Calculate the entropy of each cluster and the overall entropy over the clusters. Interpret the per cluster entropy and the overall entropy value. Classes to cluster evaluation Show the

Assignment Task

Instructions

Use Weka Explorer to apply k-means clustering on the dataset that presents a class label. Use the dataset you are using for assignment 2. Clustering should be done by ignoring the class label.

Clustering using K-means:

• Under preprocess tab, load the dataset.
• Under Cluster tab, choose Kmeans. Make sure the number of clusters corresponds to the number of class levels you have for the class attribute.
• Click on the Ignore attribute, and select the class attribute.

Cluster evaluation through inspection

• Show the final cluster centroids and the within-cluster sum of squared errors.
• Interpret the quality of the clusters through inspection.
• What does each cluster represent?
• How are the clusters similar and how they are different?
• Report each cluster characteristics.

Cluster evaluation through evaluation measures

• Calculate the entropy of each cluster and the overall entropy over the clusters.
• Interpret the per cluster entropy and the overall entropy value.

Classes to cluster evaluation

• Show the classes to clusters evaluation results.