samples, and labels points in sparse areas as noise. It finds arbitrarily shaped clusters, does not need k, and handles outliers. It struggles when clusters have very different densities and in high dimensions. HDBSCAN improves on it for varying densities.

130

Explain hierarchical clustering and dendrograms.

Show answer

Agglomerative hierarchical clustering starts with each point as its own cluster and repeatedly merges the two closest clusters. Linkage defines closeness: single, complete, average or Ward, which minimises variance. A dendrogram shows the merge history, and cutting it at a height gives the clusters. It does not need k upfront but scales poorly, roughly O(n squared) memory, so it suits smaller datasets.

131

What is PCA and how does it work?

Show answer

Principal Component Analysis finds new orthogonal axes, the principal components, that capture maximum variance in the data. It works by computing eigenvectors of the covariance matrix, or via SVD, after centring and usually standardising the data. Keeping the top components that explain, say, 95 percent of variance reduces dimensions, removes correlation and speeds up models, but the components are harder to interpret.

132

How does Isolation Forest detect anomalies?

Show answer

Isolation Forest builds random trees that split on random features and random thresholds. Anomalies are few and different, so they get isolated in fewer splits, giving shorter average path lengths. The anomaly score comes from that path length. It is fast, works without labels and scales well. The contamination parameter sets the expected share of anomalies and determines the threshold.

133

What is association rule mining? Explain support, confidence and lift.

Show answer

Association rule mining finds items that occur together, such as in market basket analysis. Support is how often an itemset appears in all transactions. Confidence of A to B is the share of transactions with A that also contain B. Lift is confidence divided by B's support; lift above 1 means A and B occur together more than by chance. Apriori and FP-Growth are common algorithms.

134

Marketing wants customer segments for campaigns. How do you approach it?

Show answer

I would agree on the purpose first, such as retention or upsell. Then build RFM features (recency, frequency, monetary) plus category preferences, log-transform skewed values and scale them. I would try k-means for k between 3 and 8, use silhouette and business review to pick k, profile each segment and give them clear names. Finally, test segment-based campaigns against a control group.

135

k-means gives one huge cluster and several tiny ones. What could be wrong?

Show answer

Likely causes are unscaled features, where one large-range feature dominates distance; heavy skew or outliers pulling centroids; or data that is not naturally spherical. I would log-transform and scale features, remove or cap extreme outliers, and try different k values. If shapes are irregular, DBSCAN, HDBSCAN or a Gaussian Mixture Model may fit better.

136

Your dataset has 800 correlated features and the model trains slowly. What would you do?

Show answer

I would remove constant and near-duplicate features, drop one from each highly correlated pair, then apply PCA inside the pipeline and keep components explaining about 95 percent of variance. Alternatively, use L1-regularised models or tree importances for supervised selection. I would compare speed and cross-validated scores; often 50 to 100 features give nearly the same accuracy.

137

Apply PCA and check explained variance.

Show answer

from sklearn.decomposition import PCA import numpy as np pca = PCA().fit(Xs) cum = np.cumsum(pca.explained_variance_ratio_) n = np.argmax(cum >= 0.95) + 1 print(n, 'components explain 95 percent') X_red = PCA(n_components=0.95).fit_transform(Xs) Passing a float to n_components keeps enough components to reach that variance.

138

Detect anomalies with Isolation Forest.

Show answer

from sklearn.ensemble import IsolationForest iso = IsolationForest(n_estimators=300, contamination=0.01, random_state=42) iso.fit(X_train) scores = -iso.score_samples(X_new) flags = iso.predict(X_new) == -1 Higher scores mean more anomalous. Tune contamination to match how many alerts reviewers can handle.

139

Build RFM features for customer segmentation in pandas.

Show answer

snap = df['order_date'].max() + pd.Timedelta(days=1) rfm = df.groupby('customer_id').agg( recency=('order_date', lambda d: (snap - d.max()).days), frequency=('order_id', 'nunique'), monetary=('amount', 'sum')) rfm[['frequency', 'monetary']] = np.log1p(rfm[['frequency', 'monetary']]) Scale these columns before clustering.

