econml

Neuro-Symbolic-Causal AI - Project Chimera | 🌌 An open research project exploring formal verification of AI agent decisions, combining symbolic reasoning, causal inference, and runtime policy enforcement.

Cusal Inference applied to timeseries, uses an event database to generate a timeseries of the outcome given a sliding window containing events. Useful to add causal outcomes of events into multivariate timeseries forecasting models.

DeepEcon: Your one-stop Python package for econometric algorithms.

A complete end-to-end AI experimentation & causal inference project using A/B testing, X-Learner, CATE estimation, and uplift segmentation on 1.5M+ synthetic SaaS behavioral records. Includes statistical analysis, causal ML workflow, uplift modeling, feature importance, and business-ready insights for AI feature rollout & monetization.

Causal Forest DML analysis of racial approval penalties in U.S. mortgage lending | 42M HMDA applications, 2020-2024 | Under review at Journal of Financial Services Research

Uplift modeling & retention API — predicting which customers a retention contact can actually save. Live on Render

Causal inference dashboard — estimating price elasticity of customer engagement on 258K Amazon products using Double ML + Causal Forest"

Causal ML pipeline for e-commerce dynamic pricing — Double Machine Learning for unbiased price elasticity, LightGBM demand forecasting (MAPE=0.418, R²=0.055), and a FastAPI pricing service delivering +30% revenue lift across 49,677 SKUs from 32M+ transactions.

Internship Project on Causal Inference (The causal effect of multi-level treatment of intervention using observational data).

Causal analysis to predict factors affecting employee resignation using the Causal Inference method on company datasets.

End-to-end causal inference study estimating the effect of smoking cessation on substantial weight gain using propensity methods, DoWhy, and doubly robust EconML estimators.

A modular Python benchmark for uplift modeling on the Criteo dataset, comparing S-Learner, T-Learner, X-Learner, DR-Learner, Causal Forest, and response-model targeting policies.

Análise para responder se a hora extra realmente causa saída de funcionários ou se outros fatores como cargo e salário explicam essa relação. Usando três métodos independentes de estimação causal, o efeito direto da hora extra foi de +21,1% na rotatividade — confirmado em testes de robustez.

Causal inference on Yelp reviews using Double Machine Learning to estimate the effect of star ratings, review length, and Elite badge status on useful votes.

Causal inference analysis of ICU beta-blocker treatment effects using propensity matching, IPW, doubly robust estimation, Double ML, and Causal Forest on eICU data

Causal inference for promotional targeting: who should receive the email? Five CATE estimators evaluated by Qini & SNIPS policy value on Hillstrom 2008.

End-to-end uplift modeling pipeline on the Criteo dataset. Compares T/S/X-Learner and Causal Forest to estimate heterogeneous treatment effects for budget-constrained marketing targeting.

ML a serviço da inferência — Double/Debiased ML para efeitos heterogêneos e forecasting hierárquico com reconciliação MinT no dataset M5.

estimate CATE and deploy uplift‑based targeting models that replace or augment traditional A/B testing.

Low-latency microservice using Causal Inference (Calibrated LightGBM T-Learner) and Risk-Adjusted Expected Monetary Value (EMV) decision gating to optimize e-commerce gross margins. Features dual Client and Server-Side tracking, zero-Pandas NumPy inference, async Pub/Sub streaming to BigQuery, FastAPI, and Cloud Run CI/CD.