def doGridSearch(self):
# params = {'base_estimator__max_depth': 10}
# self.classifier.set_params(**params)
# parameters = {'base_estimator__max_depth': np.arange(2, 12, 1)}
# parameters = {'base_estimator__max_depth':np.arange(2, 12, 1),
# 'base_estimator__min_samples_split':np.arange(2, 100, 1)}
# parameters = {'base_estimator__max_depth': [7,8,9,10], 'learning_rate':[.1, .01, .001, .0001], 'n_estimators':[100, 110,120,130,150,170,200]}
parameters = {
'base_estimator__max_depth': [3, 4, 5],
'learning_rate': np.arange(.0001, .9, .05),
'n_estimators': np.arange(10, 300, 20)
}
clf = GridSearchCV(self.classifier, parameters, refit=True, cv=self.cv)
clf.fit(self.X_train, self.y_train)
a = pd.DataFrame(clf.best_params_, index=[0])
a.to_csv('{}/images/{}/{}/{}_{}_gridsearch.csv'.format(
'.', self.datasetName, self.algoname, self.datasetName,
self.algoname))
self.classifier = clf.best_estimator_
self.classifier.set_params(**clf.best_params_)
timing.getTimingData(self.X_train,
self.y_train,
self.classifier,
self.algoname,
self.datasetName,
prefix='GS')
self.classifier.fit(self.X_train, self.y_train)
self.generateFinalAccuracy(prefix='GS')
self.generateFinalLC(prefix='GS')
def generateFinalModel(self):
params = {'n_neighbors': 1, 'metric': 'manhattan', 'weights': 'uniform'}
self.classifier.set_params(**params)
timing.getTimingData(self.X_train, self.y_train,self.classifier,self.algoname, self.datasetName)
self.classifier.fit(self.X_train, self.y_train)
self.generateFinalAccuracy()
self.generateFinalLC()
def doGridSearch(self):
# parameters = {"C": np.arange(0.001, 2.5, 0.25)}
parameters = {
"C": np.arange(0.001, 2.5, 0.25),
"gamma": np.arange(1 / self.features.shape[1], 2.1, 0.1)
}
clf = GridSearchCV(self.classifier, parameters, refit=True, cv=self.cv)
clf.fit(self.X_train, self.y_train)
js = json.dumps(clf.best_params_)
filename = '{}/images/{}/{}/{}_{}_GSP.json'.format(
'.', self.datasetName, self.algoname, self.datasetName,
self.algoname)
f = open(filename, "w")
f.write(js)
f.close()
self.classifier = clf.best_estimator_
self.classifier.set_params(**clf.best_params_)
timing.getTimingData(self.X_train,
self.y_train,
self.classifier,
self.algoname,
self.datasetName,
prefix='GS')
self.classifier.fit(self.X_train, self.y_train)
self.generateFinalAccuracy(prefix='GS')
self.generateFinalLC(prefix='GS')
def doGridSearch(self):
dimension = self.features.shape[1]
# parameters = {'solver':['lbfgs', 'adam'], "alpha": np.logspace(-5, 3, 20),'max_iter':np.arange(2, 1000, 10),
# 'hidden_layer_sizes': [(h,) * l for l in [1, 2, 3] for h in [dimension, dimension // 2, dimension * 2]]}
parameters = {
'alpha':
np.logspace(-5, 3, 20),
'max_iter': [2**x for x in range(10)],
'hidden_layer_sizes':
[(h, ) * l for l in [1, 2, 3]
for h in [dimension, dimension // 2, dimension * 2]]
}
clf = GridSearchCV(self.classifier, parameters, refit=True, cv=self.cv)
clf.fit(self.X_train, self.y_train)
js = json.dumps(clf.best_params_)
filename = '{}/images/{}/{}/{}_{}_GSP.json'.format(
'.', self.datasetName, self.algoname, self.datasetName,
self.algoname)
f = open(filename, "w")
f.write(js)
f.close()
self.classifier = clf.best_estimator_
self.classifier.set_params(**clf.best_params_)
timing.getTimingData(self.X_train,
self.y_train,
self.classifier,
self.algoname,
self.datasetName,
prefix='GS')
self.classifier.fit(self.X_train, self.y_train)
self.generateFinalAccuracy(prefix='GS')
self.generateFinalLC(prefix='GS')
def generateFinalModel(self):
params = {"C": 2.251, "gamma": 0.2625,'class_weight':'balanced'}
self.classifier.set_params(**params)
timing.getTimingData(self.X_train, self.y_train,self.classifier,self.algoname, self.datasetName)
self.classifier.fit(self.X_train, self.y_train)
