This tutorial provides an introduction to improving business metrics using the ERUPT metric and the CausalTune library in Python. It shows a practical example and the use of the ERUPT metric for optimizing clickthrough rates. It also shows how to use ERUPT to evaluate previous experiments, as well as how to evaluate the potential effect of a future experiment with different assignments using a real business example.
This tutorial covers some of the fundamental concepts and ideas of causal AI using the DoWhy library in Python. Causal inference is quite different conceptually from standard machine learning, so most people will start out with limited background knowledge. However, basic knowledge of regression analysis and statistical modeling will be helpful for following along. By Paul Hünermund and Jermain Kaminski
We've recently collaborated with Databricks to create a solution accelerator for causal machine learning in their ecosystem. Learn how to leverage tools like MLFlow to facilitate model management in the context of an end-to-end causal machine learning application, taken from data exploration and causal discovery all the way to individualized policy recommendations.
EconML’s Doubly Robust Learner model jointly estimates the effects of multiple discrete treatments.
EconML’s DML estimator uses price variations in existing data, estimates individualized responses to incentives.
EconML’s DRIV estimator uses this experimental nudge to interpret experiments with imperfect compliance
Beyond predictive models: The causal story behind hotel booking cancellations.
Estimating the effect of a member rewards program.
Introducing the do-sampler for causal inference.
Understanding the question of why.
Predict causal effect from the intervention.