140

Tell me about a time your analysis was not used by stakeholders. What did you learn?

Show answer

Situation: I built eight customer segments, but marketing did not use them. Task: understand why. Action: I met the team and learned the segments were hard to act on and lacked clear names. I reduced them to five, added a one-page profile and suggested campaign ideas for each. Result: two segments were used in the next campaign. I learned to involve users from the start.

141

How do you prioritise when several teams ask you for models at the same time?

Show answer

I estimate each request's business value, effort and urgency, then share a ranked list with my manager so priorities are agreed rather than assumed. For quick wins I offer a simple baseline first. In one quarter this helped us deliver a high-value churn model first while giving another team a dashboard-based interim answer, and both teams knew the timeline.

142

A product team wants 'similar products' recommendations with no purchase labels. How would you start?

Show answer

I would build product embeddings from content: TF-IDF or sentence embeddings of titles and descriptions, plus category and price features. Then use cosine similarity with nearest-neighbour search, such as scikit-learn NearestNeighbors or an approximate index like FAISS for large catalogues. I would review samples with the product team and later add co-view or co-purchase signals once collected.

143

Find the most similar items using cosine similarity.

Show answer

from sklearn.feature_extraction.text import TfidfVectorizer from sklearn.neighbors import NearestNeighbors X = TfidfVectorizer(min_df=2).fit_transform(df['title']) nn = NearestNeighbors(n_neighbors=6, metric='cosine').fit(X) dist, idx = nn.kneighbors(X[0]) print(df['title'].iloc[idx[0][1:]].tolist()) The first neighbour is the item itself, so skip it.

144

What is the difference between PCA and t-SNE or UMAP?

Show answer

PCA is a linear method that preserves global variance, is fast, deterministic and can transform new data, so it is used for preprocessing. t-SNE and UMAP are non-linear methods that preserve local neighbourhoods, producing good 2D visualisations of clusters. t-SNE distances between clusters are not meaningful. UMAP is faster, keeps more global structure and can transform new points.

145

What is a Gaussian Mixture Model and how is it different from k-means?

Show answer

A GMM assumes data comes from a mix of Gaussian distributions and fits their means, covariances and weights with the Expectation-Maximisation algorithm. Unlike k-means' hard assignments, GMM gives soft probabilities of belonging to each cluster and can model elliptical clusters of different sizes. You can choose the number of components using BIC or AIC.

146

How do you evaluate clustering when there are no labels?

Show answer

Use internal metrics: silhouette score, Davies-Bouldin index (lower is better) and Calinski-Harabasz index (higher is better). Check stability by re-clustering on bootstrap samples. Most important is external usefulness: profile each cluster with key statistics and check whether business teams find the segments distinct and actionable, for example different response to a campaign.

147

You need to detect unusual transactions but have no fraud labels. What do you do?

Show answer

I would start with unsupervised anomaly detection: Isolation Forest or Local Outlier Factor on features like amount relative to the customer's usual spend, time of day, new device and merchant risk. I would set the threshold based on what the review team can handle, collect investigator feedback as labels, and move to a supervised model once enough labelled cases exist.

148

The segments you built six months ago no longer match customer behaviour. What do you do?

Show answer

I would measure drift by comparing feature distributions and segment sizes over time. Then refit clustering on recent data and map new clusters to old ones by centroid similarity so business names stay consistent. I would set up a quarterly refresh with stability checks and alert when segment sizes shift sharply, and explain any renamed or merged segments to stakeholders.

149

Run DBSCAN and count clusters and noise points.

Show answer

from sklearn.cluster import DBSCAN import numpy as np db = DBSCAN(eps=0.5, min_samples=10).fit(Xs) labels = db.labels_ n_clusters = len(set(labels)) - (1 if -1 in labels else 0) print(n_clusters, 'clusters,', np.sum(labels == -1), 'noise') Pick eps from the knee of a sorted k-nearest-neighbour distance plot.

