'C': np.around(np.linspace(1, 101, 100).tolist(), 0)}]
tuned_parameters_Best = [{'kernel': ['rbf'], 'gamma': [.04], 'C': [4]}]
scores = ['precision_macro', 'recall_macro', 'balanced_accuracy']
scores2 = ['precision_macro']
scr = scores2
# t_p = tuned_parameters # *****
# t_p = tuned_parameters2b
t_p = tuned_parameters_Best
labels = ['g', 'b']
clf, b_params, c_rV, c_rT, clf_l = grid_search_tester(t_p, train_set, validation_set, test_set, cv=5, scores=scr,
verbose=True, show_param_test=False, t_names=labels)
show_best_params(b_params, c_rV, c_rT)
b_avg, b_score = get_best_performance(c_rT, scr, labels, metric='precision')
print('Should use score: {:s}'.format(b_score))
print()
print('param options: ')
print(t_p)
print('# ------------------------------- use parameters: ')
print('# ', b_params[b_score])
print('# ---------------------------------------------------')
print()
print('best score: ', b_score)
print('best avg: ', b_avg)
'kernel': ['rbf'],
'gamma':
np.around(np.linspace(.0930, .0931, 10).tolist(), 5),
'C':
np.around(np.linspace(8.0, 9.0, 10).tolist(), 1)
}]
scores = ['precision_macro', 'recall_macro', 'accuracy']
scores1 = ['precision_macro', 'accuracy', 'balanced_accuracy']
scores2 = ['balanced_accuracy']
clf, b_params = grid_search_tester(tuned_parameters2,
train_set,
validation_set,
test_set,
cv=5,
scores=scores,
verbose=True)
for score in b_params:
b_param = b_params[score]
b_C = b_param['C']
b_gamma = b_param['gamma']
b_kernel = ['kernel']
print('Score: {:s}'.format(score))