Hyperparameter Tuning MLPs for Probabilistic Time Series Forecasting
This work addresses hyperparameter tuning challenges in time series forecasting for researchers and practitioners, but it is incremental as it builds on existing MLP methods with new data and analysis.
The study investigated the impact of hyperparameters like context length and validation strategy on MLP performance in time series forecasting, finding that tuning these parameters is crucial, based on experiments with 4800 configurations per dataset across 20 datasets. It also introduced TSBench, a metadataset with 97,200 evaluations, a twentyfold increase over previous works, and demonstrated its use for hyperparameter optimization.
Time series forecasting attempts to predict future events by analyzing past trends and patterns. Although well researched, certain critical aspects pertaining to the use of deep learning in time series forecasting remain ambiguous. Our research primarily focuses on examining the impact of specific hyperparameters related to time series, such as context length and validation strategy, on the performance of the state-of-the-art MLP model in time series forecasting. We have conducted a comprehensive series of experiments involving 4800 configurations per dataset across 20 time series forecasting datasets, and our findings demonstrate the importance of tuning these parameters. Furthermore, in this work, we introduce the largest metadataset for timeseries forecasting to date, named TSBench, comprising 97200 evaluations, which is a twentyfold increase compared to previous works in the field. Finally, we demonstrate the utility of the created metadataset on multi-fidelity hyperparameter optimization tasks.