Breast Cancer Classification and Evaluation

Breast cancer classification and evaluation of classifiers using k-fold Cross-Validation.

Dataset

The dataset used is Wisconsin Breast Cancer (Original) Data Set by UC Irvine Machine Learning Repository.

Classification algorithms

• Discriminant Analysis

• K-Nearest Neighbors

• Naive Bayes

• Support Vector Machine

• Decision Tree

Evaluation

Classification Algorithm	Accuracy	Sensitivity	Specificity
Discriminant Analysis (Linear)	0.959943	0.978166	0.925311
Discriminant Analysis (Mahalanobis)	0.899857	0.847162	1.000000
K-Nearest Neighbor (NumNeighbors = 5)	0.965665	0.971616	0.954357
K-Nearest Neighbor (NumNeighbors = 25)	0.962804	0.978166	0.933610
Naive Bayes (Gaussian Distribution)	0.959943	0.951965	0.975104
Naive Bayes (Kernel Distribution)	0.964235	0.971616	0.950207
Support Vector Machine (BoxConstraint = 1)	0.967096	0.973799	0.954357
Support Vector Machine (BoxConstraint = 10)	0.962804	0.967249	0.954357
Decision Tree (AlgorithmForCategorical = Exact)	0.928469	0.941048	0.904564
Decision Tree (AlgorithmForCategorical = PCA)	0.942775	0.956332	0.917012