Sales Prediction

Task 5 · Regression

Predicts product sales from TV, Radio and Newspaper ad spend across 200 markets. A degree-2 polynomial regression captures diminishing returns and the TV × Radio synergy that a straight line misses.

Code on GitHub ↗ Notebook ↗

0.987R² (vs 0.899 linear)

0.53MAE, thousand units

TVbiggest overall impact

≈0newspaper effect

Try it: plan an ad budget

Plans:

Predicted sales

– thousand units

Total budget: –

Sales vs each channel — TV tracks sales most closely

Model comparison — Polynomial regression wins on every metric

Residual plots — The linear model's errors form a U-shape

Channel impact — TV and Radio drive sales; newspaper barely does