Abstract: Affected by the unbalanced distribution of grid storage space, resulting in large data mining cluster registration is not high, in order to improve the efficiency of data mining, proposes an improved algorithm for mining large data association based on fuzzy partition clustering analysis. The grid structure model in the cloud storage space in the data association, semantic rules of feature extraction big data flow of information, the feature extraction of adaptive weighted attribute data processing, enhance the intensity distribution, the extracted data association features optimized clustering using fuzzy partition method in the storage space, the semantic partition based on clustering results, construct discriminant statistics and test criteria for data mining clustering attribute judgment to improve the accuracy of data mining, data mining optimization. The simulation results show that using the method of performance data mining area Better, the accuracy of data classification is higher, and the accuracy of data is improved.