Amit Potdar

Graduation Semester and Year

2009

Language

English

Document Type

Dissertation

Degree Name

Doctor of Philosophy in Industrial Engineering

Department

Industrial and Manufacturing Systems Engineering

First Advisor

Jamie Rogers

Abstract

Reverse logistics has gained much attention in recent years. It is becoming a value added area of a supply chain network day by day. For enterprises, it has therefore become essential to manage the reverse flow of materials in an efficient way to gain competitive advantage. One important aspect of reverse logistics is to have a correct and timely estimation of the reverse flow of materials. Improved forecast accuracy leads to a better decision making in strategic, tactical and operational areas of an organization. Intrinsic (time series) and extrinsic (causal) forecasting are some of the well known and frequently used forecasting techniques. Very little research has been done about the forecasting aspect of reverse logistics. The initial research that has been carried out in this area was very naive. It used the basic method of probability by proportions of cumulative returns to cumulative sales. For higher forecast accuracy, more robust method is required. The purpose of this research is to develop the methodology that can be used for forecasting product returns. This methodology is developed for the consumer electronics industry. The methodology in this research is based on return reason codes (RC). The reason code based forecasting is a unique part of this research. The incoming returns are split into different categories using reason codes. These reason codes are further analyzed to forecast product returns. The computation part of this model uses a combination of two approaches namely extreme point approach and central tendency approach. Both the approaches are used separately for separate types of reason codes, and then results are added together. The extreme point approach is based on data envelopment analysis (DEA) as a first step combined with linear regression, while the central tendency approach uses a moving average. DEA is a non-parametric tool that is used to analyze performance indices. For certain types of returns, DEA evaluates relative ranks of products using 'single input and multiple outputs' model. Once this is completed, linear regression defines a correlation between relative ranks (predictor variable) and return quantities (response variable). For the remaining types of returns, we use a moving average of percent returns to estimate the central tendency. Thus, by combining two approaches for different types of returns, we have developed the model that can be used to forecast product returns for the consumer electronics industry.

Disciplines

Engineering | Operations Research, Systems Engineering and Industrial Engineering

Comments

Degree granted by The University of Texas at Arlington

Download

Share

COinS