Assignment 9 PDF

Preview

Summary

Assignment-9...

Description

Data Mining Assignment-9 Support Vector Machines

Launch the WEKA tool, and then activate the "Explorer" environment.



Open the "iris" dataset (i.e., i.e., stored in the sub folder "data" of the installed WEKA folder). For each attribute and for each of its possible values, how many instances in each class have the feature value (i.e., the class distribution of the feature values)?

Ans.)



SepalLength

SepalWidth

PetalLength

PetalWidth

Class

4.3 - 4.814: 16 4.814 - 5.329: 30 5.329 - 5.843: 34 5.843 - 6.357: 28 6.357 - 6.871: 25 6.871 - 7.386:10 7.386 - 7.9: 7

2 - 2.3: 8 2.3 - 2.6: 16 2.6 - 2.9: 33 2.9 - 3.2: 51 3.2 - 3.5: 24 3.5 - 3.8: 12 3.8 - 4.1: 4 4.1 - 4.4: 2

1 - 2.18: 50 2.18 - 3.36: 3 3.36 - 4.54: 34 4.54 - 5.72: 47 5.72 - 6.9: 16

0.1 - 0.58: 49 0.58 - 1.06: 8 1.06 - 1.54: 41 1.54 - 2.02: 29 2.02 - 2.5: 23

Iris-setosa: 50 Iris-versicolor: 50 Iris-virginca:50

Go to the "Classify" tab. Select the SMO classifier. Choose "Percentage split" (66% for training) test mode. Run the classifier and observe the results shown in the "Classifier output" window. 1)

Write down the learned classifiers (i.e., the separating hyperplanes).

Ans.) Classifier Model:SMO Kernel used: Linear Kernel: K(x,y) = Classifier for classes: Iris-setosa, Iris-versicolor BinarySMO Machine linear: showing attribute weights, not support vectors. 0.6829 * (normalized) sepallength + -1.523 * (normalized) sepalwidth + 2.2034 * (normalized) petallength + 1.9272 * (normalized) petalwidth 0.7091 Number of kernel evaluations: 352 (70.32% cached)

Classifier for classes: Iris-setosa, Iris-virginica BinarySMO Machine linear: showing attribute weights, not support vectors.

+ + + -

0.5886 * (normalized) sepallength -0.5782 * (normalized) sepalwidth 1.6429 * (normalized) petallength 1.4777 * (normalized) petalwidth 1.1668

Number of kernel evaluations: 284 (68.996% cached) Classifier for classes: Iris-versicolor, Iris-virginica BinarySMO Machine linear: showing attribute weights, not support vectors. + + + -

0.3176 * (normalized) sepallength -0.863 * (normalized) sepalwidth 3.0543 * (normalized) petallength 4.0815 * (normalized) petalwidth 4.5924

Number of kernel evaluations: 453 (61.381% cached) 2)

How many instances are incorrectly classified?

Ans.) Incorrectly Classified Instances: 2 = 3.9216%

3)

What is the MAE (mean absolute error) made by the learned classifiers?

Ans.) Mean Absolute Error (MAE): 0.2309

4)

What is the RMSE (root mean squared error) made by the learned classifiers?

Ans.) Root Mean Squared Error (RMSE): 0.2877

5)

Visualize the errors made by the learned classifiers. In the plot, see the detailed information of the incorrectly classified test instances.

Ans.)

Visualizing errors made by the classifier



Now, in the "Test options" panel select the "Cross-validation" option (10 folds). Run the classifier and observe the results shown in the "Classifier output" window. 1)

Write down the learned classifiers (i.e., the separating hyperplanes).

Ans.) Classifier Model:SMO Kernel used: Linear Kernel: K(x,y) = Classifier for classes: Iris-setosa, Iris-versicolor BinarySMO Machine linear: showing attribute weights, not support vectors. + + + -

0.6829 * (normalized) sepallength -1.523 * (normalized) sepalwidth 2.2034 * (normalized) petallength 1.9272 * (normalized) petalwidth 0.7091

Number of kernel evaluations: 352 (70.32% cached) Classifier for classes: Iris-setosa, Iris-virginica BinarySMO Machine linear: showing attribute weights, not support vectors.

+ + + -

0.5886 * (normalized) sepallength -0.5782 * (normalized) sepalwidth 1.6429 * (normalized) petallength 1.4777 * (normalized) petalwidth 1.1668

Number of kernel evaluations: 284 (68.996% cached) Classifier for classes: Iris-versicolor, Iris-virginica BinarySMO Machine linear: showing attribute weights, not support vectors.

+

0.3176 * (normalized) sepallength -0.863 * (normalized) sepalwidth

+ + -

3.0543 * (normalized) petallength 4.0815 * (normalized) petalwidth 4.5924

Number of kernel evaluations: 453 (61.381% cached) 2)

How many instances are incorrectly classified?

Ans.) Incorrectly Classified Instances: 6 = 4%

3)

Compare the MAE (mean absolute error) made by the learned classifiers to that observed in the previous experiment?

Ans.) Mean Absolute Error (MAE): 0.2311

4)

Compare the RMSE (root mean squared error) made by the learned classifiers to that observed in the previous experiment?

Ans.) Root Mean Squared Error (RMSE): 0.288

5)

Visualize the errors made by the learned classifiers. In the plot, see the detailed information of the incorrectly classified test instances.

Ans.)

Visualizing errors made by the classifier...