np.random.shuffle(idx)
X = X[idx]
y = y[idx]
# split the data
Xtrain = X[:nTrain, :]
ytrain = y[:nTrain]
Xtest = X[nTrain:, :]
ytest = y[nTrain:]
# train the decision tree
#modelDT = DecisionTreeClassifier()
#modelDT.fit(Xtrain,ytrain)
# train the boosted DT
modelBoostedDT = BoostedDT(numBoostingIters=100, maxTreeDepth=2)
modelBoostedDT.fit(Xtrain, ytrain)
# output predictions on the remaining data
#ypred_DT = modelDT.predict(Xtest)
ypred_BoostedDT = modelBoostedDT.predict(Xtest)
# compute the training accuracy of the model
#accuracyDT = accuracy_score(ytest, ypred_DT)
accuracyBoostedDT = accuracy_score(ytest, ypred_BoostedDT)
#print "Decision Tree Accuracy = "+str(accuracyDT)
print "Boosted Decision Tree Accuracy = " + str(accuracyBoostedDT)
# challenge data
challengeTrainingData = np.loadtxt('data/challengeTrainLabeled.dat',
delimiter=',')
np.random.seed(13) np.random.shuffle(idx) X = X[idx] y = y[idx] # split the data Xtrain = X[:nTrain,:] ytrain = y[:nTrain] Xtest = X[nTrain:,:] ytest = y[nTrain:] print ytest # print Xtrain # train the decision tree modelDT = DecisionTreeClassifier() modelDT.fit(Xtrain,ytrain) # train the boosted DT modelBoostedDT = BoostedDT(numBoostingIters=100, maxTreeDepth=2) modelBoostedDT.fit(Xtrain,ytrain) # output predictions on the remaining data ypred_DT = modelDT.predict(Xtest) ypred_BoostedDT = modelBoostedDT.predict(Xtest) # compute the training accuracy of the model accuracyDT = accuracy_score(ytest, ypred_DT) accuracyBoostedDT = accuracy_score(ytest, ypred_BoostedDT) print ','.join(str(e) for e in ypred_BoostedDT.astype(int)) print "Decision Tree Accuracy = "+str(accuracyDT) print "Boosted Decision Tree Accuracy = "+str(accuracyBoostedDT)
X = X[idx]
y = y[idx]
# split the data
Xtrain = X[:nTrain, :]
ytrain = y[:nTrain]
Xtest = X[nTrain:, :]
ytest = y[nTrain:]
# train the decision tree
modelDT = DecisionTreeClassifier()
modelDT.fit(Xtrain, ytrain)
#print ypred_DT
# train the boosted DT
modelBoostedDT = BoostedDT(numBoostingIters=100, maxTreeDepth=3)
model = AdaBoostClassifier(base_estimator=DecisionTreeClassifier(max_depth=2),
n_estimators=100,
random_state=13)
kfold = cross_validation.KFold(n=n, n_folds=2, random_state=13)
results = cross_validation.cross_val_score(model, X, y, cv=kfold)
modelBoostedDT.fit(Xtrain, ytrain)
model.fit(Xtrain, ytrain)
clf = SVC()
clf.fit(Xtrain, ytrain)
y_pred_rbf1 = clf.predict(Xtest)
scores = list()
scores_rbf = list()
k_model = BoostedDT(numBoostingIters=140, maxTreeDepth=5)
k_model.fit(X, y)
idx = np.arange(n) np.random.seed(13) np.random.shuffle(idx) X = X[idx] y = y[idx] # split the data Xtrain = X[:nTrain,:] ytrain = y[:nTrain] Xtest = X[nTrain:,:] ytest = y[nTrain:] # train the decision tree modelDT = DecisionTreeClassifier() modelDT.fit(Xtrain,ytrain) # train the boosted DT modelBoostedDT = BoostedDT(numBoostingIters=200, maxTreeDepth=3) modelBoostedDT.fit(Xtrain,ytrain) # output predictions on the remaining data ypred_DT = modelDT.predict(Xtest) ypred_BoostedDT = modelBoostedDT.predict(Xtest) # compute the training accuracy of the model accuracyDT = accuracy_score(ytest, ypred_DT) accuracyBoostedDT = accuracy_score(ytest, ypred_BoostedDT) print "Decision Tree Accuracy = "+str(accuracyDT) print "Boosted Decision Tree Accuracy = "+str(accuracyBoostedDT)
