INNER CODE UNIT · Python

load_csv_data

ankonzoid/LearningX · basic_ML/classification/decision_tree.py:112

    def load_csv_data(data_csv_filename, mode="clf", verbose=False):
        if verbose:
            print("Loading data from '{}' (mode={})...".format(data_csv_filename, mode))
        df = pd.read_csv(data_csv_filename)  # dataframe
        df_header = df.columns.values  # header
        header = list(df_header)
        N, d = len(df), len(df_header) - 1
        X = np.array(df.drop(['y'], axis=1))  # extract X by dropping y column
        y_raw = np.array(df['y'])  # extract y

        y_classes_raw = sorted(list(set(y_raw)))
        y_classes_dict = {}
        for i, y_class_i in enumerate(y_classes_raw):
            y_classes_dict[y_class_i] = i

        y = np.array([y_classes_dict[y_i] for y_i in y_raw])
        y_classes = list(set(y))
        assert X.shape == (N, d)  # check X.shape

View source record →