Operational Time Series Forecasting

A Hands-On Guide to Classical and Modern Methods in R and Python

Preface

Note: This chapter is under construction.

This is a Quarto book project started on September 26, 2026. Hence, it is a work in progress. Every unfinished chapter includes a note at the beginning to indicate the unfinished state.

I welcome all readers to the (currently unfinished) first edition of Operational Time Series Forecasting. This book is directed at data analysts in the industry, who deal with problems related to time series forecasting. Unlike other literature, this is not an entry-level text. While theoretical ideas are usually summarized briefly throughout, they are not introduced in too much detail to ensure the focus remains on the hands-on application of the ideas. Hence it is assumed that readers have at least a solid foundation of basic time series concepts.

From my experience, no currently existing literature is truly discussing the entire application pipeline of time series forecasting. Indeed, bits and pieces of it, such as cleaning steps or a selection of available models. But not the entire pipeline as a whole, where each step is logically linked to every step that came before. Therefore, this projects aims to fill this gap in the literature by providing a holistic view on how to solve common industry forecasting tasks, including

• data downloading,
• data cleaning,
• feature selection,
• model selection,
• model fitting,
• forecasting, and
• model backtesting.

In most cases, each chapter is accompanied by two separate labs, one for R and one for Python, showing how to obtain the results illustrated in the corresponding chapter’s examples.

While this book is currently produced in HTML format, it is a goal to provide an equivalent PDF version in the long run.

Acknowledgments

To Hương, my love and life.