Low-cost and high-resolution data classification method based on constructed prediction model

Abstract

The invention discloses a low-cost and high-resolution data classification method based on a constructed prediction model, belongs to the technical field of classification prediction, and solves the problems of long data acquisition time, high acquisition cost and the like due to the fact that more data needs to be acquired to manufacture a plurality of binary classifiers in the prior art. The method comprises the steps that a binary classifier of a continuous prediction variable is constructed based on any existing data source, and the continuous prediction variable is predicted jointly through n independent variables without time information and m independent variables with time information; According to a prediction result of the binary classifier in a continuous prediction variable range, dividing the continuous prediction variable range into three intervals, namely a left interval, a middle interval and a right interval; Constructing an associated index based on the intermediate interval and the binary classifier; And predicting the data in the left interval and the right interval by using a binary classifier, and jointly predicting the data in the middle interval by using thebinary classifier and an associated index to obtain the classification of the final data. The method is used for classifying the data intervals by using the same data source.

Description

technical field [0001] A low-cost, high-discrimination data classification method based on a constructed prediction model is used for data interval classification using the same data source, and belongs to the technical fields of statistics, machine learning algorithms, and classification prediction. Background technique [0002] In machine learning and statistics, classification is the problem of identifying correspondences within a set of categories (subpopulations) to which new observations belong, based on a training dataset containing observations (or instances) whose category membership is known. For example, assigning a given email to the "spam" or "not spam" class, and assigning a diagnosis to a given patient based on observed patient characteristics (gender, blood pressure, presence or absence of certain symptoms, etc.) . In machine learning terminology, classification prediction is considered an instance of supervised learning, i.e. learning to obtain a training s...

AI Technical Summary

Problems solved by technology

[0009] In view of the problems of the above research, the purpose of the present invention is to provide a low-cost, high-discrimination data classification method based on the constructed prediction model, to solve the classification method in the prior art to classify data, which needs to obtain more data To make multiple binary classifiers, resulting in long data acquisition time, high acquisition cost, high resource consumption and unlocking problems

Images