Book Image

Modern Time Series Forecasting with Python

By : Manu Joseph
5 (1)
Book Image

Modern Time Series Forecasting with Python

5 (1)
By: Manu Joseph

Overview of this book

We live in a serendipitous era where the explosion in the quantum of data collected and a renewed interest in data-driven techniques such as machine learning (ML), has changed the landscape of analytics, and with it, time series forecasting. This book, filled with industry-tested tips and tricks, takes you beyond commonly used classical statistical methods such as ARIMA and introduces to you the latest techniques from the world of ML. This is a comprehensive guide to analyzing, visualizing, and creating state-of-the-art forecasting systems, complete with common topics such as ML and deep learning (DL) as well as rarely touched-upon topics such as global forecasting models, cross-validation strategies, and forecast metrics. You’ll begin by exploring the basics of data handling, data visualization, and classical statistical methods before moving on to ML and DL models for time series forecasting. This book takes you on a hands-on journey in which you’ll develop state-of-the-art ML (linear regression to gradient-boosted trees) and DL (feed-forward neural networks, LSTMs, and transformers) models on a real-world dataset along with exploring practical topics such as interpretability. By the end of this book, you’ll be able to build world-class time series forecasting systems and tackle problems in the real world.
Table of Contents (26 chapters)
1
Part 1 – Getting Familiar with Time Series
6
Part 2 – Machine Learning for Time Series
13
Part 3 – Deep Learning for Time Series
20
Part 4 – Mechanics of Forecasting

Generating single-step forecast baselines

We reviewed and generated a few baseline models back in Chapter 4, Setting a Strong Baseline Forecast. But there is a small issue – the prediction horizon. In Chapter 6, Feature Engineering for Time Series Forecasting, we talked about how the machine learning model can only predict one target at a time and that we are sticking with a single-step forecast. The baselines we generated earlier were not single-step, but multi-step. Generating a single-step forecast for baseline algorithms such as ARIMA or ETS requires us to fit on history, predict one step ahead, and then fit again using one more day. Predicting in such an iterative fashion for our test or validation period requires us to do this iteration ~1,440 times (48 data points a day for 30 days) and repeat this for all the households in our selected dataset (150, in our case). This would take quite a long time to compute.

We have chosen the naïve method and seasonal naï...