Isolation Forest Implementation - Anomaly Detection

This is a course project for MSDS689 Problem Solving with Python.

Implement the original Isolation Forest algorithm by Fei Tony Liu, Kai Ming Ting, and Zhi-Hua Zhou.

Data sets

You may want to test with:

• Kaggle credit card fraud competition data set; download, unzip to get creditcard.csv

Visualization of normal versus anomaly separation

Using plot_anomalies.py, you can see the results of the isolation forest trying to detect anomalies.

http.csv, 200 trees, 99% desired TPR

Algorithm

here are the algorithms extracted from the Liu et al paper:

	average path length c(), not the one in the original paper: scoring formula: