End to End Credit Card Fraud Detection

This repository contains the procedure we followed to deploy our web app of Credit Card Fraud detection on Heroku.

Since the data for credit card fraud is not available in real form(due to confidentiality), and is availbale only in dimensionality reduced form, we will be sharing some of the test cases here.

Specification about what we used and achieved.

---

Model Training

• Architecture

  • Logistic Regression with balanced class weight.

• Inference Results

  • Accuracy: 0.976
  • Recall: 0.896

---

Web App Production

• Procfile
  • Contains the type of app.
• Requirements
  • Libraries needed to run the app.
• Templates
  • Files required for rendering purpose
• Static
  • CSS styles
• App
  • Main file which will run our Web App.

---

Web App

Our Credit Card Fraud Detection web app is available at: Credit Card Fraud Detection