The Model Domain is hosting a Model Competition/Training using Kaggle. You are asked to build a neural network model to predict fraudulent credit card transactions. The winner will be selected based on the model’s performance on an out-of-sample test dataset.
Objective:
The goal of this competition is to gain additional practice building neural networks.
Scoring Metric:
Given the class imbalance ratio, submissions will be evaluated using the Area Under the Precision-Recall Curve (AUPRC). Confusion matrix accuracy is not meaningful for unbalanced classification.
Deadline:
Entries can be submitted at any time, and you can track your progress on the public leaderboard. We encourage you to make multiple submissions to try to improve your score. Final scoring using the private leaderboard will be conducted on June 30, 2020.