filename = 'data/challengeTestUnlabeled.dat'
X2 = np.loadtxt(filename, delimiter=',')
#n,d = X.shape
#nTrain = 0.50*n #training on 50% of the data
# split the data
#Xtrain = X[:nTrain,:]
#ytrain = y[:nTrain]
#Xtest = X[nTrain:,:]
#ytest = y[nTrain:]
Xtest_unlabeled = X2
# train the boosted DT
modelBoostedDT = BoostedDT(numBoostingIters=500, maxTreeDepth=6)
modelBoostedDT.fit(X,y)
# output predictions on the remaining data
#ypred_BoostedDT = modelBoostedDT.predict(X)
#accuracyBoostedDT = accuracy_score(y, ypred_BoostedDT)
#print 'Calculated boostedDT accuracy: ',accuracyBoostedDT
#output predictions for unlabeled data
ypred_unlabeled_BoostedDT = modelBoostedDT.predict(Xtest_unlabeled)
f = open('data/predictions-BoostedDT.dat', 'w')
predict_string = ''
for i in xrange(len(ypred_unlabeled_BoostedDT)):
if i == 0:
n,d = Xdata.shape
nTrain = 0.5*n
idx = np.arange(n)
np.random.seed(22)
np.random.shuffle(idx)
Xdata = Xdata[idx]
ydata = ydata[idx]
boost_iter_list = []
for i in [100,1000,10000]:
depth_list = []
for j in [1,2,3]:
test_accuracy_list = []
modelBoostedDT = BoostedDT(numBoostingIters=i, maxTreeDepth=j)
kf = KFold(2000, n_folds=10)
for train_index, test_index in kf:
Xtrain, Xtest = Xdata[train_index], Xdata[test_index]
ytrain, ytest = ydata[train_index], ydata[test_index]
modelBoostedDT.fit(Xtrain,ytrain)
test_ypred_BoostedDT = modelBoostedDT.predict(Xtest)
test_accuracyBoostedDT = accuracy_score(ytest, test_ypred_BoostedDT)
test_accuracy_list.append(test_accuracyBoostedDT)
depth_list.append(np.mean(test_accuracy_list))
boost_iter_list.append(depth_list)
'''
output
[[0.031000000000000007, 0.016499999999999997, 0.017000000000000005],
[0.027000000000000003, 0.019000000000000003, 0.013500000000000002],
[0.026500000000000003, 0.020500000000000001, 0.013000000000000001]]
np.random.shuffle(idx) X = X[idx] y = y[idx] # split the data Xtrain = X[:nTrain,:] ytrain = y[:nTrain] Xtest = X[nTrain:,:] ytest = y[nTrain:] # train the decision tree modelDT = DecisionTreeClassifier() modelDT.fit(Xtrain,ytrain) # train the boosted DT modelBoostedDT = BoostedDT(numBoostingIters=100, maxTreeDepth=2) modelBoostedDT.fit(Xtrain,ytrain) # output predictions on the remaining data ypred_DT = modelDT.predict(Xtest) ypred_BoostedDT = modelBoostedDT.predict(Xtest) # compute the training accuracy of the model accuracyDT = accuracy_score(ytest, ypred_DT) accuracyBoostedDT = accuracy_score(ytest, ypred_BoostedDT) print "Decision Tree Accuracy = "+str(accuracyDT) print "Boosted Decision Tree Accuracy = "+str(accuracyBoostedDT) # predict data/challengeTrainLabeled.dat newBoosted = BoostedDT(numBoostingIters=100, maxTreeDepth=2)
