Breast Cancer

pink ribbon

For this project, I decided to take a different approach. Instead of creating a classification model, I implemented a K Means Clustering (KMC).

This is so as to visualize the causes of breast cancer in women.

3D Breast cancer distibution using Plotly

From the above diagram, it is obvious that the mean breast texture and perimeter are good indicators of malignant tumors.

Other improvements:

• use of PCA to get meaningful features before performing KMC
• use of cross-validation (Grid Search CV) to improve clustering

Improvements to be made:

• cross validating with classification methods to see how well clustering did
• cross-comparison with other clustering methods (e.g. Hirearchical Clustering, DBSCAN etc.)
  github notebook