The invention discloses a method for realizing decision analysis of the distribution network public transformer voltage abnormality through data mining. According to the method, a big data processingplatform is utilized to develop characteristic analysis on distribution transformer capacity distribution, load load-ratio distribution, line power supply radius and wire diameter distribution, three-phase unbalanced distribution and voltage abnormality occurrence period and customer complaint distribution through big data technologies including data mining, parallel calculation and decision treesaccording to electricity information acquisition data, power grid rack data and voltage and load operation monitoring data, voltage abnormality control measures are made, and decision-making supportis provided for distribution network planning and operation.