ADVANCED RESEARCH METHODS

INTRODUCTION TO FORECASTING USING REGRESSION ANALYSIS

Spring 1999

780-4008
Fax 780-4417

Overview:

This course introduces students to forecasting using regression analysis. Students will work individually and in teams to conduct a series of forecasting projects using autoregressive, multiple regression, and mixed models. The emphasis will be on learning the theoretical and practical applications of regression analysis and introducing the problems of forecasting.

PREREQUISITE: Familiarity with spreadsheet use. The course will use Microsoft Excel for the analytic exercises. Students may use another spreadsheet or statistics package, but instruction will focus on Excel.

Assignments

Four written assignments are required as outlined in the syllabus. These must be available at the beginning of the class when they are due for review by other members of the class. Each paper will be graded and will count 20% of the final grade. Some projects will be done in teams, some individually. 20% of the grade will be for class participation.

Learning Strategies

This course emphasizes learning by doing rather than presentations. While there will be some presentation each week in preparation for the next class, the principal focus will be on the students' projects. To get the most out of the course, you need to adopt the following learning strategies:

Practice, practice, practice. Learning quantitative analysis is rarely immediately intuitive. The only way to acquire familiarity and some degree of facility with the approaches to analysis discussed in the course is to keep practicing. Making mistakes is less important than continuing to practice the techniques.

Peer learning Through team projects and review of other class members products, much of the learning in the course will be through your peers. The benefits you get from this experience will be directly proportional to your contributions to team projects and to careful and serious reviews and critiques of other's work.

Text to text. While there will be some classroom instruction in the quantitative techniques, students will need to rely on the text for the course for a good deal of the theory behind the techniques and for the details of many of the calculations. In actual practice, analysts will often have to rely on their own reading and learning skills to brush up on old skills or learn new ones.

Text

Peter Tryfos Methods for Business Analysis and Forecasting: Text and Cases

February 24: Introduction to Forecasting and Regression

Reading for this week: Tryfos, Chapters 1-3,
Assignment due next week Maine Turnpike 1

March 3 Dummy Variables

Reading for this week: Tryfos, Chapters 4 (to p. 107)
Assignment due next week: The Maine Turnpike 2

March 10: Autoregressive models

Reading for this week. Tryfos, Chapter 10
Assignment due next week: The Maine Turnpike 3.

March 17: Non linear models

Reading for this week: Tryfos, Chapter 5.
Assignment due next week: Population Forecast

March 24 No Class

March 31 Population forecast

Assignment 1 (Team Projects)

Using the monthly data for the Maine Turnpike traffic from 1956-1996, calculate an annual total traffic series. Estimate models for annual Turnpike traffic using traffic alone and using the data on Gross State Product. Write a brief memo showing your models and giving the appropriate regression statistics. Include graphs showing estimated and actual values. Which model do you recommend be used?

Assignment 2 (Individual Projects)

Prepare an analysis of monthly traffic patterns on the Maine Turnpike. Develop a regression model that forecasts Turnpike traffic by month for 1997 and 1998, as well as a moving average model.

Assignment 3 (Individual Projects)

Prepare an autoregressive model of Turnpike traffic. Write a memo comparing the results of the various approaches to forecasting annual turnpike traffic that you have undertaken and presenting the results of a forecast for traffic from 1997-2000. (Your monthly forecasts in Assignment 2 should be aggregated to annual figures). Also comment on which models work best for forecasting peak summer traffic.

Assignment 4 (Team Projects)

Population forecast. Read the Population Projects case in Tryfos, and prepare a forecast of the population for the region for 2008 (assume that the "current year" in the data is 1998. Write a memo showing the male and female population in 2008 as well as the contribution of immigration and natural increase over the period. What conclusions do you have about the demand for K-12 school space over the next ten years?

NOTE: In doing this assignment, you will have to use both the data in the case and the forecasts you construct of birth and death rates to construct the various elements of the population forecast. Spend some time familiarizing yourself with the case data in order to know what data is available.