General feature library generation method and device, storage medium and electronic equipment

Abstract

The invention relates to a general feature library generation method and device, and belongs to the technical field of data processing, and the method comprises the steps: obtaining a plurality of data sources, and carrying out the preprocessing of the data sources, and obtaining a plurality of original indexes; Grouping the original indexes according to the variables of the original indexes, andperforming first operation processing on a grouping result to obtain a plurality of basic indexes; Dividing each basic index into a plurality of index classes according to the generality of each basicindex, and carrying out second operation processing on each index class to obtain a plurality of derivative indexes; And summarizing the original indexes, the basic indexes and the derivative indexesto obtain the general feature library. According to the method, the value of the model index is obtained from the general feature library, so that the problem that the index has a large amount of repeated work in the processing stage is avoided, and the value taking speed of the index is increased.

Description

technical field [0001] The present disclosure relates to the technical field of data processing, and in particular, to a method for generating a general feature library, a device for generating a general feature library, a computer-readable storage medium, and electronic equipment. Background technique [0002] With the development of e-commerce, the application of machine learning algorithms in e-commerce is becoming more and more extensive, such as e-commerce anti-fraud, credit model construction, search engine recommendation, etc. [0003] At present, the general data processing process of machine learning algorithms can include: First, analyze the valuable potential indicator characteristics related to the business in the massive data according to different business needs, and pre-set the content of the indicators to be obtained according to the analysis experience; Secondly, perform data extraction, transformation and loading (ETL, Extract Transform Load) to generate pr...

AI Technical Summary

Problems solved by technology

[0005] However, the above-mentioned modeling process has the following disadvantages: on the one hand, it will restrict the value range of the indicators in the model and lack comprehensiveness; Measurability is reduced; on the other hand, for different business requirements modeling, there is a lot of repetitive work in the index processing stage; further, the meaning of indicators for modeling different business requirements is not universal, which is not convenient for the integration and application of models

Images