11-02, 14:35–15:15 (America/New_York), Radio City (Room 6604)
With an average of 3.2 new papers published on Arxiv every day in 2022, causal inference has exploded in popularity, attracting large amount of talent and interest from top researchers and institutions including industry giants like Amazon or Microsoft.
There’s a very good reason for this upsurge in popularity. In our contemporary data culture we got accustomed to thinking that traditional machine learning methods can provide us with answers to any interesting business or scientific questions.
This view turns out to be incorrect. Many interesting business and scientific questions are causal in their nature and traditional machine learning methods are not suitable to address them.
In this talk, dedicated to data scientists and machine learning engineers with at least 3 years of experience, we’ll show why this is the case, we’ll introduce the fundamental tools for causal thinking and show how to translate them into code.
We’ll discuss a popular use case of churn prevention and demonstrate why only causal models should be used to solve it.
All in Python, repo included!
Traditional machine learning methods leverage associations between variables in order to learn the patterns of variability in the dataset of interest.
This is great when we want to predict the next most likely token or classify a data point. Yet, when decision-making is at stakes, these models usually cannot provide us with a clear solution.
A person with a relatively high probability of churn, might react negatively to a promotional content we send them and churn, because of this content. This scenario cannot be effectively modeled in a traditional churn prediction framework and requires a causal approach.
In the talk we’ll demonstrate why this is the case. We’ll discuss theoretical and practical underpinnings of causal models and demonstrate how to implement them in Python.
The talk is addressed to people who want to enrich their data science toolbox and learn about one of the currently hottest sub-fields of artificial intelligence.
In the talk we’ll focus on building the practical understanding of the topic and we’ll use a mixture of hands-on and theoretical approaches.
The talk is open to everyone, yet to fully enjoy the content, it’s recommend that you:
• Have a solid understanding of Python fundamentals
• Understand the basics of graph theory (nodes, directed and undirected edges)
• Understand the basics of probability theory (including conditional probability)
• Have a good understanding of standard machine learning techniques (incl. tree-based models)
The goal of this talk is to give you practical understanding of what causal modeling is and how to implement it in Python.
You’ll learn which packages to use and get access to the repository including code example(s) that you will be able to adapt to your own use cases.
Outline:
00-05 min – We have machine learning, so why even bother?
05-12 min – The Ladder of Causation
12-17 min – From statistical control to graphs and back
17-25 min – The Four Steps of Causal Inference – introduction to the DoWhy library
25-33 min – Heterogeneous treatment effects: Making it personal (with code)
33-35 min – Summary
35-40 min – Q&A
Previous knowledge expected
Aleksander Molak is a Machine Learning Researcher, Educator, Consultant and Author who gained
experience working with Fortune 100, Fortune 500, and Inc. 5000 companies across Europe, the USA,
and Israel, designing and building large scale machine learning systems.
On a mission to democratize causality for businesses and machine learning practitioners, Aleksander is a prolific writer, creator, international speaker and the author of a best selling book Causal Inference and Discovery in Python.
He’s a founder of Lesprie.io a company that provides machine learning trainings for corporate
teams, the leader of CausalPython.io community and the host of the Causal Bandits Podcast
Aleksander has provided workshops , and trainings for companies across industries, including
market leaders like Mercedes Benz innovative disruptors like e:fs TechHub, and more.