# Perform Grid Search on the SVR model to find the optimal parameters to achieve a lower MSE.
# The poly kernel was excluded as it never converged to a solution. The same was done for C values 10, 100 and 1000 for the linear kernel.
from sklearn.svm import SVR
from sklearn.model_selection import GridSearchCV
regressor = SVR(kernel = 'linear')
parameters = [{'C': [1], 'kernel': ['linear']},
              {'C': [1, 10, 100, 1000], 'kernel': ['rbf']},
              {'C': [1, 10, 100, 1000], 'kernel': ['sigmoid']}]
grid_search = GridSearchCV(estimator = regressor,
                           param_grid = parameters,
                           scoring = 'neg_mean_squared_error',
                           n_jobs = -1)

# Fit GridSearchCV to the training dataset.
grid_search = grid_search.fit(X_train, y_train)