All paths
Intermediate path
From causal effect to population twin behavior
Featured tutorial: translate a heterogeneous treatment effect into agent-level rules without double-counting baseline risk.
- Lesson 1 · Prediction versus interventionSeparate who is likely to buy from who buys because of the offer.8 min
- Lesson 2 · Estimate CATE for defined segmentsEstimate uplift for six segments with adjustment for loyalty and expected demand.10 min
- Lesson 3 · Map effect estimates to twin parametersTurn absolute uplifts into twin purchase rules.10 min
- Lesson 4 · Preserve uncertainty and validity rangesSample effects and record where the estimate is valid.8 min
- Lesson 5 · Aggregate outcomes and validateAggregate twin outcomes and compare with held-out observations.10 min
Capstone
Apply a discount intervention to three customer twins and compute expected incremental purchases.