Aiming to realize the reasonable allocation of water resources in irrigation district, the composition of big data resources of irrigation district was analyzed, and the corresponding data mining algorithms were proposed. The cosine correlation coefficient was introduced, and the fuzzy hierarchical cluster analysis was used to realize the case classification and characteristic analysis of department water distribution based on the data of regional water resources, economy, population and industrial water consumption. A variable step length exhaustive method was used to calculate the dynamic weight of relevant parameters of water distribution in irrigation district. Fuzzy distance was used to match the most similar irrigation district. Weighted influence factors and exponential smoothing method were used to estimate the water demand of irrigation area based on casebased reasoning. The proposed methodology demonstrated the effectiveness in the analysis of department water distribution characteristics and water demand prediction of mediumsized irrigation districts in 11 cities and administrative regions of Zhejiang Province in 2018. The results showed that water distribution among different regions in Zhejiang Province was divided into four categories, showing different characteristics of water distribution among different departments; the relative error of water demand prediction in irrigation area was not more than 9.39%. The established theoretical method could provide decision support for making reasonable regional inter industry water distribution scheme and estimating irrigation water requirement.