/Spectral-Clustering

Primary LanguagePython

Spectral-Clustering

A python implementation of Spectral Clustering, using PEMS04 dataset

Usage

 python main.py

Steps of Spectral Clustering

Step1: Construct adjacency matrix by distance.csv

There are two kinds of adjacency matrices, one is 0-1 adjacency matrices which indicating the neighbor relationship between nodes, and the other is distance adjacency matrix created based on the Gaussian kernel function.

Step2: Calculate the Laplacian matrix of the adjacency matrix

Step3: Calculate the eigenvector of the Laplacian matrix

Step4: Clustering using kmeans

Step5: Calculate the calinski harabasz score