leftrightcompare

# load training data
print  "Loading training data..."
x_train_left = np.genfromtxt('leftFeature.csv', delimiter=',')
x_train_right = np.genfromtxt('rightFeature.csv', delimiter=',')

train_labels_raw = np.genfromtxt('BinaryTrain_sbj_list.csv', delimiter=',', skip_header=1)
y_train = train_labels_raw[:, 1]
perf_left = []
perf_right = []
acc_left = []
acc_right = []
for i in range(5):




 #feature selection
 #print "feature selection..."
 #from sklearn import metrics
 #from sklearn.ensemble import ExtraTreesClassifier
 #model = ExtraTreesClassifier()
 #model.fit(x_train, y_train)
 # display the relative importance of each attribute
 #print(model.feature_importances_)

# adaboost classifier
# initialize the  (defaults to using shallow decision trees
# as the weak learner to boost) and optimize parameters using random search

 print "Training adaboost DT..."

 bdt1 = AdaBoostClassifier(DecisionTreeClassifier(max_depth=1),
                          n_estimators=300,learning_rate=0.8)
 bdt2 = AdaBoostClassifier(DecisionTreeClassifier(max_depth=1),
                          n_estimators=300,learning_rate=0.8)
 bdt1.fit(x_train_left, y_train)
 bdt2.fit(x_train_right, y_train)
 print "Generating CV scores for Adaboost left feature training data..."

 scores = cross_validation.cross_val_score(bdt1, x_train_left, y_train, cv=5)
 print("Accuracy: %0.2f (+/- %0.2f)" % (scores.mean(), scores.std() * 2))
 acc_left.append(scores.mean())

 print "Generating CV scores for Adaboost all train data "
 scores = cross_validation.cross_val_score(bdt2, x_train_right, y_train, cv=5)
 print("Accuracy: %0.2f (+/- %0.2f)" % (scores.mean(), scores.std() * 2))
 acc_right.append(scores.mean())

 AUC_Ada_left = []

 for train_index, test_index in cross_validation.KFold(n=150, n_folds=5, shuffle=True,
                               random_state=None):
    x_train_cv, x_test_cv = x_train_left[train_index], x_train_left[test_index]
    y_train_cv, y_test_cv = y_train[train_index], y_train[test_index]
    bdt_cv = AdaBoostClassifier(DecisionTreeClassifier(max_depth=1),
                          n_estimators=300,learning_rate=0.8)
    bdt_cv.fit(x_train_cv, y_train_cv)
    predictionAda = bdt_cv.predict(x_test_cv)

    AUC_Ada_left.append(roc_auc_score(y_test_cv, predictionAda))


 print np.mean(AUC_Ada_left)
 perf_left.append(np.mean(AUC_Ada_left))



 AUC_Ada_right = []

 for train_index, test_index in cross_validation.KFold(n=150, n_folds=5, shuffle=True,
                               random_state=None):
    x_train_cv, x_test_cv = x_train_right[train_index], x_train_right[test_index]
    y_train_cv, y_test_cv = y_train[train_index], y_train[test_index]
    bdt_cv = AdaBoostClassifier(DecisionTreeClassifier(max_depth=1),
                          n_estimators=300,learning_rate=0.8)
    bdt_cv.fit(x_train_cv, y_train_cv)
    predictionAda = bdt_cv.predict(x_test_cv)

    AUC_Ada_right.append(roc_auc_score(y_test_cv, predictionAda))


 print np.mean(AUC_Ada_right)
 perf_right.append(np.mean(AUC_Ada_right))
# set up bagging around each AdaBoost set
print "=========================="
print np.mean(perf_left)
print np.mean(acc_left)
print "=========================="
print np.mean(perf_right)
print np.mean(acc_right)