150

What is Local Outlier Factor and how does it differ from Isolation Forest?

Show answer

Local Outlier Factor compares the local density of a point with the density of its neighbours; a point in a much sparser area than its neighbours gets a high outlier score. It catches local anomalies inside varied-density data but is distance-based, so it needs scaling and is slower on large data. Isolation Forest is tree-based, faster and scales better, but is less sensitive to local density differences.

151

What is k-fold cross-validation and why is it better than a single split?

Show answer

k-fold cross-validation splits training data into k folds, trains on k-1 folds and validates on the remaining one, repeating k times so every row is validated once. Averaging the k scores gives a more stable estimate than one random split and shows variance across folds. Five or ten folds are common. Use StratifiedKFold for classification and GroupKFold when rows from one user must stay together.

152

What is a holdout set and why should it be touched only once?

Show answer

A holdout or test set is kept aside for the final evaluation after all model and hyperparameter choices are done. If you check it repeatedly and adjust the model, you are indirectly tuning on it, so its score becomes optimistic. Use cross-validation or a separate validation set for decisions, and report the holdout score once as the realistic estimate.

153

Run stratified 5-fold cross-validation with multiple metrics.

Show answer

from sklearn.model_selection import StratifiedKFold, cross_validate cv = StratifiedKFold(n_splits=5, shuffle=True, random_state=42) res = cross_validate(pipe, X, y, cv=cv, scoring=['roc_auc', 'average_precision', 'f1'], n_jobs=-1) for k in ['test_roc_auc', 'test_average_precision', 'test_f1']: print(k, res[k].mean().round(3), res[k].std().round(3))

154

Tune a pipeline with GridSearchCV.

Show answer

from sklearn.model_selection import GridSearchCV params = {'clf__C': [0.01, 0.1, 1, 10], 'clf__penalty': ['l2']} gs = GridSearchCV(pipe, params, cv=5, scoring='roc_auc', n_jobs=-1) gs.fit(X_train, y_train) print(gs.best_params_, gs.best_score_) best = gs.best_estimator_ The step__param naming targets parameters inside a pipeline step.

155

How do you explain the precision-recall trade-off to a business manager?

Show answer

I use their own example. For fraud: 'If we flag more transactions, we catch more fraud, which is recall, but more genuine customers get blocked, which hurts precision. Here is a table of what each threshold means per day: frauds caught, customers disturbed, reviews needed.' Then we choose the threshold together based on cost. Managers relate much better to counts than to percentages.

156

When should you use TimeSeriesSplit or GroupKFold instead of normal k-fold?

Show answer

Use TimeSeriesSplit when data is ordered in time, so each validation fold comes after its training data and no future information leaks back. Use GroupKFold when multiple rows belong to the same entity, such as several visits per patient, so the same entity never appears in both train and validation. Normal k-fold in these cases gives over-optimistic scores.

157

Compare grid search, random search and Bayesian optimisation.

Show answer

Grid search tries every combination of listed values; it is thorough but expensive and wastes trials on unimportant parameters. Random search samples combinations and usually finds good settings faster in high-dimensional spaces. Bayesian optimisation, as in Optuna or Hyperopt, builds a model of past trials to pick promising settings next, using fewer trials. Successive halving methods add early pruning of weak trials.

158

What are learning curves and validation curves?

Show answer

A learning curve plots training and validation scores against training set size; if both converge at a low score the model underfits, and if a gap remains, more data may help. A validation curve plots scores against one hyperparameter, such as max_depth, showing where overfitting starts. Both help decide whether to collect data, add features or change complexity.

159

What is the log loss metric and why is it useful?

Show answer

Log loss, or cross-entropy, is the negative average log of the probability assigned to the true class. It heavily penalises confident wrong predictions, such as predicting 0.99 for an event that does not happen. It evaluates probability quality, not just ranking, so it suits problems where probabilities drive decisions. Lower is better, and a constant base-rate prediction is the baseline to beat.

160

What is the difference between offline evaluation and online A/B testing of a model?

