Hierarchical Time Series Forecasting Via Latent Mean Encoding
This work addresses a specific challenge in temporal hierarchical forecasting for business applications, representing an incremental improvement.
The paper tackles the problem of coherently forecasting a target variable across multiple temporal scales for business decision-making by proposing a hierarchical architecture that encodes average behavior in hidden layers, achieving improved accuracy over established methods like TSMixer on the M5 dataset.
Coherently forecasting the behaviour of a target variable across both coarse and fine temporal scales is crucial for profit-optimized decision-making in several business applications, and remains an open research problem in temporal hierarchical forecasting. Here, we propose a new hierarchical architecture that tackles this problem by leveraging modules that specialize in forecasting the different temporal aggregation levels of interest. The architecture, which learns to encode the average behaviour of the target variable within its hidden layers, makes accurate and coherent forecasts across the target temporal hierarchies. We validate our architecture on the challenging, real-world M5 dataset and show that it outperforms established methods, such as the TSMixer model.