Function generate_graph_seq2seq_io_data

scripts/generate_training_data.py:12–53

Generate samples from :param df: :param x_offsets: :param y_offsets: :param add_time_in_day: :param add_day_in_week: :param scaler: :return: # x: (epoch_size, input_length, num_nodes, input_dim) # y: (epoch_size, output_length, num_nodes, output_dim

(
        df, x_offsets, y_offsets, add_time_in_day=True, add_day_in_week=False, scaler=None
)

def generate_graph_seq2seq_io_data(
        df, x_offsets, y_offsets, add_time_in_day=True, add_day_in_week=False, scaler=None
):
    """
    Generate samples from
    :param df:
    :param x_offsets:
    :param y_offsets:
    :param add_time_in_day:
    :param add_day_in_week:
    :param scaler:
    :return:
    # x: (epoch_size, input_length, num_nodes, input_dim)
    # y: (epoch_size, output_length, num_nodes, output_dim)
    """

    num_samples, num_nodes = df.shape
    data = np.expand_dims(df.values, axis=-1)
    data_list = [data]
    if add_time_in_day:
        time_ind = (df.index.values - df.index.values.astype("datetime64[D]")) / np.timedelta64(1, "D")
        time_in_day = np.tile(time_ind, [1, num_nodes, 1]).transpose((2, 1, 0))
        data_list.append(time_in_day)
    if add_day_in_week:
        day_in_week = np.zeros(shape=(num_samples, num_nodes, 7))
        day_in_week[np.arange(num_samples), :, df.index.dayofweek] = 1
        data_list.append(day_in_week)

    data = np.concatenate(data_list, axis=-1)
    # epoch_len = num_samples + min(x_offsets) - max(y_offsets)
    x, y = [], []
    # t is the index of the last observation.
    min_t = abs(min(x_offsets))
    max_t = abs(num_samples - abs(max(y_offsets)))  # Exclusive
    for t in range(min_t, max_t):
        x_t = data[t + x_offsets, ...]
        y_t = data[t + y_offsets, ...]
        x.append(x_t)
        y.append(y_t)
    x = np.stack(x, axis=0)
    y = np.stack(y, axis=0)
    return x, y


def generate_train_val_test(args):