Matheus Facure

Matheus Facure is an economist, data scientist and causal inference specialist at Nubank. He currently works with adding causal inference capabilities to machine learning models with application in the credit card business, and acts as a consultant for other business areas inside Nubank, such as personal loans and marketing.

projects by Matheus Facure

Causal Inference and Personalization

5 weeks · 5-7 hours per week average · INTERMEDIATE

In this series of liveProjects, you’ll explore a variety of causal inference techniques to help optimize the discounting strategy of an e-commerce business. Causal inference is a groundbreaking field of data science that’s breaking out of academic offices and into practical application across industries. It provides a mathematical basis for determining how one variable (the “treatment”) will impact another, allowing predictions of how certain actions might affect business outcomes and KPIs.

Double Machine Learning Approach

1 week · 6-8 hours per week · INTERMEDIATE

In this liveProject, you’ll utilize machine learning for treatment effect estimation in order to help an e-commerce company deliver targeted discounting to the most profitable customers. You’ll build a model that predicts the effect of a discount on a customer following a causal model, maximizing profits for the business.

Personalizing Discounts

1 week · 6-8 hours per week · INTERMEDIATE

In this liveProject, you’ll use causal inference to investigate data on randomized discounts and determine if an e-commerce company should offer personalized discounting. You’ll estimate a different treatment effect for each customer in the hopes to see if some are positive, and figure out which customers should get what discounts. You’ll utilize Python and machine learning to build this personalization system, and implement a causal model for personalization.

Difference in Differences

1 week · 4-6 hours per week · INTERMEDIATE

In this liveProject, you’ll use difference-in-differences as an alternative to A/B testing to assess the discounting strategy of an e-commerce company. This approach compares growth between a group that has been experimented on, and data from a control group. You’ll use Python and causal inference to examine the growth trajectory in profits from customers in a high discounts group against customers that didn’t get a discount.

Regression Discontinuity Design

1 week · 4-6 hours per week · INTERMEDIATE

In this liveProject, you’ll use regression discontinuity design (RDD) as a form of natural experiment that works as an alternative to A/B testing. RDD measures treatment effects at points of discontinuity to get an idea of the effectiveness of a program without needing to rigorously A/B test it. You’ll use this technique to assess the discounting strategy of an e-commerce company, in a way that will allow you to start applying causal inference to other practical problems.

Estimating the Profitability of Discounts

1 week · 2-4 hours per week · INTERMEDIATE

In this liveProject, you’ll utilize causal inference techniques to help an e-commerce company estimate the impact of discounts on profits. You’ll learn how bias gets in our way of inferring the effect of discounts, and how to adjust for it for more accurate results. Leverage adjustment techniques, like linear regression, to adjust for bias and see how linear regression can improve the quality of your data. Your final task is to present a recommendation on whether the company should distribute its discounts or not.