min_max_scaler = preprocessing.MinMaxScaler(feature_range=(-1, 1)) # Normalizing data between -1 and 1 X = pd.DataFrame(min_max_scaler.fit_transform(X)) y = data.iloc[:,-1].copy() y[y == 0] = -1 # Filtering data: X_new, y_new = remove_noise(X, y) # Comparing methods: method = ["nn_clas", "parallel", "extreme_search"] print("Dataset: {}".format(data_name)) f = open("results.txt", "a") f.write("Dataset: %s\r\n" % data_name) for model in method: y_hat, y_test, result, runtime = chip_clas(X_new, y_new, method = model, kfold = 10) print(" \n Method: {0} \n Avarege AUC: {1:.4f} \n Std. Deviation {2:.4f} \n Avarege Runtime: {3:.4f} \n".format(model, result.mean()[0], result.std()[0], runtime.mean()[0])) results = {'Method': model, 'Avarege AUC': result.mean()[0], 'Std deviation': result.std()[0], 'Avarege runtime': runtime.mean()[0]} json.dump(results, f) f.write('\n\n') f.close()
X = data.iloc[:,:-1] y = data.iloc[:,-1] ''' data_name = "Banknote Auth." url = "https://archive.ics.uci.edu/ml/machine-learning-databases/00267/data_banknote_authentication.txt" data = pd.read_csv(url, header=None) X = data.iloc[:, :-1] min_max_scaler = preprocessing.MinMaxScaler( feature_range=(-1, 1)) # Normalizing data between -1 and 1 X = pd.DataFrame(min_max_scaler.fit_transform(X)) y = data.iloc[:, -1].copy() y[y == 0] = -1 # Filtering data: X_new, y_new = remove_noise(X, y) y_hat, y_test, result, runtime = chip_clas(X_new, y_new, method="extreme_search", kfold=10) print( " \n Method: {0} \n Avarege AUC: {1:.4f} \n Std. Deviation {2:.4f} \n Avarege Runtime: {3:.4f} \n" .format("pseudo_support_edges", result.mean()[0], result.std()[0], runtime.mean()[0]))