How Long Does It Take to Learn Machine Learning Properly?

Your ML timeline estimate

Focused hours per week: ____

Block A - Python + pandas ........ skip if confident
Block B - Statistics + algebra ... revise what you forgot
Block C - Core supervised ML ..... do not skip
Block D - Unsupervised basics .... clustering, PCA
Block E - 3 projects ............. the biggest block

Write a target week for each block.
Review every 4 weeks and adjust honestly.

You are progressing properly when you can:

[ ] Explain why you split train and test data
[ ] Pick a metric that fits the business problem
[ ] Spot overfitting from train vs validation scores
[ ] Build a pipeline without data leakage
[ ] Explain a model's prediction in simple words
[ ] Repeat the whole process on a new dataset alone

Course certificates are not on this list on purpose.

import numpy as np

# Vectors and dot product
w = np.array([0.5, -1.2, 2.0])
x = np.array([3.0, 1.0, 0.5])
print("prediction:", w @ x)      # 0.5*3 - 1.2*1 + 2.0*0.5 = 1.3

# Mean, variance, standard deviation
data = np.array([12, 15, 11, 19, 13])
print(data.mean(), data.var(), data.std())

# Correlation between two features
a = np.array([1, 2, 3, 4, 5])
b = np.array([2, 4, 5, 4, 5])
print(np.corrcoef(a, b)[0, 1])

import numpy as np

# Fit y = m*x + c with plain gradient descent
x = np.array([1, 2, 3, 4, 5], dtype=float)
y = np.array([3, 5, 7, 9, 11], dtype=float)  # true: m=2, c=1

m, c, lr = 0.0, 0.0, 0.01
for step in range(5000):
    pred = m * x + c
    error = pred - y
    m -= lr * (2 * (error * x).mean())
    c -= lr * (2 * error.mean())

print(round(m, 2), round(c, 2))  # close to 2.0 1.0

from sklearn.datasets import load_wine
from sklearn.model_selection import cross_val_score
from sklearn.pipeline import make_pipeline
from sklearn.preprocessing import StandardScaler
from sklearn.linear_model import LogisticRegression
from sklearn.tree import DecisionTreeClassifier
from sklearn.ensemble import RandomForestClassifier

X, y = load_wine(return_X_y=True)
models = {
    "logistic": make_pipeline(StandardScaler(), LogisticRegression(max_iter=1000)),
    "tree": DecisionTreeClassifier(random_state=0),
    "forest": RandomForestClassifier(n_estimators=200, random_state=0),
}
for name, model in models.items():
    scores = cross_val_score(model, X, y, cv=5)
    print(f"{name:9s} {scores.mean():.3f} +/- {scores.std():.3f}")

from sklearn.datasets import load_iris
from sklearn.preprocessing import StandardScaler
from sklearn.cluster import KMeans
from sklearn.decomposition import PCA

X, _ = load_iris(return_X_y=True)
X_scaled = StandardScaler().fit_transform(X)

km = KMeans(n_clusters=3, n_init=10, random_state=0)
labels = km.fit_predict(X_scaled)

pca = PCA(n_components=2)
X_2d = pca.fit_transform(X_scaled)
print("explained variance:", pca.explained_variance_ratio_.round(2))
print("cluster sizes:", [list(labels).count(k) for k in range(3)])

Week ___  Block: A / B / C / D / E

Concept of the week: ________________

Mon  Learn concept (theory + docs)
Tue  Implement from scratch or by hand
Wed  Use scikit-learn on a small dataset
Thu  Try a second dataset, compare results
Fri  Write notes: when to use / when not to
Sat  Project time (2-3 hours)
Sun  Review + plan next week

Experiment # ___   Date: ________
Dataset: ______________________
Target: _______________________
Baseline model + score: _________

Change tried: __________________
New score (validation): __________
Train score: ___________________
Overfitting?  Yes / No
Keep the change?  Yes / No
Why: ___________________________

ML project quality checklist

[ ] Clear business question in README
[ ] Data source and licence stated
[ ] EDA with 4-6 explained charts
[ ] Train / validation / test split done correctly
[ ] Baseline model reported
[ ] Final model with chosen metric explained
[ ] Error analysis: where the model fails
[ ] Reproducible: requirements.txt + random seeds
[ ] Simple demo or saved model with usage example