Python implementation of the extended isolation forest algorithm for model-free anomaly detection. Made as a project in the Data Mining course at Radboud University Nijmegen, NL. Documentation can be found in the file itself.
QUICK START GUIDE: Import and fit an extended isolation forest via:
from extended_iForest import iForest
forest = iForest(X)
with X being your data as a numpy array. Forest size is set to 100 by default and subsampling size to 256.
Compute an anomaly score via:
score = forest.anomaly_score(x)
with x being one data point.
Multiple anomaly scores are most easily computed via:
anom_scores = [forest.anomaly_score(point) for point in x]