Summary of Forecasting with Deep Learning: Beyond Average Of Average Of Average Performance, by Vitor Cerqueira et al.
Forecasting with Deep Learning: Beyond Average of Average of Average Performance
by Vitor Cerqueira, Luis Roque, Carlos Soares
First submitted to arxiv on: 24 Jun 2024
Categories
- Main: Machine Learning (stat.ML)
- Secondary: Machine Learning (cs.LG)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The proposed framework assesses univariate time series forecasting models from multiple perspectives, addressing the limitations of current practices that average performance over all samples. The framework is applied to a state-of-the-art deep learning approach (NHITS) and classical forecasting techniques, showcasing NHITS’ superiority in certain conditions, such as multi-step ahead forecasting, but being outperformed by methods like Theta when dealing with anomalies. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Forecasting models are used to predict future events. To make sure these predictions are accurate, we need to evaluate them correctly. Currently, this is done by giving a single score based on how well the model performed overall. However, this method can hide important information about which models perform better in specific situations. This paper proposes a new way of evaluating forecasting models that looks at their performance from different angles, like predicting one step ahead or multiple steps ahead. The authors test their approach by comparing a deep learning model (NHITS) with classical forecasting methods. They find that NHITS generally performs well, but its strengths vary depending on the situation. |
Keywords
* Artificial intelligence * Deep learning * Time series
