Operators, product owners, and people affected by the trees & boosting decision.
Classical ML · weeks 7
Trees, forests & gradient boosting
A marketplace must catch rare fraud without blocking trustworthy sellers.
Opening production brief
A measurable problem, before a model.
impurity gain, bagging, feature subsampling, pseudo-residuals with version, owner, event time, and availability contracts.
A marketplace must catch rare fraud without blocking trustworthy sellers. Every solution must state latency, cost, capacity, and fallback limits.
Offline evidence plus a deployable resolution: Calibrated random-forest and XGBoost baselines with cost-sensitive thresholds.
Frame and baseline
Turn the production problem into explicit data, metric, and baseline contracts.
- 01↗
Trees as learned decision rules
A tree learns readable local rules; constraints keep those rules from becoming a lookup table.
65–90 min · checkpoint · deterministic lab - 02↗
Random forests and diversity
A forest is strong when its trees are individually useful and collectively disagree in the right ways.
65–90 min · checkpoint · deterministic lab
Build and stress
Formalize the machinery, test counterexamples, and expose system limits.
- 03↗
Gradient boosting as error correction
Boosting learns by repeatedly asking: what structured error remains?
65–90 min · checkpoint · deterministic lab - 04↗
XGBoost: regularized boosting at scale
XGBoost scales boosting by turning gradient statistics into regularized decisions about leaves and splits.
65–90 min · checkpoint · deterministic lab
Resolve and operate
Join model behavior to architecture, observability, rollout, and rollback.
- 05↗
Calibrating and explaining ensembles
Calibrate probabilities for decisions, and read explanations as model accounting—not causal proof.
65–90 min · checkpoint · deterministic lab - HW↗
Rare-fraud ensemble decision system
Compare a tree, forest, gradient boosting, and XGBoost with temporal validation and operational thresholding.
Rubric · staged hints · reference resolution
Prerequisite guidance
1 concepts make this faster.
Prerequisites are guidance, never hard gates.