if __name__ == '__main__': if len(sys.argv) == 1: data = load_data.load_data() # Set up datasets for cross validation rs = cross_validation.ShuffleSplit(150000, n_iterations=3, test_fraction=.30) C, train_results, cv_results = prepare_params(100, 10000, 10) # Run through the cross validation iterations for train_index, cv_index in rs: train_data, cv_data = split_data(data[1], train_index, cv_index) model = Logistic_ComplexI_Model() train_X, train_y = model.load_train_data(train_data) cv_X, cv_y = model.load_data(cv_data) for c in C: print c model.train_model(train_X, train_y, {'c':c}) train_probs = model.run_model(train_X) train_results[c].append(calc_score(train_probs, train_y)) cv_probs = model.run_model(cv_X) cv_results[c].append(calc_score(cv_probs, cv_y))
if __name__ == '__main__': if len(sys.argv) == 1: data = load_data.load_data() # Set up datasets for cross validation rs = cross_validation.ShuffleSplit(150000, n_iterations=3, test_fraction=.30) C, train_results, cv_results = prepare_params(0.0000001, 1000, 50) # Run through the cross validation iterations for train_index, cv_index in rs: train_data, cv_data = split_data(data[1], train_index, cv_index) model = Logistic_Inverse_Model() train_X, train_y = model.load_train_data(train_data) cv_X, cv_y = model.load_data(cv_data) for c in C: print c model.train_model(train_X, train_y, {'c':c}) train_probs = model.run_model(train_X) train_results[c].append(calc_score(train_probs, train_y)) cv_probs = model.run_model(cv_X) cv_results[c].append(calc_score(cv_probs, cv_y))
# probs = self.clf.predict_proba(X) # return probs[:,1] if __name__ == '__main__': if len(sys.argv) == 1: data = load_data.load_data() # Set up datasets for cross validation rs = cross_validation.ShuffleSplit(150000, n_iterations=3, test_fraction=.30) C, train_results, cv_results = prepare_params(0.0000000001, 0.0001, 10) # Run through the cross validation iterations for train_index, cv_index in rs: train_data, cv_data = split_data(data[1], train_index, cv_index) model = Logistic_Pca_Model() train_X, train_y = model.load_train_data(train_data) cv_X, cv_y = model.load_data(cv_data) for c in C: print c model.train_model(train_X, train_y, {'c': c}) train_probs = model.run_model(train_X) train_results[c].append(calc_score(train_probs, train_y)) cv_probs = model.run_model(cv_X) cv_results[c].append(calc_score(cv_probs, cv_y))
# cv_test_data = self.pcan.execute(X) return (X, y) if __name__ == '__main__': if len(sys.argv) == 1: data = load_data.load_data() # Set up datasets for cross validation rs = cross_validation.ShuffleSplit(150000, n_iterations=3, test_fraction=.30) C, train_results, cv_results = prepare_params(100, 10000, 10) # Run through the cross validation iterations for train_index, cv_index in rs: train_data, cv_data = split_data(data[1], train_index, cv_index) model = Logistic_ComplexI_Model() train_X, train_y = model.load_train_data(train_data) cv_X, cv_y = model.load_data(cv_data) for c in C: print c model.train_model(train_X, train_y, {'c': c}) train_probs = model.run_model(train_X) train_results[c].append(calc_score(train_probs, train_y)) cv_probs = model.run_model(cv_X) cv_results[c].append(calc_score(cv_probs, cv_y))