def KNN(xTrain, xTest, yTrain, yTest):
print("---KNN report---")
clf = KNeighborsClassifier()
param_grid = {'n_neighbors': [x for x in np.arange(1, 30) if x % 2 == 1]}
CV = GridSearchCV(estimator=clf,
param_grid=param_grid,
cv=5,
n_jobs=CPU_CORES)
t = time.time()
CV.fit(xTrain, yTrain)
elapsed_time = time.time() - t
print("best number of neighbors:" + str(CV.best_params_['n_neighbors']))
print("training time: " + str(elapsed_time) + 's')
generateReport(CV, xTest, yTest)
def NeuralNets(xTrain, xTest, yTrain, yTest):
print("---Neural Nets report---")
param_grid = {'learning_rate_init': [10e-4, 10e-3, 10e-2]}
clf = MLPClassifier(hidden_layer_sizes=(640, ),
algorithm='sgd',
early_stopping=True,
nesterovs_momentum=False,
learning_rate='constant')
CV = GridSearchCV(estimator=clf, param_grid=param_grid, cv=5)
t = time.time()
CV.fit(xTrain, yTrain)
elapsed_time = time.time() - t
print("best learning_rate_init " +
"{0:.3f}".format(CV.best_params_['learning_rate_init']))
print("training time: " + str(elapsed_time) + 's')
generateReport(CV, xTest, yTest)
def RandomForest(xTrain, xTest, yTrain, yTest):
print("---Random Forest report---")
clf = RandomForestClassifier(n_estimators=100)
param_grid = {
'max_features':
[x for x in [1, 2, 4, 6, 8, 12, 16, 20] if x < len(xTrain[0])]
}
CV = GridSearchCV(estimator=clf,
param_grid=param_grid,
cv=5,
n_jobs=CPU_CORES)
t = time.time()
CV.fit(xTrain, yTrain)
elapsed_time = time.time() - t
print("best number of max_features:" +
str(CV.best_params_['max_features']))
print("training time: " + str(elapsed_time) + 's')
generateReport(CV, xTest, yTest)
def BoostedDecisionTree(xTrain, xTest, yTrain, yTest):
print("---Boosted Decision Tree report---")
param_grid = {
'n_estimators': [2, 4, 8, 16, 32, 64, 128, 256, 512, 1024, 2048]
}
# classify
clf = AdaBoostClassifier()
CV = GridSearchCV(estimator=clf,
param_grid=param_grid,
cv=5,
n_jobs=CPU_CORES)
t = time.time()
CV.fit(xTrain, yTrain)
elapsed_time = time.time() - t
print("best number of n_estimators:" +
str(CV.best_params_['n_estimators']))
print("training time: " + str(elapsed_time) + 's')
generateReport(CV, xTest, yTest)
def linearSVC(xTrain, xTest, yTrain, yTest):
print("---LinearSVM report---")
svc = LinearSVC(C=1.0,
class_weight=None,
max_iter=-1,
random_state=None,
tol=0.001,
verbose=False)
param_grid = {'C': [0.1, 1, 2, 3, 4, 5, 6, 7]}
CV = GridSearchCV(svc, param_grid=param_grid, cv=5, n_jobs=CPU_CORES)
t = time.time()
CV.fit(xTrain, yTrain)
elapsed_time = time.time() - t
print("best number of C:" + str(CV.best_params_['C']))
print("training time: " + str(elapsed_time) + 's')
generateReport(CV, xTest, yTest)
def XGBoost(xTrain, xTest, yTrain, yTest):
xTrain = np.array(xTrain)
xTest = np.array(xTest)
yTrain = np.array(yTrain)
yTest = np.array(yTest)
print("---XGBoost report---")
param_grid = {
'n_estimators': [2, 4, 8, 16, 32, 64, 128, 256, 512, 1024, 2048]
}
clf = xgb.XGBClassifier()
CV = GridSearchCV(estimator=clf, param_grid=param_grid, cv=5)
t = time.time()
CV.fit(xTrain, yTrain)
elapsed_time = time.time() - t
print("best number of n_estimators:" +
str(CV.best_params_['n_estimators']))
print("training time: " + str(elapsed_time) + 's')
generateReport(CV, xTest, yTest)
def SVM(xTrain, xTest, yTrain, yTest):
print("---SVM with Linear kernel (LibSVM) report---")
svc = svm.SVC(C=1.0,
cache_size=200,
class_weight=None,
coef0=0.0,
gamma='auto',
kernel='linear',
max_iter=-1,
probability=False,
random_state=None,
shrinking=True,
tol=0.001,
verbose=False)
param_grid = {'C': [0.1, 1, 2, 3, 4, 5, 6, 7]}
CV = GridSearchCV(svc, param_grid=param_grid, cv=5, n_jobs=CPU_CORES)
t = time.time()
CV.fit(xTrain, yTrain)
elapsed_time = time.time() - t
print("best number of C:" + str(CV.best_params_['C']))
print("training time: " + str(elapsed_time) + 's')
generateReport(CV, xTest, yTest)