Two analysts, the same data, the same algorithm — and a 13-point gap in accuracy. The difference was what they showed the model. Learn to turn a raw behavioural export into a modeling table on purpose: collapsing event logs to the customer grain, applying the recency-frequency-intensity template to every behaviour, encoding categories without leaking the outcome, and enforcing a temporal line that no feature may cross.
Topics and managerial frameworks covered in Feature Engineering:
Students manipulate these interactive visual simulations directly inside the chapter reading before answering gated check-in questions:
Students allocate a fixed effort budget between engineering features and upgrading the algorithm, and watch the resulting AUC. The returns are modelled on the case table in this chapter: the first systematic pass at features buys a great deal, later passes...
Students drag the prediction date and the observation window, then judge a list of candidate features. A feature is legal only if every value it needs existed strictly before the prediction date.
One slider (number of category levels) drives three encodings side by side, so the tradeoff becomes arithmetic rather than assertion: one-hot explodes the column count and starves each level of observations; ordinal stays at one column but invents an order...
This chapter connects directly to these full MktPraxis analytical apps (with pre-loaded business case scenarios, CSV upload, and reproducible Python export):
Predict a numeric outcome using multiple drivers.
Pre-loaded Business Scenarios: EDM Events: Lead Time, Ticket Type, and Revenue Per Patron · Customer Valuation: RFM Metrics and Segment Labels
Predict binary outcomes like Churn vs. Retain.
Pre-loaded Business Scenarios: Influencer Posts: What Makes a Story Drive a Sale? · B2B Email Sequences: Predicting Demo Bookings · Digital Promo A/B Test: High Incentive vs Standard Offer · Subscription Churn: Predicting Customer Attrition · Event Marketing: Predicting Registration Conversion · RFM-Based Purchase Prediction: Who Buys Next? · SignalPath MarTech: Which Leads Actually Close? · Will They Come Back? Predicting Season-Membership Renewal
Compare binary classification algorithms side-by-side on training and holdout data. Select from Logistic Regression, Decision Tree, k-NN, Naive Bayes, and Linear SVM.
Pre-loaded Business Scenarios: CloudMetrics SaaS — Customer Churn Prediction · NexaFlow — B2B Lead Conversion Scoring · PixelReach Ads — Click Prediction for Programmatic Bidding
Transform continuous data into meaningful categories using binning. See how the same data tells different stories.
Pre-loaded Business Scenarios: Quarterly Campaign Performance Review · Loyalty Program Tier Design · App Re-engagement Campaign Targeting · Customer Segmentation for Targeted Nurture