Optimize the Right Thing
summary
An open-source Python tool Empulse package integrates with Scikit-Learn and popular boosters (XGBoost CatBoost LightGBM) to translate a domain cost matrix into metrics, scorers, and cost-aware classifiers. It supports class- and instance-dependent costs, uses sample weights or direct loss optimization (cost-sensitive boosting, logistic, tree variants), compiles a faster backend, and reports concrete wins (example shows 35% savings versus a baseline). Benchmarks claim better results when customer values are highly heterogeneous.
AI-written summary · about
chapters (9)
- 0:00Cost-sensitive learning intro
- 3:15Why traditional machine learning falls short
- 8:15Cost matrices in practice
- 12:06Costs in model training
- 17:21Models with business KPIs
- 23:22Max profit and thresholds
- 28:20Advanced metric API
- 35:04Ampulse solution
- 37:10Q&A
related talks
28mGet the best from your scikit-learn classifier
41mOptimizing Machine Learning Models
18mCompetitive Analysis of the Top Gradient...
39mPushing Cython to its Limits in Scikit-learn
01h 28mMastering gradient boosting with CatBoost
01h 27mCatBoost - The New Generation of Gradient Boosting
30mBeyond Grid Search: XGBoost and Optuna as the ultimate ML Optimization Combo
25mBalancing speed and accuracy in model development