__author__ = 'stanley' from src.Utility.CrossValidation import CorssValidation from numpy import * from sklearn import svm nbaData = genfromtxt('../../NBA2012-15/Classification/NBA12_14.csv', delimiter=',') nba15test = genfromtxt('../../NBA2012-15/Classification/NBA15.csv', delimiter=',') # SVM may perform better with standardize features # For easy implementation, I convert data to the base 10 logarithm nba15test_scaled = nba15test nba15test_scaled[:, 1:] = log10(nba15test_scaled[:, 1:]) label = nbaData[:, 0] features = log10(nbaData[:, 1:]) classifier = svm.SVC(kernel='rbf', probability=True, C=1) #CV validation = CorssValidation() validation.cv(features, label, classifier, nba15test, 'ROC for RBF SVM C=1')
from numpy import * nbaData = genfromtxt('../../NBA2012-15/Classification/NBA12_14.csv', delimiter=',') nba15test = genfromtxt('../../NBA2012-15/Classification/NBA15.csv', delimiter=',') label = nbaData[:,0] features = nbaData[:,1:] # show 2D result classifier = LDA(n_components=2) f_reduced = classifier.fit(features,label).transform(features) show = ScatterWithHistPlot() show.plot(f_reduced, label) #CV validation = CorssValidation() validation.cv(features,label,classifier,nba15test,'ROC for LDA')
__author__ = 'stanley' from src.Utility.CrossValidation import CorssValidation from sklearn import linear_model from numpy import * nbaData = genfromtxt('../../NBA2012-15/Classification/NBA12_14.csv', delimiter=',') nba15test = genfromtxt('../../NBA2012-15/Classification/NBA15.csv', delimiter=',') label = nbaData[:,0] features = nbaData[:,1:] # May perform better with standardize features # For easy implementation, I convert data to the base 10 logarithm nba15test_scaled = nba15test nba15test_scaled[:,1:] = log10(nba15test_scaled[:,1:]) label = nbaData[:,0] features = log10(nbaData[:,1:]) classifier =linear_model.LogisticRegression(penalty='l2',C=100) #CV validation = CorssValidation() validation.cv(features,label,classifier,nba15test,'ROC for Logistic l2 C=100')
__author__ = 'stanley' from src.Utility.CrossValidation import CorssValidation from numpy import * from sklearn import svm nbaData = genfromtxt('../../NBA2012-15/Classification/NBA12_14.csv', delimiter=',') nba15test = genfromtxt('../../NBA2012-15/Classification/NBA15.csv', delimiter=',') # SVM may perform better with standardize features # For easy implementation, I convert data to the base 10 logarithm nba15test_scaled = nba15test nba15test_scaled[:,1:] = log10(nba15test_scaled[:,1:]) label = nbaData[:,0] features = log10(nbaData[:,1:]) classifier = svm.SVC(kernel='rbf', probability=True, C=1) #CV validation = CorssValidation() validation.cv(features,label,classifier,nba15test, 'ROC for RBF SVM C=1')