INTELLIGENT MODELS FOR PREDICTING LEVELS OF CLIENT SATISFACTION.

This paper presents the development of artificial neural network models for predicting client satisfaction levels arising from the performance of contractors, based on data from a UK-wide questionnaire survey of clients. Important independent variables identified by the models indicate that tong-term relationships may encourage higher satisfaction levels. Moreover, the performance of contractors was found to only partly contribute to determining levels of client satisfaction. Attributes of the assessor (i.e. client) were also found to be of importance, confirming that subjectivity is to some extent prevalent in performance assessment. The models demonstrate accurate and consistent predictive performance for "unseen" independent data. It is recommended that the models be used as a platform to develop an expert system aimed at advising project coalition (PC) participants on how to improve performance and enhance satisfaction levels. The use of this tool will ultimately help to create a performance-enhancing environment, leading to harmonious working relationships between PC participants. [ABSTRACT FROM AUTHOR]