Decision making is a process where multiple alternatives need to be analysed. When the complexity and the importance of the decisions increase, formalisation of the process is required. The goal of Multicriteria Decision Aid is to help decision makers facing a multicriteria decision problem that involves taking into account multiple criteria, frequently in conflict, in order to analyse the possible alternatives. Several techniques have been developed in the field of Multicriteria Decision Aid. This thesis is focussed on the PROMETHEE method. The PROMETHEE I method is used to obtain a partial ranking of all the possible alternatives considered, while the PROMETHEE II is used to obtain a complete ranking. Databases with the evaluations of every alternative on every criterion are required in the PROMETHEE method to analyse the different alternatives considered in multicriteria problems. Nevertheless, it is not rare to find databases with some missing values. The aim of this thesis is to analyse how missing values should be managed in the PROMETHEE method. The software Visual PROMETHEE already includes a procedure for dealing with the problem concerning the missing values. The technique is presented in this thesis and other methods are proposed. All the techniques are evaluated and compared to each other to identify which ones perform better in the PROMETHEE method when the databases contain missing values. The methods presented in this study are developed using MATLAB and the same software is used to assess the performance of the techniques. In order to evaluate the methods presented in the thesis, a methodology has been applied using four real databases. First of all, the PROMETHEE II method is used to determine the ranking of the alternatives of a database. After the elimination of one or several values of the database, the ranking of the alternatives is determined again applying all the different techniques. Finally, both results are compared using the Pearson correlation coefficient, the Spearman rank correlation coefficient and the Kendall rank correlation coefficient. The process is performed multiple times and under different conditions. Examination of the results showed that no method performs better than the others in all the conditions considered. Therefore, the procedure implemented in the software Visual PROMETHEE, being the simplest technique to apply, seems to be the most appropriate in the general cases. However, one of the proposed methods led to better results when the criteria of the problem are strongly related to each other. It should be noted that both techniques can be combined in a database, applying one or the other method depending on the particular missing value considered.

Outgoing