Developments in fuzzy cognitive maps and their applications to enterprise resource planning readiness assessment

Fuzzy Cognitive Maps (FCMs) have been widely used for modelling and analysing causal relationships and feedback between components of a complex system. This research develops two efficient and effective algorithms for learning FCMs. This research also develops a structured FCM-based methodology for assessing, managing and planning change readiness for implementing a complex information system involving many interrelated readiness-relevant factors. Modelling and analysing a system using the FCM technique can be done by using either experts’ knowledge or historical data about the system. This research has two main objectives: (1) developing new, accurate and fast algorithms for analysing historical data to develop FCM models, and (2) developing a FCM-based methodology using experts’ knowledge for analysing and managing the readiness of an organisation for implementing a complex information system. To achieve the first objective, this research investigates whether the new forms of evolutionary algorithms provide a faster and more accurate algorithm for automated FCM learning using historical data. Two new types of fast and accurate FCM learning algorithms are developed by using two evolutionary optimisation algorithms called the Cultural Algorithm (CA) and the Imperialist Competitive Algorithm (ICA). The two new FCM learning algorithms resulted from using these two algorithms are the Cultural Learning Algorithm (CLA) and the Imperialist Competitive Learning Algorithm (ICLA) respectively. Comparing these two new FCM learning algorithms with many existing and well-known learning algorithms, such as Real Coded Genetic Algorithm (RCGA) and divide and conquer RCGA, shows that in most cases they outperform existing algorithms in terms of accuracy and speed. The experimental comparison between CLA and ICLA also shows that the accuracy of CLA is slightly better than ICLA, whereas ICLA is more efficient than the CLA. To achieve the second objective, this research addresses two issues: (1) how causal relationships between interrelated readiness-relevant factors or activities can be modelled for assessing an organisation’s readiness for a major change, and (2) how an organisation can develop the most cost effective plan for improving the readiness. This research addresses these two issues in the context of a new Enterprise Resource Planning (ERP) system implementation. To address the first issue, a structured FCM-based methodology for managing interrelated readiness-relevant factors or activities to implement a significant change is developed. The methodology can (1) identify the readiness-relevant factors, (2) determine how these factors influence each other, (3) assess how these factors contribute to the overall readiness, (4) cluster the factors into manageable groups and (5) prioritise the factors according to their causal interrelationships to find the most influential factors for allocating readiness improvement management effort. In addition, based on the contribution of the factors on the overall readiness and their interrelationships, the factors can be categorised into four management zones for effective allocation of limited management efforts. To address the second issue, an FCM-based problem solving methodology is developed to find the most cost effective plan for improving the readiness. This methodology finds the best readiness improvement plan as a multi objective trade-off between the two objectives of readiness maximisation and cost minimisation using the Nondominated Sorting Genetic Algorithm II evolutionary algorithm. The result of solving this problem is a set of optimal improvement plans from which the best plan for meeting an organisation’s specific needs can be determined. This research has made significant methodological and practical contributions to both FCM and ERP research. As a methodological contribution to FCM research, this research develops two new, effective, and efficient automated FCM learning algorithms. As a methodological contribution to ERP research, this research develops a structured readiness management methodology with methodological advances including (1) modelling causal relationships between interrelated ERP readiness-relevant factors or activities, (2) reflecting the imprecise nature of the readiness assessment process, (3) prioritising readiness-relevant factors based on their causal relationships with other factors, (4) reducing the complexity of the readiness assessment model by clustering factors into manageable groups, and (5) analysing the readiness assessment model to find the most cost effective areas and factors for improving the overall readiness. As practical contribution to ERP research, this research shows how the new structured readiness management methodology can help an organisation (1) identify the main areas required for management focus and (2) analyse the readiness-relevant factors to determine the individual factors which should be improved for achieving the highest readiness improvement. These are significant contributions to the practice of ERP implementation. The methodology developed and the outcomes produced in this research are valuable for an organisation to effectively manage its readiness relevant factors or activities during the pre-implementation stage of an ERP system.