Forecasting for Social Good: Relative performance of methods for forecasting major projects

Litsiou, Konstantia, Polychronakis, Yiannis and Nikolopoulos, Konstantinos (2018) Forecasting for Social Good: Relative performance of methods for forecasting major projects. In: International Symposium on Forecasting, 16 June 2018 - 20 June 2018, Boulder, Colorado, USA.

Preview

Available under License In Copyright. Download (654kB) | Preview

Abstract

Forecasting for Social good, most notably the socio-economic impact of major high-impact projects - like Olympic games or space exploration -is a very difficult but also extremely important task; not only for the resources allocated in such project but predominantly for the great expectations around them. This study evaluates the performances of Unaided Judgment (UJ), Structured Analogies (SA) and semi-Structured Analogies (s-SA) as well as Interaction Groups (IG) in forecasting the impact of such projects. The empirical evidence reveals that the use of s-SA Analogy leads to accuracy improvement compared to UJ. This improvement in accuracy is greater when introducing pooling of analogies through interaction in IG. A smaller scale experiment run to compare Delphi with IGs with inconclusive results.