Difference minimization random grouping method and system

Abstract

The invention provides a difference minimization random grouping method which comprises: acquiring data information of a plurality of preset groups, wherein the data information comprises at least onecontrol factor and the number of currently grouped objects under a plurality of horizontal parameter categories corresponding to the control factors; constructing a first parameter matrix according to the number of the currently grouped objects under each horizontal parameter category; obtaining basic information of the to-be-grouped objects, and according to the basic information, sequentially classifying the to-be-grouped objects into corresponding horizontal parameter categories in each preset group; constructing a second parameter matrix corresponding to each preset group according to thenumber of the currently grouped objects under each horizontal parameter category after the to-be-grouped objects are added; and obtaining a division probability result of the to-be-grouped objects according to the first parameter matrix and the second parameter matrixes. The method and system solve the problem that the influence of global imbalance on the algorithm processing result is not considered in the traditional minimization random algorithm, so that the algorithm has stronger balance and adaptability.

Description

Technical field [0001] The invention relates to the technical field of machine learning, in particular to a random grouping method and system for minimizing differences. Background technique [0002] The minimization method was improved on the basis of the partial coin toss by Taves in 1974. At present, the research on the minimization random algorithm at home and abroad focuses on the statistical theory, statistical efficiency, bias control, and prediction of the algorithm. In terms of performance or application, the main focus is on the correct application of the algorithm, and there are very few studies on the analysis and minimization of stochastic algorithm defects and their improvements. [0003] The core idea of ​​the traditional minimization random algorithm minimization method is that every time a new object to be grouped is added in the experiment, it is first calculated and divided into all possible treatment groups, which leads to the inconsistency caused by the experim...

AI Technical Summary

Problems solved by technology

[0004] In view of this, the embodiment of the present invention provides a random grouping method and system for minimizing differences to solve the traditional minimization random algorithm in the prior art, which does not consider the global changes of the objects to be grouped on all levels of each control factor and Issues affecting the balance between different treatment groups

Images