A new dynamic multi-attribute decision making method based on Markov chain and linear assignment

This paper presents a new Dynamic Multi-Attribute Decision-Making method based on Markovian property, which can predict the performance of each alternative in the future and at the same time allows modeling interrelationship among different periods. To this aim, the criteria and decision alternatives in different periods are determined at first, and the information of decision matrices over the decision-making horizon is gathered. To increase the robustness of the results, criteria weights are extracted using the Entropy method in each period and alternatives performance is evaluated using different Multi-Attribute Decision-Making methods. To attain the final rank of alternatives in each period, the results of different methods are aggregated by the Correlation coefficient and standard deviation method. Following this, the rank transformation matrices of alternatives during the evaluation horizon are extracted and the stable rank probability of alternatives is calculated based on limiting probability. Eventually, the overall rank of alternatives is determined using a linear assignment-based method. The proposed model has been used in the promotion of the sales staff in a private company to show the model effectiveness in a real-world problem. Results are compared with some well-known methods (five methods, to be exact). Finally, the trustworthiness and acceptability of the method are assessed based on features discussed in the literature.