LSTM-MSNet: Leveraging Forecasts on Sets of Related Time Series With Multiple Seasonal Patterns

Abstract

Generating forecasts for time series with multiple seasonal cycles is an important use-case for many industries nowadays. Accounting for the multi-seasonal patterns becomes necessary to generate more accurate and meaningful forecasts in these contexts. In this paper, we propose Long Short-Term Memory Multi-Seasonal Net (LSTM-MSNet), a decompositionbased, unified prediction framework to forecast time series with multiple seasonal patterns. The current state of the art in this space are typically univariate methods, in which the model parameters of each time series are estimated independently. Consequently, these models are unable to include key patterns and structures that may be shared by a collection of time series. In contrast, LSTM-MSNet is a globally trained Long Short-Term Memory network (LSTM), where a single prediction model is built across all the available time series to exploit the crossseries knowledge in a group of related time series. Furthermore, our methodology combines a series of state-of-the-art multiseasonal decomposition techniques to supplement the LSTM learning procedure. In our experiments, we are able to show that on datasets from disparate data sources, like e.g. the popular M4 forecasting competition, a decomposition step is beneficial, whereas in the common real-world situation of homogeneous series from a single application, exogenous seasonal variables or no seasonal preprocessing at all are better choices. All options are readily included in the framework and allow us to achieve competitive results for both cases, outperforming many state-ofthe-art multi-seasonal forecasting methods.

Traditional vs. Machine-Learning Methods for Forecasting Sandy Shoreline Evolution Using Historic Satellite-Derived Shorelines

“…Therefore ML models are able to extract patterns from irregular collections of time series which cannot be distinguished at individual level. Recent studies [34,[36][37][38][39] have demonstrated that Neural Networks (NNs) are particularly well suited to take advantage of a dataset which consists of large amounts of time-series. Moreover, whereas classical models are typically limited to univariate inputs, NNs are able to use multiple inputs, and therefore these types of models can be enhanced by (additional) explanatory variables.…”

MuSDRI: Multi-Seasonal Decomposition Based Recurrent Imputation for Time Series

“…For example, [36] proposes a hierarchical prediction method that combines the Exponential Smoothing (ES) model with the Long Short Term Memory (LSTM) neural network to extract local as well as global information from time series data. [37] proposes a model called Long Short-Term Memory Multi-Seasonal Net (LSTM-MSNet), which is able to predict the changes of time series with multiple seasonal patterns. [38] proposes a network traffic prediction model which combines STL and LSTM.…”

Short-Term Net Load Forecasting with Singular Spectrum Analysis and LSTM Neural Networks

“…Notable techniques include Seasonal and Trend decomposition using Loess (STL) [24], which could be extended for multiple seasonalities, trigonometricbased decomposition (TBATS) [25], and Singular Spectrum Analysis (SSA) [26]. In [27], a decomposition-based LSTM model was proposed to deal with multiple seasonal patterns. In that model, the time series is decomposed into various seasonal components, which are used as exogenous regressors in an LSTM.…”