According to the characteristics of multi-objective, fuzziness and uncertainty of water resource allocation scheme, projection pursuit method was used to project high dimensional data into two-dimensional space, and combined with the clustering algorithm and multi-agent genetic algorithm，a projection pursuit clustering comprehensive evaluation model (MAGA-PPC) of water resources allocation was established. The validity of the model is verified by the correlation analysis of the index to ensure that the optimal solution is obtained. Meanwhile, the Hanjiang-to-Weihe water transfer project in Shaanxi province was taken as an example, and the evaluation results show that MAGA-PPC can comprehensively and systematically quantify the differences in social, economic, ecological, resources aspects of various programs, and this evaluation result is consistent with the development and utilization of water resources in the study area. Moreover, MAGA-PPC has good practicability in the evaluation of rational allocation of water resources.