The algorithm first fits a multivariate gaussian distribution to the multi-dimensional data. Then it calculates PDF for each data point. The points whose probability is very low is flagged as anomalous data and requires further qualitative investigation.
- X.csv : 2-D data
- images folder : contains all the image used in the jupter notebook
The course available at courseera for Machine Learning by Andrew Ng is on Matlab. This repository has codes in python for the exercise. Related Material: https://towardsdatascience.com/andrew-ngs-machine-learning-course-in-python-anomaly-detection-1233d23dba95