Clustering-Based Stability and Seasonality Analysis for Optimal Inventory Prediction

Clustering-Based Stability and Seasonality Analysis for Optimal Inventory Prediction

Manish Joshi (North Maharashtra University, India), Pawan Lingras (Saint Mary's University Halifax, Canada), Gajendra Wani (Bhusawal Arts, Science and Commerce College, India) and Peng Zhang (Saint Mary's University Halifax, Canada)
Copyright: © 2014 |Pages: 18
DOI: 10.4018/978-1-4666-4936-1.ch001
OnDemand PDF Download:
$30.00
List Price: $37.50

Abstract

This chapter exemplifies how clustering can be a versatile tool in real life applications. Optimal inventory prediction is one of the important issues faced by owners of retail chain stores. Researchers have made several attempts to develop a generic forecasting model for accurate inventory prediction for all products. Regression analysis, neural networks, exponential smoothing, and Autoregressive Integrated Moving Average (ARIMA) are some of the widely used time series prediction techniques in inventory management. However, such generic models have limitations. The authors propose an approach that uses time series clustering and time series prediction techniques to forecast future demand for each product in an inventory management system. A stability and seasonality analysis of the time series is proposed to identify groups of products (local groups) exhibiting similar sales patterns. The details of the experimental techniques and results for obtaining optimal inventory predictions are shared in this chapter.
Chapter Preview
Top

1. Introduction

This chapter exemplifies how clustering can be a versatile tool in real life applications. Traditionally, clustering techniques group similar patterns, which can be used for creating profiles. We show how clustering can be a very useful first step in prediction. Optimal inventory prediction is one of the important issues faced by owners of retail chain stores. Determination of how, when and what quantities of products are to be reordered is a key to running a profitable business. Researchers have made several attempts to develop a generic forecasting model for accurate inventory prediction for all products. The demand quantity of a particular item or a group of related items can be considered as a time series. Time series prediction techniques that predict future values of a time series plays a critical role in forecasting quantity demand in business operations. Regression analysis, neural networks, exponential smoothing and autoregressive integrated moving average (ARIMA) are some of the widely used time series prediction techniques in inventory management.

Many researchers focus on finding a generic forecasting solution for all the products. However, products are distinguished by their seasonal sales patterns and volatilities in sales demand. One generic solution may not always be able to predict the most accurate demand for each product. Hence, these approaches are not always successful.

Limitations of generic forecasting model are overcome by developing specialized or targeted forecasting models. A variety of prediction techniques are combined with clustering to develop inventory prediction models. Clustering facilitate to explore stability underneath temporal variations. This chapter describes stability and seasonality analysis used to develop inventory prediction model.

We also elaborate on the effectiveness of stability analysis by applying it to a larger volume of data and further analyze stability for multiple years. We demonstrate that a group of stable products obtained using stability analysis is similar for different years. We also share our observations regarding how stable products from a particular year carry forward to subsequent years. A cross tabulation is used for trend analysis that emphasizes importance of stability analysis for multiple years.

The details of the experimental techniques and results for obtaining optimal inventory predictions are shared in this chapter. We elaborate usefulness of clustering in sales forecasting of objects that show similar sales patterns.

The experimental data set is obtained from an independent small retail chain of specialty stores. Information of customers, products, and their business operations from January 2005 to December 2009 is used for experimentation. More than 600,000 sales transactions are recorded in 60 months. In total, there are 25,378 distinct customers and 15,045 different products. Table 1 shows specific characteristics of the data set.

Table 1.
An overview of experimental data set
Attribute20052006200720082009
Number of products57827567803489489409
Number of customers42036159105011154813247
Number of transactions557749985275664131499147995

Complete Chapter List

Search this Book:
Reset