PITable-MAX is a maximum frequent pattern mining algorithm based on projection two-dimensional tables. The algorithm only needs to traverse the transaction database once and store the data information in the projection database, and the related data information can be extracted from the projection database to generate a projection two-dimensional table when mining the transaction item, thus reducing the occupancy of memory space. Although the algorithm adopts recursive method, the combination strategy can reduce the size of conditional projection two-dimensional tables to reduce the space, and the number of mining items is reduced by combining the reduction strategy to improve the efficiency of mining. The experiment verifies the feasibility and superiority of the algorithm.