Show answer

Offline evaluation measures metrics like AUC on historical holdout data; it is fast and cheap but cannot capture user behaviour changes or feedback loops. Online A/B testing exposes a share of real traffic to the new model and measures business outcomes like conversion or loss rate. Offline gains do not always translate online, so important models are validated with A/B tests or shadow deployments.

161

A new model has slightly better AUC but worse precision in the top 1,000 cases the team acts on. Which do you pick?

Show answer

I would pick the model that does better where decisions are actually made. If the team only reviews the top 1,000 cases, precision at 1,000 or recall at that cut-off matters more than global AUC. I would make that the primary metric in future comparisons, and confirm the result holds across several time periods before switching.

162

Grid search with 5 hyperparameters is taking two days. How do you speed it up?

Show answer

Switch to random search or Optuna with pruning, which find good settings with far fewer trials. Narrow ranges using a quick coarse search, tune the most important parameters first, use early stopping for boosting, run trials in parallel with n_jobs, and tune on a stratified sample of the data before a final refit on the full set. HalvingRandomSearchCV is another quick option.

163

Your manager asks for a single number showing the model's business value. What do you give?

Show answer

I would convert model outcomes into money or time using a cost matrix agreed with the business, for example savings from caught fraud minus the cost of reviewing false alerts. I would compare this against the current process or a rule-based baseline on holdout data and show a range rather than one point. A short note on assumptions keeps the number honest.

164

Plot a learning curve in scikit-learn.

Show answer

from sklearn.model_selection import LearningCurveDisplay import numpy as np LearningCurveDisplay.from_estimator(model, X, y, cv=5, train_sizes=np.linspace(0.1, 1.0, 6), scoring='roc_auc', n_jobs=-1) plt.show() A persistent gap between curves suggests more data or regularisation; two low curves suggest a richer model or better features.

165

Check calibration with a reliability curve and Brier score.

Show answer

from sklearn.calibration import CalibrationDisplay from sklearn.metrics import brier_score_loss proba = clf.predict_proba(X_test)[:, 1] print('Brier', brier_score_loss(y_test, proba)) CalibrationDisplay.from_predictions(y_test, proba, n_bins=10) plt.show() Points on the diagonal mean predicted probabilities match observed frequencies.

166

Tell me about a time you discovered your own mistake after sharing results.

Show answer

Situation: after presenting a 12 percent uplift from a pricing model, I found my evaluation used a random split on time-series data. Task: correct it quickly. Action: I re-ran with a time-based split, which showed 7 percent, and informed my manager the same day with the corrected analysis. Result: the team appreciated the honesty, the rollout plan used the right number, and I added split checks to our review template.

167

What is nested cross-validation and when do you need it?

Show answer

Nested CV has an inner loop that tunes hyperparameters and an outer loop that evaluates the whole tuning process on unseen folds. It gives an unbiased performance estimate when you tune heavily on small data, because a normal CV score used both to tune and to report is optimistic. It is expensive, so it is mainly used for small datasets or research reporting.

168

How do you compare two models fairly?

Show answer

Use the same data splits and folds, the same preprocessing inside pipelines and the same metric tied to the business goal. Look at the mean and standard deviation across folds, not one number, and use a paired test like a corrected t-test or bootstrap on the fold differences. Also compare cost factors: latency, memory, interpretability and maintenance.

169

What fairness checks would you run on a model that affects people?

Show answer

Compare metrics across groups such as gender, age band or region: approval rate (demographic parity), true positive rate (equal opportunity) and false positive rate. Check calibration per group too. Remove or audit proxy features like pin code that may stand in for protected attributes. Document findings in a model card, and involve legal or compliance teams, especially for lending and hiring use cases.

170

Your cross-validation score is 0.85 but the test score is 0.70. What could explain it?

Show answer

Possible causes: leakage inside CV, such as preprocessing or feature selection fit outside folds; repeated tuning that overfit the CV folds; a test set from a different time period or population; or duplicate entities across folds but not test. I would check pipeline structure, switch to time-based or group splits that match how test data differs, and compare feature distributions between the two sets.

