Running Clustering Algorithm in Weka

1. Running Clustering Algorithm in Weka Presented by Rachsuda Jiamthapthaksin Computer Science Department University of Houston

2. What is Weka? • Data mining software in Java • Supervised learning (classification) • Unsupervised learning (clustering) • Tools • Exploration • Visualization • Experiment • Statistical summary

3. Download Weka • http://www.cs.waikato.ac.nz/ml/weka/ • Window(weka-3-5-6jre.exe) • Linux

4. Getting Start

5. Memory Limitation in Weka • Run Chooser from DOS to increase memory • C:\> java -Xmx128m -classpath .;/progra~1/weka-3-5/weka.jar weka.gui.GUIChooser

6. Weka GUI

7. Explorer

8. Open Files (.csv, .arff)

9. Dataset’s Description Dataset’s statistics Attributes

10. Remove Class Attribute Non-class attributes

11. Select A Clustering Algorithm

12. Select A Clustering Algorithm

13. Select A Clustering Algorithm

14. Parameters’ Setting

15. Run A Clustering Algorithm

16. DBSCAN Results === Run information === Scheme: weka.clusterers.DBScan -E 0.9 -M 6 -I weka.clusterers.forOPTICSAndDBScan.Databases.SequentialDatabase -D weka.clusterers.forOPTICSAndDBScan.DataObjects.EuclidianDataObject Relation: iris-weka.filters.unsupervised.attribute.Remove-R5 Instances: 150 Attributes: 4 sepallength sepalwidth petallength petalwidth Test mode: evaluate on training data === Model and evaluation on training set === DBScan clustering results ======================================================================================== Clustered DataObjects: 150 Number of attributes: 4 Epsilon: 0.9; minPoints: 6 Index: weka.clusterers.forOPTICSAndDBScan.Databases.SequentialDatabase Distance-type: weka.clusterers.forOPTICSAndDBScan.DataObjects.EuclidianDataObject Number of generated clusters: 1 Elapsed time: .06 ( 0.) 5.1,3.5,1.4,0.2 --> 0 ( 1.) 4.9,3,1.4,0.2 --> 0 ( 2.) 4.7,3.2,1.3,0.2 --> 0 ( 3.) 4.6,3.1,1.5,0.2 --> 0 ( 4.) 5,3.6,1.4,0.2 --> 0 … (146.) 6.3,2.5,5,1.9 --> 0 (147.) 6.5,3,5.2,2 --> 0 (148.) 6.2,3.4,5.4,2.3 --> 0 (149.) 5.9,3,5.1,1.8 --> 0 Clustered Instances 0 150 (100%)

17. Simplify A Tested Dataset

18. Simplify A Tested Dataset

19. Parameters’ Setting

20. DBSCAN Clustering Results === Run information === Scheme: weka.clusterers.DBScan -E 0.3 -M 50 -I weka.clusterers.forOPTICSAndDBScan.Databases.SequentialDatabase -D weka.clusterers.forOPTICSAndDBScan.DataObjects.EuclidianDataObject Relation: iris-weka.filters.unsupervised.attribute.Remove-R1-2,5 Instances: 150 Attributes: 2 petallength petalwidth Test mode: evaluate on training data === Model and evaluation on training set === DBScan clustering results ======================================================================================== Clustered DataObjects: 150 Number of attributes: 2 Epsilon: 0.3; minPoints: 50 Index: weka.clusterers.forOPTICSAndDBScan.Databases.SequentialDatabase Distance-type: weka.clusterers.forOPTICSAndDBScan.DataObjects.EuclidianDataObject Number of generated clusters: 2 Elapsed time: .03 ( 0.) 1.4,0.2 --> 0 ( 1.) 1.4,0.2 --> 0 ( 2.) 1.3,0.2 --> 0 ( 3.) 1.5,0.2 --> 0 … (146.) 5,1.9 --> 1 (147.) 5.2,2 --> 1 (148.) 5.4,2.3 --> 1 (149.) 5.1,1.8 --> 1 Clustered Instances 0 50 ( 33%) 1 100 ( 67%)

21. Run k-Means in Weka

22. Parameters’ Setting

23. k-Means Clustering Results === Run information === Scheme: weka.clusterers.SimpleKMeans -N 2 -S 10 Relation: iris-weka.filters.unsupervised.attribute.Remove-R1-2,5 Instances: 150 Attributes: 2 petallength petalwidth Test mode: evaluate on training data === Model and evaluation on training set === kMeans ====== Number of iterations: 6 Within cluster sum of squared errors: 5.179687509974782 Cluster centroids: Cluster 0 Mean/Mode: 4.906 1.676 Std Devs: 0.8256 0.4248 Cluster 1 Mean/Mode: 1.464 0.244 Std Devs: 0.1735 0.1072 Clustered Instances 0 100 ( 67%) 1 50 ( 33%)

24. ArffViewer: Convert Dataset’s Extension

25. Open A Dataset’s file

26. Select A Dataset’s File

27. View the Dataset

28. Manipulate the Dataset (Optional)

29. Save As .Arff File

30. Weka Documentation