An introduction to Causal Inference with Python – making accurate estimates of cause and effect from data, using PyWhy and DoWhy
Everyone wants to understand why things happen, and what would happen if you did things differently. You’ve probably read that statistics can only tell you if things are correlated, or associated, and this explains nothing about cause and effect. But in fact there’s a process called Causal Inference which does answer these questions, can tell you if A causes B and more importantly, can tell you what would happen, if… This talk will help you to frame and tackle these questions using your data and some popular Python libraries.
See this talk and many more by getting your ticket to PyCon AU now!I want a ticket!
Causal inference is used by statisticians, econometricians, and data scientists to understand cause-and-effect relationships. It seeks to determine the causal impact of a particular treatment or intervention on an outcome of interest, while accounting for potential confounding factors in a principled and accurate way. Causal Inference is often used with historical, observational data, or where it’s unethical, too expensive, or impractical to conduct a randomized controlled trial (RCT).
Causal inference methods often rely on statistical Machine Learning models and techniques to estimate causal effects. Using these models, hypothetical or counterfactual scenarios can be explored - what would have happened, under different conditions. The models also provide insight into why things happened - and which factors were most responsible. Python is one of the most popular languages for Causal Inference. PyWhy is a collection of Open-Source libraries for Causal Inference, including DoWhy and EconML. This talk will explore how to use these tools with your existing data to form and answer Causal questions, and some tips and tricks to help you interpret and validate your results. The content will be aimed at people with basic Python coding experience and doesn’t require any advanced maths.
Dave has been working on applied AI and Machine Learning for over 20 years in both academia and industry, beginning with an undergraduate degree in computer science and artificial intelligence and then a PhD in computer vision for mobile robot navigation. He has been working in engineering data science since 2019, for a variety of government and private clients. The need to find good answers for their questions drove him towards Causal Inference, the topic of his talk. In 2023, he helped to create an app called Causal Wizard, which enables subject matter experts without programming experience to make use of ML Causal Inference tools.