Demand Forecasting Models With Time Series and Random Forest

Demand Forecasting Models With Time Series and Random Forest

Halit Alper Tayali
Copyright: © 2021 |Pages: 24
DOI: 10.4018/978-1-7998-5879-9.ch004
OnDemand:
(Individual Chapters)
Available
$33.75
List Price: $37.50
10% Discount:-$3.75
TOTAL SAVINGS: $3.75

Abstract

This chapter presents the recent methodological developments in demand management and demand forecasting subjects of the operations management. The background section provides detailed information on the domain of production management, operational analytics, and demand forecasting while providing introductory information on time series forecasting and related machine learning methodologies. The novel contribution of the chapter is the exploration developed in the solutions and recommendations section while examining the effect of stationarity in the time series forecasting methodologies of machine learning with improved benchmark results.
Chapter Preview
Top

Introduction

The contemporary business organization, or a company, is based on three core functions, namely marketing, operations, and finance. The discipline of operations management, which serves as the major function of both service and manufacturing organizations, aims to organize the activities related to producing goods and services. The scientific literature on the operations management focuses on transforming inputs to outputs for to economically sustain the companies, while making decisions based on available and limited resources. Therefore, a key issue in operations management is to augment the decision processes while sustaining a business certainly requires good decisions. Heizer, Render, & Munson (2017) address the strategic decisions that operations managers make in relation to design of goods and services, managing quality, strategies of process, location, layout, human resources, supply-chain, inventory, scheduling and maintenance. These operational decisions typically include processes of quantitative modelling, calculation, forecasting and prediction to enhance profitability, and service to society.

The chaotic and global business dynamics change at an even more accelerating speed as new technologies persist to evolve, and companies are one of the central figures in the 4th Industrial Revolution for the significant impacts of digitization immediately unfold here. The wide scope of the Industry 4.0 includes concepts such as automation, internet of things, and knowledge-based, decision support or embedded systems. The automation efforts in the Industry 4.0 include methodologies of artificial intelligence and one of its major interdisciplinary subfields, machine learning. Artificial intelligence is seen as the governor of the Industry 4.0. Although the humans are the scriptwriters, machines can now write their own automated scripts as well. One can easily argue that the student becomes the master. However, according to a computer scientist, a computer just performs calculations and remembers the results (Guttag, 2013). Therefore, assigning subjective attributes and prediction tasks to artificial intelligence might be a little bit of overshooting for the distinction between the scope of the objectivity of the machine, and that of the subjectivity of the human remains a mystery. In fact, this is quite like asking whether the theory of mathematics is an invention or a discovery. Another challenging question is based on the definition of forecasting, the art and science of future event prediction (Heizer et al., 2017); but, is what the machines do really art or science? If the answer is yes, then art or science might also have to include the term machine within their linguistic definitions.

The humans make predictions about the future all the time -even at when we are unaware of this phenomenon. The ability to create scenarios and thinking ahead is one of the distinguishing attributes of the humankind and perhaps the biggest reason that the society is evolving towards an automated future. There is very little doubt that the issue of prediction accuracy might have been easily resolved as the physiological nature of the prediction phenomenon gets solved. Humans use their prediction and forecasting abilities in the organizations that they form. Tetlock & Gardner (2015) explain the super forecasting and prediction abilities of humans. However, the concept of forecasting is often confused with that of the prediction and Everett & Ebert (1986) clearly separate these two concepts: Forecasting requires quantitative modelling, yet prediction requires skill, experience, and judgement on top of that. However, although forecasting casts past data, which have been combined in a predetermined way, systematically forward, and prediction uses subjective considerations that does not need a predetermined combination, both prediction and forecasting are processes of estimation. Krajewski, Malhotra, & Ritzman, (2016) define forecast as a future event prediction for the purpose of planning.

Key Terms in this Chapter

Model: A mathematical representation of a real and complex system that formulates the relation between an input and an output.

Technological Forecast: A long-term forecast that examines the characteristics of advancements, policies, strengths, weaknesses, opportunities, or threats in relation to technology.

Decision Analysis Models: Mathematical representations of quantified business processes for the analysis of decision alternatives for better decision-making.

Economics Theory: The theory that investigates the interdependent relationships of people or organizations using the flow of money, goods, and services as well as various concepts such as interest rate, inflation, competition, growth, currency exchange, unemployment, social capital, and technological leaps.

Data Science: An interdisciplinary field of study of data in relation to mathematics, statistics, operations research, data analysis, data mining, machine learning, computer science, engineering, visualization, data privacy, and big data.

Pattern Recognition: Discovering or detecting the explicit relation or repetitive structures in a dataset using data science.

Data Mining: A set of machine learning and statistical methods or techniques for drawing insight or extracting information from data or revealing patterns that are not implicitly available.

Economic Forecast: A long-term forecast for planning the behaviour of the economic actors.

Data Analysis: Examining data for to extract information with the help of a set of computational methods or techniques that range from basic descriptive statistical techniques to more sophisticated data mining.

Artificial Intelligence: Programming computers for performing complex tasks that can be done by human beings with intelligence.

Complete Chapter List

Search this Book:
Reset