Intelligent data analysis methods for the evaluation of the efficiency of management teams

Abstract

Employing teams of managers for the management of projects is an established trend in modern business. The effectiveness of projects depends directly on the effectiveness of the management teams. When forming a team, it is obviously necessary to take into account the professional qualifications of its members. But it is also important to think of the role each member can play in the team, and whether all the members can effectively work together. According to the model suggested by R. M. Belbin, each team member combines to some extent the roles of the plant, the monitor evaluator, the resource investigator, the co-ordinator, the implementer, the shaper, the completer finisher, and the teamworker. The vector and degree of intensity of these roles forms the role profile of each team member. This article analyses the possibility of applying intelligent data analysis methods for assessing the effectiveness of teams based on the role profiles of their members. We suggest algorithms for obtaining predictive estimates of the efficiency of management teams based on neural network technologies, as well as algorithms for determining association rules for building effective teams. Association rules help to identify cause and effect patterns in the role composition of effective management teams, which can be used when forming new teams or changing compositions of the existing ones. The database for the neural network algorithm is a system whose input is a correlation matrix or a matrix of proximity of team members’ role profiles, and whose outputs are five components of the team’s functional efficiency (planning, organisation, motivation, control, and coordination). Using this database the neural network is trained to recognise outputs depending on the inputs. All the algorithms have been implemented in software and were successfully tested. The article describes the main characteristics of the developed software and the results of a computational experiment.