• Aim at understanding the features of different types of dermatology diseases

  • Techniques include PCA and K-Means clustering

    1. Screen plot to check importance of principal components, and how much information will be thrown away if use only first 2 -3 PCs to visualize data Scree

    2. Underlying features for the PC – are they mostly clinical or histopathological? Features

    3. 3D scatter plot for K-Means clustering of the first three PCs to entertain human eyes 3d

    4. Are there better ways to cluster this data? Looks like K-Means is the winner.compare clustering