Recent Advances in Time Series Forecasting

Dinesh C.S. Bisht, Mangey Ram

• Publisher: CRC Press
• Year: 2021
• Print ISBN: 9780367607753
• eBook ISBN: 9781000433845
• Edition: 1
• Topic: Mathematics
• Subtopic: Operations
• Index: Mathematics

Future predictions are always a topic of interest. Precise estimates are crucial in many activities as forecasting errors can lead to big financial loss. The sequential analysis of data and information gathered from past to present is call time series analysis. This book covers the recent advancements in time series forecasting.

The book includes theoretical as well as recent applications of time series analysis. It focuses on the recent techniques used, discusses a combination of methodology and applications, presents traditional and advanced tools, new applications, and identifies the gaps in knowledge in engineering applications.

This book is aimed at scientists, researchers, postgraduate students and engineers in the areas of supply chain management, production, inventory planning, and statistical quality control.

1 Time Series Econometrics: Some Initial Understanding

A. Mishra

CHRIST (Deemed to be a University)

DOI: 10.1201/9781003102281-1

1.1 Introduction

In our daily lives, when we talk about a topic that is related to economics and finance, generally we relate it to time. Most of the macroeconomic variables, which have enormous significance in our lives, are measured in the time dimension. We are very familiar with these variables, such as GDP of a nation, inflation rate and unemployment rate. Likewise, in finance many things are calibrated on the time dimension, such as stock prices and return on investment. Measuring variables with regular frequency gives us a significant platform for useful analysis. Time series econometric analysis has massive implications for making inferences in dynamic circumstances; hence, its uses are surging day by day. These days, the innovations of machine learning and data analysis are expanding the need for time series econometric analysis beyond the dimension of economics and finance. For example, time series analysis data of a cardiac patient can be used to forecast the exact time of a possible cardiac attack in the future through the application of various sophisticated time series econometric modeling.

1.1.1 Learning Objectives

After completing this chapter, students should be able to:

• Understand the meaning of the time series.
• Understand the random walk phenomenon.
• Recognize the stationarity and unit root process.
• Understand the concept of spurious regression.
• Identify the relevancy of a stationary time series.

1.2 Time Series, What Is It?

When we say “time series”, one thing is obvious—it has two dimensions. The first is related to time, and the second is that it is a series, which is a collection of variables. Hence, a time series can be defined as accumulating random variables according to time or in chronological order.

In the time series, the quantitative characteristic is collected or arranged with an equal time interval. For example, suppose we are arranging the GDP growth of a nation in years. Then, we cannot change the time interval to months. We have to follow the same sequence of time for each variable. If we follow the same time sequence, the series will be considered a time series of data. One important aspect that we have to keep in mind is that the time interval could be any period; it may be yearly, quarterly, monthly, weekly, daily or even hourly. Accordingly, we should classify the data as a yearly time series, monthly time series, weekly time series or daily time series.

Figure 1.1 shows a graph of daily trading time series data for the Indian stock exchange from 2010 onwards.

Figure 1.1 Indian stock exchange, Sensex daily trading data from 2010 onward.

1.2.1 Four Components of a Time Series

Usually, any time series includes the four components of trend, cyclical, seasonal and irregular, which are discussed below. These components have their relevancy in the forecasting process. When analyzing the time series data from a statistical point of view, we always pay attention to these four components, and our primary intention is to decompose these components so that we can forecast. There are two ways to deal with these components, the addition method and the multiplication method. We now discuss the components of a time series in detail.

1. Trend (T)
2. Cyclical (C)
3. Seasonal (S)
4. Irregular (I)

1.2.2 Trend Component

The trend is the long-run array of a time series. A trend can be upward or downward, conditional on whether the time series exhibits a surging long-term outline or a diminishing long-term array. On the other hand, if a time series does not demonstrate an upward or downward outline, it is considered stationary in the mean.

1.2.3 Cyclical Component

Sometimes there is a continuous fluctuation in the trend line. Sometimes it moves up, and sometimes it falls. Such continuous movement is considered a cyclical pattern. The length of a cyclical pattern is determined by the type of industry and business that has been selected for the analysis.

1.2.4 Seasonal Component

Seasonality happens when the time series shows systematic variations throughout an identical time sequence. The time sequence could be monthly or quarterly each year. For instance, there is a market sale surge throughout the month of Diwali and the festive session in India and during Christmas in Christian parts of the world.

1.2.5 Irregular Component

The irregular element of a time series will always be random and impulsive in nature. Every time series has some distinctive constituent that transforms it into an unsystematic variable. In forecasting, the main motto is to “calibrate” each and every component of the time series in a precise manner except for the irregular ...

Table of contents

• Cover
• Half Title
• Series Page
• Title Page
• Copyright Page
• Contents
• Preface
• Editors
• Chapter 1 Time Series Econometrics: Some Initial Understanding
• Chapter 2 Time Series Analysis for Modeling the Transmission of Dengue Disease
• Chapter 3 Time Series Analysis of COVID-19 Confirmed Cases in Select Countries
• Chapter 4 Bayesian Estimation of Bonferroni Curve and Zenga Curve in the Case of Dagum Distribution
• Chapter 5 Band Pass Filters and their Applications in Time Series Analyses
• Chapter 6 Deep Learning Approaches to Time Series Forecasting
• Chapter 7 ARFIMA and ARTFIMA Processes in Time Series with Applications
• Chapter 8 Comparative Study of Time Series Forecasting Models for COVID-19 Cases in India
• Chapter 9 Time Series Forecasting Using Support Vector Machines
• Chapter 10 A Comprehensive Review of Urban Floods and Relevant Modeling Techniques
• Chapter 11 Fuzzy Time Series Techniques for Forecasting
• Chapter 12 Artificial Neural Networks (ANNs) and their Application in Soil and Water Resources Engineering
• Index