Feature selection method for power distribution network data optimization

The invention discloses a feature selection method for power distribution network data optimization, and belongs to the field of power data analysis and processing. The method comprises the followingsteps of quantifying a certain type of influence factors of power distribution network faults according to the related data source to enable the influence factors to become fault characteristic variables; preprocessing and discretizing the data matrix according to the average value of each fault expression; wherein the output feature number n is provided by a user from the outside, and then inputting a category data matrix; calculating the target value A through mutual information maximization according to the correlation of each fault; repeatedly updating the remaining output features in a circulating manner; setting a target value B model function as a ratio of the characteristic correlation index to the redundancy index, and maximizing the ratio; and sequentially and circularly sortingout the characteristic vec