Business Problem

👋 Welcome aboard

You’ve just joined the Fraud Intelligence team at DoorDash as a data scientist. Your teammates include engineers, fraud investigators, and operations analysts, all working together to protect the platform from abuse and ensure fair experiences for Dashers, customers, and merchants.

✍️ Project Description

DoorDash is a three-sided marketplace connecting customers, Dashers (delivery drivers), and merchants. As the platform has grown, so has the sophistication of fraud. Preventing it requires nuance — identifying bad actors without harming good ones.

Your mission is to design a custom machine learning system that flags potentially fraudulent customers. E.g. fraudsters who have taken over legitimate accounts.

Model accuracy isn’t the only concern:

• False positives can mean deactivating legitimate Dashers or blocking loyal customers — directly affecting livelihoods and satisfaction.
• False negatives can lead to widespread abuse, lost revenue, and degraded platform trust.

🎯 Key Objectives

Build a data science solution that addresses the following:

1. Fraud Model: Build a machine learning model that predicts the likelihood of fraud using the customer transaction history. Perform exploratory data analysis, data preparation, feature preparation, model selection and evaluation to build a robust model pipeline.

📊 Downloads

Boilerplate Template

To help you get started, we’ve provided a boilerplate in the form of a Jupyter Notebook.

Dataset

You’ll be working with a simulated dataset containing users and transaction data. Use these datasets to address the project’s key objectives using data science methods.