In this assignment we had to perform different clustering algorithms using different pre-processing techniques with different numbers of clusters on different evaluation parameters.
I have used Pycaret in this assignment.
The dataset that I have used is Iris Dataset.
The three clustering techniques I have used are:
- K means Clustering
- Hierarchical Clustering
- DBSCAN
The Parameters I have taken are:
- Silhouette
- Calinski-Harabasz
- Davies-Bouldin
The Pre-Processing techniques I have used are:
- Normalization
- PCA (Principle Component Analysis)
- Transform
- Scale
The absence of a 'scale' column in the DBSCAN results is because DBSCAN does not utilize distance-based similarity metrics in the same manner as KMeans or hierarchical clustering algorithms, hence it is not as directly influenced by scaling.
The graphs of the different clustering algorithms are given below: