def support_vector_regression(input_dict):
""" Epsilon-Support Vector Regression, using the RBF kernel. """
clf = r.svr()
output_dict = {}
output_dict['out'] = clf
return output_dict
def do_regressions(x_train, y_train, x_test, y_test):
print("\nLinear regression:")
since = time.time()
mse_lr, r2_lr = regression.linear_regression(x_train, y_train, x_test,
y_test)
print('\tMean squared error linear regression: %.2f' % mse_lr)
print('\tCoefficient of determination linear regression: %.2f' % r2_lr)
print("\tExecution time:", time.time() - since, "s")
print("\nSVR:")
since = time.time()
mse_svr, r2_svr = regression.svr(x_train, y_train, x_test, y_test)
print('\tMean squared error SVR: %.2f' % mse_svr)
print('\tCoefficient of determination SVR: %.2f' % r2_svr)
print("\tExecution time:", time.time() - since, "s")
xte = te[:,0:len(te[0])-1]
bte = standard_scaler.transform(xte)
ate = min_max_scaler.transform(xte)
yte = te[:,len(te[0])-1]
mses = []
ffnnerr = 1e10
for i in range(0,10):
ffnnerr = min(ffnnerr,regression.ffnnr(atr, ytr, ate, yte))
print("MIN err " + str(ffnnerr))
mses.append(regression.mlr(xtr, ytr, xte, yte))
mses.append(regression.knn(xtr, ytr, xte, yte))
mses.append(regression.rfr(xtr, ytr, xte, yte))
mses.append(regression.svr(xtr, ytr, xte, yte))
mses.append(ffnnerr)
mses.append(regression.rbfnr(xtr, ytr, xte, yte))
import numpy as np
import regression
from sklearn import preprocessing
path = "./Bank/Bank8FM/"
tr = np.genfromtxt(path + "bank8FM.data", delimiter=' ')
te = np.genfromtxt(path + "bank8FM.test", delimiter=' ')
min_max_scaler = preprocessing.MinMaxScaler()
standard_scaler = preprocessing.StandardScaler()
xtr = tr[:,0:len(tr[0])-1]
btr = standard_scaler.fit_transform(xtr)
atr = min_max_scaler.fit_transform(xtr)
ytr = tr[:,len(tr[0])-1]
