You're reading from Modern Time Series Forecasting with Python Explore industry-ready time series forecasting using modern machine learning and deep learning

Product type Paperback

Published in Nov 2022

Publisher Packt

ISBN-13 9781803246802

Length 552 pages

Edition 1st Edition

Languages

Python

Tools

Time Series Forecasting

Concepts

Data Science

Author (1):

Manu Joseph

View More author details

Table of Contents (26) Chapters

Preface

1. Part 1 – Getting Familiar with Time Series

2. Chapter 1: Introducing Time Series FREE CHAPTER

3. Chapter 2: Acquiring and Processing Time Series Data

4. Chapter 3: Analyzing and Visualizing Time Series Data

5. Chapter 4: Setting a Strong Baseline Forecast

6. Part 2 – Machine Learning for Time Series

7. Chapter 5: Time Series Forecasting as Regression

8. Chapter 6: Feature Engineering for Time Series Forecasting

9. Chapter 7: Target Transformations for Time Series Forecasting

10. Chapter 8: Forecasting Time Series with Machine Learning Models

11. Chapter 9: Ensembling and Stacking

12. Chapter 10: Global Forecasting Models

13. Part 3 – Deep Learning for Time Series

14. Chapter 11: Introduction to Deep Learning

15. Chapter 12: Building Blocks of Deep Learning for Time Series

16. Chapter 13: Common Modeling Patterns for Time Series

17. Chapter 14: Attention and Transformers for Time Series

18. Chapter 15: Strategies for Global Deep Learning Forecasting Models

19. Chapter 16: Specialized Deep Learning Architectures for Forecasting

20. Part 4 – Mechanics of Forecasting

21. Chapter 17: Multi-Step Forecasting

22. Chapter 18: Evaluating Forecasts – Forecast Metrics

23. Chapter 19: Evaluating Forecasts – Validation Strategies

24. Index

Why subscribe?

25. Other Books You May Enjoy

Informer

Recently, Transformer models have shown superior performance in capturing long-term patterns than standard RNNs. One of the major factors of that is the fact that self-attention, which powers Transformers, can reduce the length that the relevant sequence information has to be held on to before it can be used for prediction. In other words, in an RNN, if the timestep 12 steps before holds important information, that information has to be stored in the RNN through 12 updates before it can be used for prediction. But with self-attention in Transformers, the model is free to create a shortcut between lag 12 and the current step directly because of the lack of recurrence in the structure.

But the same self-attention is also the reason why we can’t scale vanilla Transformers to long sequences. In the previous section, we discussed how long-term forecasting is a challenge because of two reasons: the expressiveness required to truly capture the variation and computational...