In the country's economic and social development, banks play an important role in decision-making. The bankruptcy of banks will have varying degrees of impact on enterprises and individuals. For the five tasks given in the title, this paper uses PCA combined with optimization algorithm projection pursuit evaluation + hierarchical clustering, variance analysis, machine learning algorithm, FCM clustering and error inspection model to solve the problem, and achieved relatively good results. Good solution. In order to evaluate the bank's efficiency and solve the dividing line of bank failure efficiency, use MATLAB to first reduce the dimension of up to 19,720 lines of big data to 5 using the PCA algorithm, and then use the projection pursuit algorithm combined with simulated annealing to calculate the optimal index weight, and then Further weighted to calculate the evaluation value; then use hierarchical clustering to aggregate the data into two categories, the midpoint between the center points of the two categories, the corresponding data is the cut-off point, and the connection line of each indicator data is the bank failure efficiency score. boundaries.