def errRate(self, x_test, y_test):
""" Error rate for this members local data. """
y_pred = self.predict(x_test)
if y_pred is None:
return np.nan
err_rate = compute_errorRate(y_test, y_pred)
return err_rate
def scoreLocalData(self, partial_model):
""" Error rate for partial_model/global model on this members local training data. """
if partial_model is None:
return 1
y_pred = partial_model.predict(self.__x_train)
# validation = partial_model.model.evaluate(self.__x_train,self.__y_train)
errRate = compute_errorRate(self.__y_train, y_pred)
# print("errRate: ", errRate, "validation: ", validation)
return 1 - errRate / 2
def run_experiments(X_train, y_train, X_test, y_test, X_star_train,
X_star_test):
# normalization of the training data
scaler = StandardScaler()
scaler.fit(X_train)
X_train = scaler.transform(X_train)
X_test = scaler.transform(X_test)
'''
scaler = StandardScaler()
scaler.fit(X_star_train)
X_star_train = scaler.transform(X_star_train)
X_star_test = scaler.transform(X_star_test)
'''
results = OrderedDict()
if 1:
y_predicted = fit_SVM(X_train, y_train, X_test)
results["svm"] = [util.compute_errorRate(y_test, y_predicted)]
X_train_mod = np.column_stack((X_train, X_star_train))
X_test_mod = np.column_stack((X_test, X_star_test))
scaler = StandardScaler()
scaler.fit(X_train_mod)
X_train_mod = scaler.transform(X_train_mod)
X_test_mod = scaler.transform(X_test_mod)
# print(X_train_mod.shape)
y_predicted = fit_SVM(X_train_mod, y_train, X_test_mod)
results["svm_pi"] = [util.compute_errorRate(y_test, y_predicted)]
if 1:
y_predicted = KT_LUPI(X_train, X_star_train, y_train, X_test)
results["svm_kt_lupi"] = [util.compute_errorRate(y_test, y_predicted)]
if 1:
y_predicted = RobustKT_LUPI(X_train, X_star_train, y_train, X_test)
results["svm_robust_kt_lupi"] = [
util.compute_errorRate(y_test, y_predicted)
]
print(results)
return results
#scaler = StandardScaler()
scaler.fit(X_train)
X_train = scaler.transform(X_train)
X_test = scaler.transform(X_test)
'''
scaler = StandardScaler()
scaler.fit(X_star_train)
X_star_train = scaler.transform(X_star_train)
X_star_test = scaler.transform(X_star_test)
'''
if 1:
y_predicted = fit_SVM(X_train, y_train_label, X_test)
print("SVM Error Rate:")
errRateSVM[i] = util.compute_errorRate(y_test_label, y_predicted)
print(errRateSVM[i])
if 1:
X_train_mod = np.column_stack((X_train, X_star_train))
X_test_mod = np.column_stack((X_test, X_star_test))
#scaler = StandardScaler()
scaler.fit(X_train_mod)
X_train_mod = scaler.transform(X_train_mod)
X_test_mod = scaler.transform(X_test_mod)
#print(X_train_mod.shape)
y_predicted = fit_SVM(X_train_mod, y_train_label, X_test_mod)
print("SVM with extra features Error Rate:")
errRateSVM_PI[i] = util.compute_errorRate(y_test_label, y_predicted)
print(errRateSVM_PI[i])
def errRateEnsembleModel(self, x_test, y_test):
y_pred = self.predictEnsemble(x_test)
if y_pred is None:
return np.nan
errRate = compute_errorRate(y_test, y_pred)
return errRate
def errRateFedEnsembleGlobalModel(self, x_test, y_test, model=None):
if not model:
model = self.trainEnsemble(x_test)
y_pred = self.predictFedEnsembleGlobalModel(model, x_test)
errRate = compute_errorRate(y_test, y_pred)
return errRate
def errRateGlobalModel(self, x_test, y_test, model=None):
if not model:
model = self.trainGlobalModel()
y_pred = self.predictGlobalModel(x_test, model)
errRate = compute_errorRate(y_test, y_pred)
return errRate
def alliance_test_loss(self, alliance_model):
""" Use alliance global validation data. """
print("alliance_test_loss")
y_pred = alliance_model.predict(self.x_test)
error_rate = compute_errorRate(self.y_test, y_pred)
return 1 - error_rate / 2
