Applying K-Means algorithm to the 20 newsgroup data (link: http://qwone.com/~jason/20Newsgroups/)
- Step 1: From the already classified data, I have selected 30 from each group and mixed them under directory
workdata - Step 2: Using tf-idf model, extract features for the documents and form the document vectors
- Step 3: Apply K-Means algo to these n-dim doc vectors
- Step 4: The labels generated for the cluster membership and the doc vectors are passed to the LDA function
- Step 5 The linear discriminant analysis process reduces the n-dim to 2-dim and plots the clusters via a scatter plot (
Cluster_Visualization.png) - Step 6 The doc list of each cluster is also output (sample available in
Clusters_of_Docs.txt)