self.generateFinalAccuracy()
self.generateFinalLC()
def generateFinalModel(self):
params = {'max_depth': 22, 'class_weight': 'balanced'}
self.classifier.set_params(**params)
timing.getTimingData(self.X_train, self.y_train, self.classifier,
self.algoname, self.datasetName)
self.classifier.fit(self.X_train, self.y_train)
self.generateFinalAccuracy()
self.generateFinalLC()
def generateFinalModel(self):
params = {
'base_estimator__max_depth': 20,
'n_estimators': 100,
'learning_rate': .01
}
self.classifier.set_params(**params)
timing.getTimingData(self.X_train, self.y_train, self.classifier,
self.algoname, self.datasetName)
self.classifier.fit(self.X_train, self.y_train)
self.generateFinalAccuracy()
self.generateFinalLC()
def generateFinalModel(self):
params = {
"alpha": 0.003359818286283781,
"hidden_layer_sizes": [32, 32, 32],
"max_iter": 128
}
self.classifier.set_params(**params)
timing.getTimingData(self.X_train, self.y_train, self.classifier,
self.algoname, self.datasetName)
self.classifier.fit(self.X_train, self.y_train)
self.generateFinalAccuracy()
self.generateFinalLC()
def generateFinalModel(self):
params = {
"alpha": 0.0004832930238571752,
"hidden_layer_sizes": [20, 20, 20],
"max_iter": 256
}
self.classifier.set_params(**params)
timing.getTimingData(self.X_train, self.y_train, self.classifier,
self.algoname, self.datasetName)
self.classifier.fit(self.X_train, self.y_train)
self.generateFinalAccuracy()
self.generateFinalLC()
def doGridSearch(self):
parameters = {"n_neighbors":np.arange(1,50,1), "metric":['manhattan', 'chebyshev', 'euclidean']}
clf = GridSearchCV(self.classifier, parameters, refit=True, cv=self.cv)
clf.fit(self.X_train, self.y_train)
a = pd.DataFrame(clf.best_params_, index=[0])
a.to_csv('{}/images/{}/{}/{}_{}_gridsearch.csv'.format('.', self.datasetName, self.algoname,
self.datasetName, self.algoname))
self.classifier = clf.best_estimator_
self.classifier.set_params(**clf.best_params_)
timing.getTimingData(self.X_train, self.y_train, self.classifier, self.algoname, self.datasetName,prefix='GS')
self.classifier.fit(self.X_train, self.y_train)
self.generateFinalAccuracy(prefix='GS')
self.generateFinalLC(prefix='GS')
def doGridSearch(self):
parameters = {
'max_depth': np.arange(8, 24, 1),
"min_samples_split": np.arange(2, 100, 5)
}
clf = GridSearchCV(self.classifier, parameters, refit=True, cv=self.cv)
clf.fit(self.X_train, self.y_train)
a = pd.DataFrame(clf.best_params_, index=[0])
a.to_csv('{}/images/{}/{}/{}_{}_gridsearch.csv'.format(
'.', self.datasetName, self.algoname, self.datasetName,
self.algoname))
self.classifier = clf.best_estimator_
self.classifier.set_params(**clf.best_params_)
timing.getTimingData(self.X_train,
self.y_train,
self.classifier,
self.algoname,
self.datasetName,
prefix='GS')
self.classifier.fit(self.X_train, self.y_train)
self.generateFinalAccuracy(prefix='GS')
self.generateFinalLC(prefix='GS')
def doGridSearch(self):
parameters = {
'base_estimator__max_depth': np.arange(5, 22, 1),
'learning_rate': np.arange(.0001, .9, .05),
'n_estimators': np.arange(10, 300, 20)
}
clf = GridSearchCV(self.classifier, parameters, refit=True, cv=self.cv)
clf.fit(self.X_train, self.y_train)
a = pd.DataFrame(clf.best_params_, index=[0])
a.to_csv('{}/images/{}/{}/{}_{}_gridsearch.csv'.format(
'.', self.datasetName, self.algoname, self.datasetName,
self.algoname))
self.classifier = clf.best_estimator_
self.classifier.set_params(**clf.best_params_)
timing.getTimingData(self.X_train,
self.y_train,
self.classifier,
self.algoname,
self.datasetName,
prefix='GS')
self.classifier.fit(self.X_train, self.y_train)
self.generateFinalAccuracy(prefix='GS')
self.generateFinalLC(prefix='GS')