Please use this identifier to cite or link to this item: http://hdl.handle.net/1959.3/26128

Download PDF (Adobe Acrobat PDF, -1 bytes)

Title

Objective assessment of forecasting assignments using some function of prediction errors

Author(s)

Clarke, Stephen R.

Abstract

One way of examining forecasting methods via assignments is to give each student a real or simulated set of data, with a requirement to forecast future values. However checking the accuracy of calculations for the host of possible methods can be onerous. One solution is to make part or the entire assessment dependent on the accuracy of the forecasts obtained. This mirrors real life, where forecasts are judged not by the method used but by how accurate the predictions turn out. This paper investigates how this might work with an actual example. Using simulated data from a model which incorporates trend, seasonality, Easter effect and randomness, we use a function of the mean square error of the forecasts to determine the final mark for a variety of methods. Results indicate that the students who have put in more work, and/or fitted the better models, would obtain the better marks.

Publication type

Conference paper

Research centre

Swinburne University of Technology

Source

Proceedings IASE/ISI Satellite Conference on Assessing Student Learning in Statistics, Guimarães, Portugal, 19-21 August 2007, p. 1-7

Publication year

2007

Keyword(s)

Forecasting methods;  Future;  Mathematical applications;  Mathematical models;  Mathematics education;  Models;  Randomness

Publisher

International Association for Statistical Education

Publisher URL

http://www.stat.auckland.ac.nz/~iase/publications.php?show=sat07

Full text

Peer reviewed