171

An A/B test of a new recommendation model shows no significant uplift despite better offline metrics. What next?

Show answer

I would check the test first: enough sample size and duration, correct randomisation, and no bugs in serving such as stale features. Then investigate whether offline metrics really reflect user value, for example optimising clicks rather than purchases, or popularity bias. Segment-level analysis may show gains in some groups. I would revise the offline metric to better match the online goal.

172

Tune LightGBM with Optuna.

Show answer

import optuna from sklearn.model_selection import cross_val_score def objective(trial): p = {'num_leaves': trial.suggest_int('num_leaves', 15, 255), 'learning_rate': trial.suggest_float('learning_rate', 0.01, 0.2, log=True), 'min_child_samples': trial.suggest_int('min_child_samples', 10, 200)} m = lgb.LGBMClassifier(n_estimators=500, **p) return cross_val_score(m, X, y, cv=5, scoring='roc_auc').mean() study = optuna.create_study(direction='maximize') study.optimize(objective, n_trials=50)

173

Get a bootstrap confidence interval for a test metric.

Show answer

import numpy as np from sklearn.metrics import roc_auc_score rng = np.random.default_rng(0); scores = [] yt, pp = np.array(y_test), np.array(proba) for _ in range(1000): i = rng.integers(0, len(yt), len(yt)) if len(np.unique(yt[i])) == 2: scores.append(roc_auc_score(yt[i], pp[i])) print(np.percentile(scores, [2.5, 97.5]))

174

How do you handle pressure to report better model numbers than the data supports?

Show answer

I explain the risk clearly: inflated numbers lead to bad decisions and lose trust when production results arrive. I present the honest metric with confidence intervals, and then offer a concrete plan to improve it, such as new data sources. In one case I showed that a 'better' number came from leakage, and leadership agreed to report the lower but reliable figure.

175

Explain the bias in AUC when classes are extremely imbalanced, with a numeric intuition.

Show answer

With 100 fraud and 100,000 genuine cases, a false positive rate of 1 percent looks tiny on the ROC curve, but it means 1,000 false alerts versus at most 100 true ones, so precision is below 10 percent. ROC-AUC can be 0.97 while the model is impractical. Precision-recall curves expose this because precision directly depends on the count of false positives.

176

What does it mean to deploy a machine learning model?

Show answer

Deployment makes a trained model available to produce predictions for real users or systems. It can be batch, scoring data on a schedule and writing results to a table, or real-time, serving predictions through a REST API, or streaming, scoring events as they arrive. It includes packaging the model with its preprocessing, setting up infrastructure, logging, monitoring and a retraining plan.

177

How do you save and load a scikit-learn model?

Show answer

Save the whole fitted pipeline so preprocessing travels with the model. import joblib joblib.dump(pipe, 'model_v1.joblib') pipe = joblib.load('model_v1.joblib') preds = pipe.predict(new_df) Load with the same scikit-learn version used for training, and never load pickle or joblib files from untrusted sources because they can execute code. ONNX or skops are safer portable options.

178

What is MLOps?

Show answer

MLOps applies DevOps practices to machine learning: version control for code, data and models, automated training pipelines, testing, CI/CD for model releases, model registries, monitoring and automated retraining. Its aim is to move models from notebooks to production reliably and keep them healthy. Common tools are MLflow, DVC, Airflow, Kubeflow, and cloud platforms like SageMaker, Vertex AI and Azure ML.

179

Why containerise an ML model with Docker?

Show answer

Docker packages the model, code, Python version and exact library versions into one image that runs the same on a laptop, a test server or Kubernetes. This removes 'works on my machine' problems, especially with scikit-learn version mismatches. Images can be scanned, versioned and deployed through CI/CD, and scaled horizontally behind a load balancer on platforms like Kubernetes, ECS or Cloud Run.

180

How do you work with data engineers and software engineers on an ML project?

