Machine Learning for Business Analytics, 2nd Edition

17 HANDLING TIME SERIES

In this chapter, we describe the context of business time series forecasting and introduce the main approaches that are detailed in Chapters 18 and 19 and, in particular, regression‐based forecasting and smoothing‐based methods. Our focus is on forecasting future values of a single time series. These three chapters (Chapters 17–19) are meant as an introduction to the general forecasting approach and methods.

In this chapter, we discuss the difference between the predictive nature of time series forecasting and the descriptive or explanatory task of time series analysis. A general discussion of combining forecasting methods or results for added precision follows. Additionally, we present a time series in terms of four components (level, trend, seasonality, and noise), and methods for visualizing the different components and exploring time series data. We close with a discussion of data partitioning (creating training and validation sets), which is performed differently from cross‐sectional data partitioning.

Time Series in JMP: The methods introduced in the three chapters in this part of the book are available in the standard version of JMP. However, JMP Pro is required to evaluate performance based on validation metrics for regression‐based time series models.

17.1 INTRODUCTION¹

Time series forecasting is performed in nearly every organization that works with quantifiable data. Retail stores use it to forecast sales. Energy companies use it to ...