This part, contains project sample for Classification in Python using Sci-Kit tools.
In order to run the project please follow this link.
Data set for this section is different, please visit the dataset here: Clustering Dataset Web site
The goal is to group people by the comments they have made on social medias. This goal is achieved by using K-Means Clustering and Hierarchical-Clustering. And then, the result are compared together via Silhouette Score. As the results shows, 3 class worked best in both cases. However, the K-Means has higher score.
K-Means:
| N-Cluster (K) | Silhouette Score |
|---|---|
| 2 | 0.31 |
| 3 | 0.35 |
| 4 | 0.34 |
| 5 | 0.29 |
| 6 | 0.30 |
| 7 | 0.27 |
| 8 | 0.31 |
| 9 | 0.31 |
| best k = 3 | 0.35 |
Hierarchical-Clustering:
| N-Cluster | Silhouette Score |
|---|---|
| 2 | 0.24 |
| 3 | 0.29 |
| 4 | 0.25 |
| 5 | 0.25 |
| 6 | 0.25 |
| 7 | 0.27 |
| 8 | 0.29 |
| 9 | 0.28 |
| best k = 3 | 0.29 |
This is PCA plot of the clustered data for K-Means Clustering:
This is PCA plot of the clustered data for Hierarchical Clustering: