Guidelines for MindBridge Data Science Challenge

Overview

In this competition you will use machine learning methods to identify anomalies in a financial data set. Anomaly detection in financial data involves identifying unusual patterns or behaviors that deviate significantly from expected norms, crucial for detecting potential risks like fraud, inefficiencies, or policy violations.

You will use a modified version of the credit card fraud detection data, where we have added additional anomalies. You should build a model to detect anomalies in the data. Data will be provided on the competition start date.

Evaluation

Submissions are evaluated based on the following criteria:

• Performance, e.g., area under the precision-recall curve (PR AUC)
• Innovation and creativity
• Presentation

Submission

Submission of your results will be on Brightspace. Link to follow.

Code requirements

A sample notebook will be provided. Please add your code to the notebook. The code should be self-contained – please include any function or class you write in the notebook. You can use common python packages such as pandas, sci-kit learn, etc.

Submission format

Jupyter notebook (Python).

The notebook should be named using the following format: DS_competition_<FirstName>_<LastName>_<StudentID>.ipynb

Rules

Participation

• Undergrad and grad students of all levels can participate (no postdocs, profs, or MindBridge employees)
• Limited to individual participation

Competition timeline

• Lecture: Wednesday, 28 May 2025
• Start date / Data availability: Wednesday, 28 May 2025 — Immediately after the lecture
• Duration: 8 weeks
• Submission deadline: Wednesday, 23 Jul 2025, 11:59pm
• Winners announcement: Wednesday, 6 August 2025

Prizes

• 1^st place: Macbook pro
• 2^nd place: ipad pro
• 3^rd place: ipad

Plus guaranteed interview for a paid MindBridge internship.