The Machine Learning “Advent Calendar” Day 8: Isolation Forest in Excel

The Isolation Forest algorithm is an unsupervised machine learning method used for anomaly detection. It operates by building numerous random decision trees to isolate individual data points, based on the principle that anomalies are easier to isolate than normal points. The number of splits required to isolate a point, known as its path length or depth, is averaged across all trees to create a stable measure. This average depth is then normalized into an anomaly score, where lower depths result in scores closer to 1, indicating a likely anomaly. The content provides a detailed, step-by-step walkthrough of how to implement this entire process within Microsoft Excel.