'\u03BB',
'parameters',
col_names_nn_Keras_regressor,
row_names_nn_Keras_regressor,
True,
savefig=True,
figname='Images/NN_reg_accuracy_' + wine_type + '.png')
#refit best NN regressor
print(clf.best_params_)
nnKerasRegBest = KerasRegressor(build_fn=build_network,
n_outputs=1,
output_activation=None,
loss="mean_squared_error",
verbose=0)
nnKerasRegBest.set_params(**clf.best_params_)
hist = nnKerasRegBest.fit(Xtrain, ytrain, validation_data=(Xtest, ytest))
pred_nnKerasRegBest_train = nnKerasRegBest.predict(Xtrain)
pred_nnKerasRegBest_test = nnKerasRegBest.predict(Xtest)
print('Neural network regressor MSE train: %g' %
mean_squared_error(ytrain, pred_nnKerasRegBest_train))
print('Neural network regressor MSE test: %g' %
mean_squared_error(ytest, pred_nnKerasRegBest_test))
print('Neural network regressor MAD train: %g' %
MAD(ytrain, pred_nnKerasRegBest_train))
print('Neural network regressor MAD test: %g' %
MAD(ytest, pred_nnKerasRegBest_test))
print('Neural network regressor accuracy train: %g' %
accuracy_score(ytrain, np.rint(pred_nnKerasRegBest_train)))
print('Neural network regressor accuracy test: %g' %
accuracy_score(ytest, np.rint(pred_nnKerasRegBest_test)))
