How Power BI Helps You Land a Data Scientist Job Faster

import pandas as pd

# X_test: test features DataFrame, model: fitted classifier
scores = X_test[["customer_id", "region", "plan"]].copy()
scores["churn_prob"] = model.predict_proba(X_test[features])[:, 1]
scores["risk_band"] = pd.cut(scores["churn_prob"],
                             bins=[0, 0.3, 0.6, 1.0],
                             labels=["Low", "Medium", "High"],
                             include_lowest=True)
scores.to_csv("churn_scores.csv", index=False)

churn_scores.csv
customer_id  text      unique key
region       text      for slicers
plan         text      for slicers
churn_prob   decimal   0 to 1, model output
risk_band    text      Low / Medium / High

High Risk Customers =
CALCULATE ( COUNTROWS ( Scores ), Scores[risk_band] = "High" )

Avg Churn Probability =
AVERAGE ( Scores[churn_prob] )

High Risk Share =
DIVIDE ( [High Risk Customers], COUNTROWS ( Scores ) )

Risk Rank =
RANKX ( ALL ( Scores ), Scores[churn_prob], , DESC, Dense )
// Use as a calculated column, then filter the table visual
// to Risk Rank <= 50 to show the top 50 customers.

// Create a What-if parameter "Offer Cost" (0 to 2000, step 100)
Expected Net Value =
VAR Targeted = [High Risk Customers]
VAR SaveRate = 0.2          // assumption, state it on the page
VAR AvgRevenue = 6000       // assumption, state it on the page
RETURN
    Targeted * SaveRate * AvgRevenue - Targeted * 'Offer Cost'[Offer Cost Value]

Page title: Who to call this month
Subtitle: Churn model scores, refreshed weekly
Note: Save rate and average revenue are assumptions —
      change them with the business team before use.

Project plan: churn model + Power BI report
1. Pick a public telecom or subscription churn dataset
2. Baseline model (logistic regression), then a tree model
3. Evaluate with recall and precision on held-out data
4. Export scores with IDs and business fields
5. Build 2 report pages: overview and action list
6. Publish screenshots and the .pbix in the repo

## Dashboard
The Power BI report shows churn risk by region and plan,
a ranked list of high-risk customers, and an expected-value
estimate for a retention offer. Assumptions are labelled
on the page. Screenshots: /images/overview.png

2-minute interview demo script
0:00 Problem and business question
0:20 Data and model, one sentence each
0:40 Model quality on held-out data
1:00 Dashboard: overview page, filter one region
1:30 Action list and expected value
1:50 One limitation and next step