def convert_sparse_matrix()

in src/pipeline/prepare.py [0:0]

• 17 lines of code
• 3 McCabe index (conditional complexity)

def convert_sparse_matrix(df, nb_rows, nb_customer, nb_products):
    # dataframe to array
    df_val = df.values

    # determine feature size
    nb_cols = nb_customer + nb_products
    print("# of rows = {}".format(str(nb_rows)))
    print("# of cols = {}".format(str(nb_cols)))

    # extract customers and ratings
    df_X = df_val[:, 0:2]
    # Features are one-hot encoded in a sparse matrix
    X = lil_matrix((nb_rows, nb_cols)).astype('float32')
    df_X[:, 1] = nb_customer + df_X[:, 1]
    coords = df_X[:, 0:2]
    X[np.arange(nb_rows), coords[:, 0]] = 1
    X[np.arange(nb_rows), coords[:, 1]] = 1

    # create label with ratings
    Y = df_val[:, 2].astype('float32')

    # validate size and shape
    print(X.shape)
    print(Y.shape)
    assert X.shape == (nb_rows, nb_cols)
    assert Y.shape == (nb_rows, )

    return X, Y