This repo contains all 1-time codes required for SMD related processing

• ServerMachineDataset folder contains the original dataset from the authors.
• Use create_smd_attack_dataset.py and create_smd_normal_dataset.py to convert the data files from txt to csv appropriately.
• The resultant dataset is smd.
• initial_eda_for_smd.py contains the initial EDA code for SMD including obtaining all-zero counts for columns, preparatory codes for box plots, and the necessary codes for pairplots
• boxplots_varianceOfMean_pairplots.ipynb is the Jupyter notebook for obtaining necessary visualization plots.

kmeans clustering code :

• kmeans_clustering.py contains the code to perform kmeans clustering on SMD dataset, and the Elbow method.
• pair-wise-euclidean-dist.py contains the code to obtain pairwise Euclidean distance between machines in cluster to form similarity chain

kruskal based clustering code :

• similarity-measure-clustering.py - Main code to obtain JS_distance, cluster based on JSD and derive similarity chains
• kruskal_algo.py - Kruskal-based clustering code
• mst_using_kruskal.py - MST to obtain similarity chains