Exploratory data analysis and clustering using pytorch

• Dataset from Kaggle - histopathologic cancer detect

Task performed

• Dataset size and class distribution

• Perform some sanity checks

  • Checking for corrupted files
  • File typing check
  • Checking image channels

• Visualizing the dataset

• Visualizing the distribution of channel pixels by class

• Identifying very dark and very light images and removing them

• Identifying duplicates in our data

• Transforming the images into a Feature Matrix

• Using KMeans to cluster our images

Technologies applied