Analysing decision variables that influence preliminary feasibility studies using data mining techniques

Abstract

The development of infrastructure contributes to the social and economic improvement of a country, and generally requires huge and immediate investments. To decide on appropriate infrastructure projects, many countries use preliminary feasibility studies (PFS). However, a preliminary feasibility study takes a relatively long time to complete. During this time, decision‐making parameters such as the estimated project cost as well as the project environment may change. To identify the decision parameters that affect the feasibility analysis, data mining techniques are applied to analyse the Go/No Go decision‐making process in infrastructure projects. The data mining analysis uses PFS data obtained from large‐scale infrastructure projects in Korea. Classification models found 57 hidden rules in the PFS. Prediction models were also developed for Go/No Go decision making using an artificial neural network (ANN) and logistic regression analysis. In order to validate the results, the study evaluated the accuracies and errors of both the classification and the prediction model.