Forecasting Time Series in Python – eBook

Description

Forecasting future outcomes is a problem that appears in many research and industry settings, from business, economic, climate science to science and engineering.

Time Series Forecasting in Python is a practical, model-first guide to understanding, building, evaluating, and deploying forecasting systems. Written for master’s-level students and data scientists moving into time series modelling, the book combines statistical foundations with modern machine learning and real-world workflow design.

The book begins with the essentials: time series notation, exploratory analysis, decomposition, forecast horizons, forecast origins, and evaluation. It then develops classical forecasting methods including benchmark models, exponential smoothing, ARIMA, dynamic regression, and state space models. Modern approaches follow, with chapters on feature-based machine learning, global forecasting models, deep learning, recurrent neural networks, foundation models, probabilistic forecasting, and Bayesian methods.

Later chapters move from models to forecasting systems, covering AutoML, hierarchical forecasting, monitoring and maintenance, intermittent demand, reconciliation, forecasting at scale, decision-focused forecasting, and clustering. The final part focuses on practice through case studies, common pitfalls, and reproducible workflows.

Throughout, the emphasis is on clear assumptions, honest out-of-sample evaluation, uncertainty, benchmarks, and decision relevance. Python examples are used throughout to simulate data, fit models, create forecasts, evaluate performance, and produce figures. The result is a complete roadmap for developing forecasting skills that are both technically sound and practically useful.

Includes zipped file containing all code used throughout the book.

Table of Contents:

I Foundations

1. What Is Time Series Forecasting?
2. Time Series Data and Notation
3. Exploratory Analysis and Decomposition
4. Forecasting Principles and Evaluation

II Classical Forecasting Methods

5. Benchmark Methods
6. Exponential Smoothing
7. ARIMA Models
8. Dynamic Regression and Transfer Function Models
9. State Space Models

III Modern Forecasting Methods

10. Feature-Based Machine Learning
11. Global Forecasting Models
12. Deep Learning for Forecasting
13. Recurrent Neural Networks for Forecasting
14. Probabilistic Forecasting
15. Bayesian Time Series Forecasting
16. Time Series Foundation Models for Forecasting

IV Forecasting Systems and Practice

17. Hierarchical Forecasting
18. Forecast Monitoring and Model Maintenance
19. Intermittent Demand Forecasting
20. Forecast Reconciliation
21. Forecasting at Scale
22. Automated Machine Learning for Time Series Forecasting
23. Decision-Focused Forecasting
24. Using Clustering to Support Forecasting

V Forecasting in Practice

25. Case Studies
26. Common Pitfalls
27. Reproducible Forecasting Workflows
28. Looking Forward