Cable television user audience rating error correction method under large sample

Abstract

The invention discloses a cable television user audience rating error correction method under a large sample. The method comprises the following specific steps: step 1, for a simple random sample of which the sample size of the cable television user is n, the sample audience rating p is unbiased estimation of P, calculating the variance V (p) of the sample audience rating p and the sample unbiasedestimation of V (p); 2, reflecting the accuracy of the audience rating of the sample by using the allowable relative error r; 3, obtaining a relational expression between the sample size and the relative error by utilizing the variance of the sample audience rating p obtained in the step 1 and the unbiased estimation of the sample audience rating p; and 4, correcting the audience rating error according to the relational expression. The method is suitable for cable television big data companies, cable television network companies, television stations and national functional departments, and assists the cable television big data companies, the cable television network companies, the television stations and the national functional departments in quantitatively calculating the sample size required by investigation with the highest precision and the minimum cost during audience rating investigation, thereby further correcting audience rating investigation errors.

Description

technical field [0001] The invention relates to the technical field of data investigation, and more specifically relates to a method for correcting errors in audience ratings of cable TV users under a large sample. Background technique [0002] With the advent of the era of big data and the comprehensive popularization and rapid development of computer technology, the ability to collect, store and analyze data has been continuously improved, and the magnitude of data has continued to increase. Analyzing a large amount of data will cause time and economic costs. massive consumption of resources. For example, there are currently more than 10 million cable TV users' ratings data in the country, and the data structure is complex and the data comes from multiple sources. How to analyze and extract information from such a large number of cable TV subscribers' viewing data. In addition, ratings are an overall proportion estimation problem. Among the existing 200-300 channels, th...

AI Technical Summary

Problems solved by technology

In addition, ratings are an overall proportion estimation problem

Among the existing 200-300 channels, the average audience rating p is about 3 / 1000, which is a low proportion problem and requires higher precision

Images