MGTS 4200
QUANTITATIVE METHODS FOR BUSINESS ANALYSIS AND FORECASTING
Winter 1999---Prof. Peter Tryfos

Introduction

The course provides a survey of the principal quantitative methods for analysis and forecasting. By "analysis," we understand the detection of patterns, relationships and trends, and by "forecasting," the projection of these patterns, relationships and trends. Problems of analysis and forecasting arise in all areas of business, but the methods themselves tend to be common for all.

The course emphasizes the assumptions, conclusions, and applications of the methods rather than their mathematical derivation or calculation. Applications are illustrated by means of substantial cases. The cases are drawn mainly from the Canadian environment, and refer to such diverse issues and problems as insurance, real estate, employment equity, credit screening, market research, energy management, demographics, and others.

In this era of the computer, methods are rarely implemented by hand, especially in view of the large data sets typically encountered in business studies. The number of computer programs assisting business analysis is rapidly growing and their quality and "friendliness" continually improving. Instruction will be provided for the use of Excel, SAS and, if feasible, SPSS. These programs are capable of implementing all the methods surveyed in the course, but students are free to use any other program they prefer.

Because the course examines methods with applications in all areas of business, it may be of interest to students concentrating in a functional area (finance, marketing, etc.), as well as to those concentrating in Management Science.
Text

Peter Tryfos, Methods for Business Analysis and Forecasting: Text and Cases, John Wiley & Sons, 1998.

Topics
(Approximate number of weeks in parentheses. Text references in brackets.)

 Topic Number of weeks Description Text reference 1 0.5 Introduction. Exact and approximate relationships. Estimation of relationships among variables. Ch. 1 2 1 Regression. LS estimation. Contribution of explanatory variables. Ch. 2 3 1 Attributes as explanatory variables. Dummy variables. Secs. 4.1-4.4 4 1 Non-linear relationships. Non-linear regression. Ch. 5 5 1 Regression and time series. Estimation of trend and seasonality. Seasonal adjustment. Ch. 6 6 1 Lagged explanatory variables. Distributed lag models. Ch. 7 7 1 Regression subject to constraints. Least absolute deviations. Stepwise, all-subsets regression. Elements of experimental design. Ch. 8 8 1 Autoregressive models. Secs. 10.1-10.5 9 1.5 The classical linear model and inference therefrom. Analysis of variance. ARIMA models. Ch. 3, Secs. 4.5-4.9, 10.6-10.9 10 1 Violations of the classical linear model and simple remedies for non-linearity, non-constant variance, and non-independence. Specification errors. Ch. 9 11 2 Attributes as dependent variables. The classification problem. Probability models. Discriminant analysis. Chs. 11 and 12 12 1 Estimation methods for systems of equations. Chs. 13 13 (Time permitting) Other multivariate methods: factor and cluster analysis. Supplements
Computing

Students are free to use whatever software package they prefer. However, support is provided only for Excel and SAS. The use of these programs and of SPSS will be illustrated in parallel with the lectures and cases. No previous knowledge of computer programming is required.

The student's grade will depend on three assignments (weight 30%), two case analyses (weight 40%), and a final exam (weight 30%).

The assignments consist of the following problems from the text:

 Assignment Number Do problems: 1 2.3, 2.14, 4.3, 4.10, 5.4, 5.12 2 6.10, 6.12, 7.4, 7.8, 8.7, 8.8 3 10.8, 10.14(b), 11.3, 11.9, 12.2, 12.12(a)-(e)

The dates at which these assignments are due will be announced in class.

The cases are included in the course text. They may be done individually or in teams of two students. A written analysis of about five pages for each, excluding appendices and computer output, is expected. In addition, students are expected to present in class their analysis of one of these two cases. The intent is to hear and discuss in class one case per week. The schedule of the cases and the random assignment of students to cases will be detailed in the second week of the course.

The final exam is open-book.