Show answer

I agree early on interfaces: which tables or APIs provide features, the input and output schema, latency targets and who owns monitoring. I share a short design document and keep my code in the same repository standards as theirs, with tests and reviews. In my last project weekly syncs with data engineering helped us fix a late-arriving data issue before launch rather than after.

181

What is data drift and concept drift?

Show answer

Data drift, or covariate shift, means the distribution of input features changes, such as average order value rising after a pricing change. Concept drift means the relationship between features and target changes, such as fraud patterns evolving. Data drift can be detected without labels using PSI or KS tests; concept drift usually needs labels and shows up as falling performance. Both trigger investigation and retraining.

182

What is a model registry and why is it useful?

Show answer

A model registry is a central store of model versions with metadata: training data version, parameters, metrics, owner and stage such as staging or production. It makes promotions and rollbacks controlled and auditable, and lets serving systems load a model by name and version. MLflow Model Registry, SageMaker Model Registry and Vertex AI Model Registry are common examples.

183

Compare batch scoring and real-time inference.

Show answer

Batch scoring runs on a schedule, for example nightly churn scores for all customers written to a table; it is simple, cheap and tolerant of slow models. Real-time inference returns predictions on request, such as fraud checks during payment, and needs low latency, high availability, scaling and fresh features. Choose real-time only when the decision must happen at the moment of the event.

184

What are shadow deployment and canary release for models?

Show answer

In shadow deployment the new model receives a copy of live traffic and its predictions are logged but not used, so you can compare it with the current model safely. In a canary release a small share of real traffic, such as 5 percent, is served by the new model, and the share grows if metrics stay healthy. Both reduce the risk of a bad model affecting all users.

185

What should you monitor for a model in production?

Show answer

Monitor service health such as latency, error rate and throughput; input data quality such as missing values, schema changes and out-of-range values; feature and prediction distribution drift; and real performance once labels arrive. Also track business KPIs the model influences. Set alerts with thresholds and owners, and keep a dashboard so issues are found before users report them.

186

A model's accuracy dropped 10 percent over three months in production. How do you investigate?

Show answer

I would first check data pipelines for broken features, nulls or schema changes, then compare current feature distributions with training data using PSI to find drift. I would check whether the label definition or business process changed, and analyse errors by segment. If drift is real, retrain on recent data, possibly with time-weighted samples, and set up automatic drift alerts and scheduled retraining.

187

A new model version performs badly right after release. How do you handle it?

Show answer

Roll back immediately to the previous version from the model registry, which should be a one-step action. Then investigate: compare the input features served with those used in validation, check for training-serving skew, version mismatches or a wrong artifact. Add the failure case to automated pre-release checks, and use canary or shadow releases next time so issues affect only a small share of traffic.

188

Your team retrains models manually in notebooks every month. How would you improve it?

Show answer

I would convert the notebook into a versioned training pipeline with separate steps for data extraction, validation, feature building, training and evaluation, scheduled with Airflow or a cloud pipeline. Each run would log to MLflow, compare against the production model on a fixed holdout, and register the new model only if it passes thresholds. Promotion would stay a reviewed step, at least initially.

189

Serve a scikit-learn model with FastAPI.

Show answer

from fastapi import FastAPI from pydantic import BaseModel import joblib, pandas as pd app = FastAPI() model = joblib.load('model_v1.joblib') class Item(BaseModel): age: int income: float city: str @app.post('/predict') def predict(item: Item): df = pd.DataFrame([item.model_dump()]) return {'score': float(model.predict_proba(df)[0, 1])} Run it with: uvicorn app:app --host 0.0.0.0 --port 8000

190

Write a simple Dockerfile for an ML API.

Show answer

FROM python:3.12-slim WORKDIR /app COPY requirements.txt . RUN pip install --no-cache-dir -r requirements.txt COPY app.py model_v1.joblib ./ EXPOSE 8000 CMD ["uvicorn", "app:app", "--host", "0.0.0.0", "--port", "8000"] The exec-form CMD avoids a shell wrapper. Pin versions in requirements.txt so training and serving match.

