MTNetForecaster


MTNetForecaster

Introduction

MTNet is a memory-network based solution for multivariate time-series forecasting. In a specific task of multivariate time-series forecasting, we have several variables observed in time series and we want to forecast some or all of the variables' value in a future time stamp.

MTNet is proposed by paper A Memory-Network Based Solution for Multivariate Time-Series Forecasting. MTNetForecaster is derived from tfpark.KerasMode, and can use all methods of KerasModel. Refer to tfpark.KerasModel API Doc for details.

For the detailed algorithm description, please refer to here.

Method

Arguments

__init__

MTNetForecaster(target_dim=1,
                 feature_dim=1,
                 long_series_num=1,
                 series_length=1,
                 ar_window_size=1,
                 cnn_height=1,
                 cnn_hid_size=32,
                 rnn_hid_sizes=[16, 32],
                 lr=0.001,
                 loss="mae",
                 cnn_dropout=0.2,
                 rnn_dropout=0.2,
                 metric="mean_squared_error",
                 uncertainty: bool = False,
                 )

fit, evaluate, predict

Refer to fit, evaluate, predict defined in tfpark.KerasModel API Doc

Reference

Yen-YuChang, Fan-YunSun, Yueh-HuaWu, Shou-DeLin, A Memory-Network Based Solution for Multivariate Time-Series Forecasting.