ATUS-cluster-analysis

To-do

• Clean data from ATUS repo
• Explore data transformations and scaling
• Figure out out how to account for survey weights
• Test clustering methods: hierarchical, kmeans, model-based
• Choose optimum number of clusters based on C(g), silhouette width, and gap statistic
• Create master analysis
  • clean up Marlo analysis
• Develop method to show relative change in demographics
• Analyse difference in clusters: by activity and by demographics
• Fixed label switching problem
• Clean up plots and match cluster colors across plots