def get_the_plots(tups):
ratios = []
rf_scores = []
lr_scores = []
tree_scores = []
for i in xrange(len(tups)):
n,k,p = tups[i]
X,y = yield_points(n,k,p)
d = {}
d['X'] = X
d['y'] = y
data = split_into_folds(d)
rfmodel = RandomForest_c(10, 'gini')
score = np.mean([AUC(rfmodel.predict_p(fold['X_train'], fold['y_train'], fold['X_test']), fold['y_test']) for fold in data['folds']])
rf_scores.append(score)
print score
logregmodel = LogisticRegression_c(C=0.1, penalty='l2')
score = np.mean([AUC(logregmodel.predict_p(fold['X_train'], fold['y_train'], fold['X_test']), fold['y_test']) for fold in data['folds']])
lr_scores.append(score)
print score
dtmodel = Tree_c(10, 'gini')
score = np.mean([AUC(dtmodel.predict_p(fold['X_train'], fold['y_train'], fold['X_test']), fold['y_test']) for fold in data['folds']])
tree_scores.append(score)
print score
ratios.append(p)
line1, = plot(ratios,rf_scores,'r', label='RF')
line1, = plot(ratios, lr_scores, 'g', label='LR')
line1, = plot(ratios, tree_scores, 'k', label='Tree')
legend()
xlabel('Number of points in the dataset')
ylabel('Performance score')
show()
def evaluate_and_save(method, is_method, data_file, save_file_name):
data = split_into_folds(standardise_outputs(standardise_inputs(load_dictionary(data_file))))
#print data
if is_method:
score = np.mean([AUC(method.predict_p(fold['X_train'], fold['y_train'], fold['X_test']), fold['y_test']) for fold in data['folds']])
else:
score = np.mean([method.get_stat(fold['X_train']) for fold in data['folds']])
save_file_dir = os.path.split(save_file_name)[0]
if not os.path.isdir(save_file_dir):
os.makedirs(save_file_dir)
with open(save_file_name, 'w') as save_file:
save_file.write('%s' % score)