The present invention discloses an automated data mining preprocessing method. The method is characterized by comprising: establishing a database and a preprocessing rule base, establishing a new data table in the database, naming the data table in a standardized mode, sampling to-be-preprocessed data, then importing the sampled data into the new data table, and performing mathematical statistics on a value of each field of the preprocessed data after sampling at the same time; and extracting keywords A, B and C of the data table, querying whether these keywords exist in the preprocessing rule base, and if no, adding the keywords of the data table and all the fields into the preprocessing rule base, and then processing all the preprocessed data by adopting a binning method and a data smoothing method to generate a new rule, and adding the new rule into the original rule base. According to the method disclosed by the present invention, by scoring and feeding back a preprocessing result, a field mapping function is adjusted, so as to improve the preprocessing quality.