191

Log a training run to MLflow.

Show answer

import mlflow, mlflow.sklearn mlflow.set_experiment('churn') with mlflow.start_run(): pipe.fit(X_train, y_train) auc = roc_auc_score(y_val, pipe.predict_proba(X_val)[:, 1]) mlflow.log_params({'C': 1.0, 'model': 'logreg'}) mlflow.log_metric('val_auc', auc) mlflow.sklearn.log_model(pipe, 'model', registered_model_name='churn-model')

192

Write a unit test for a model prediction function.

Show answer

import joblib, pandas as pd def test_predict_shape_and_range(): model = joblib.load('model_v1.joblib') df = pd.DataFrame([{'age': 30, 'income': 50000.0, 'city': 'Pune'}, {'age': 45, 'income': 0.0, 'city': 'UnknownCity'}]) p = model.predict_proba(df)[:, 1] assert p.shape == (2,) assert ((p >= 0) & (p <= 1)).all() Run with pytest in CI so broken artifacts never reach production.

193

Tell me about a time you took a model from notebook to production.

Show answer

Situation: our demand model ran only in a notebook on my laptop. Task: make it run daily for 300 stores. Action: I refactored it into modules with a scikit-learn pipeline, added tests and input checks, containerised it with Docker and scheduled it with Airflow writing forecasts to the warehouse. Result: forecasts became available by 6 a.m. daily without manual work, and failures triggered alerts instead of silent gaps.

194

Tell me about a time a production model failed and how you communicated it.

Show answer

Situation: our daily scoring job silently produced all-zero scores for two days after an upstream column was renamed. Task: fix it and inform affected teams. Action: I rolled back to cached scores, patched the feature mapping, and sent a short note explaining impact, cause and fix. I then added schema validation and a check on score distribution. Result: the issue never repeated, and teams trusted our alerts more.

195

What is training-serving skew and how do you prevent it?

Show answer

Training-serving skew is when features at prediction time are computed differently from features used in training, causing silent accuracy loss. Causes include separate code paths, different data sources, or time windows that use future data in training. Prevent it by sharing one feature-computation code path or a feature store, logging served features and comparing their distributions with training data.

196

Your real-time model API takes 400 ms but the requirement is under 50 ms. What do you do?

Show answer

Profile first to see whether time goes to feature lookup, preprocessing or the model. Then cache or precompute features, load the model once at startup, reduce trees or features, use a faster library like LightGBM or ONNX Runtime, and batch requests if possible. Run several workers and keep the service close to its data store. Re-measure p95 and p99 latency, not just the average.

197

Labels for your loan model arrive only after 6 months. How do you monitor it meanwhile?

Show answer

I would rely on proxy signals: input data drift with PSI, prediction score distribution, approval rates by segment, and early indicators like first missed payment at 30 or 60 days. Once enough early outcomes arrive, I would check performance on these. I would also set up a delayed evaluation job that computes full metrics as labels mature and compares them with the validation baseline.

198

Your model must run on a mobile app with no internet. What do you consider?

Show answer

Model size, memory and latency on low-end phones become key. I would choose a compact model, reduce features, and convert it to a mobile format such as ONNX, TensorFlow Lite or Core ML, applying quantisation if needed. I would make sure all preprocessing is replicated on-device, test accuracy after conversion, and plan how model updates will ship with app releases.

199

Calculate the Population Stability Index for a feature.

Show answer

import numpy as np def psi(expected, actual, bins=10): cuts = np.quantile(expected, np.linspace(0, 1, bins + 1)) cuts[0], cuts[-1] = -np.inf, np.inf e = np.histogram(expected, cuts)[0] / len(expected) + 1e-6 a = np.histogram(actual, cuts)[0] / len(actual) + 1e-6 return float(np.sum((a - e) * np.log(a / e))) Rule of thumb: below 0.1 stable, 0.1 to 0.25 moderate, above 0.25 significant drift.

200

What is a feature store and when is it needed?