| [Visitor (112.21.*.*)]answers [Chinese ] | Time :2021-09-08 | Nonparametric test
As mentioned earlier, data analysis can be performed using nonparametric testing for data that does not conform to the normal distribution. Here, data that does not conform to the normal distribution can be divided into two kinds: 1, high-measurement data that do not conform to the normal distribution (fixed-distance data and high-sequencing data); According to the above two data types, nonparametric testing mainly includes the following three aspects:
Test the distribution pattern of the sample
The distribution pattern of high-measurement data series is tested for single variables by examining the difference between the distribution of data series and the standard distribution pattern. If there is no significant difference between the current data series and the standard distribution pattern, the current sequence is considered to satisfy the distribution pattern. Common test techniques for determining the distribution pattern of single sample data are: single sample K-S test, single sample run test, two distribution test, and square test.
The significance test of distribution pattern difference
For high-measurement data series that do not conform to the normal distribution, the common differential significance test methods are: 1, the difference significance test of the two independent samples;
The differential significance test of low measurement data
For class data that do not conform to the normal distribution or low-quality sequencing data, the test method is to calculate the frequency of intersections by using cross-sectional branch, to implement the square test by the square distance, and to analyze whether there are significant differences between different categories of data based on frequency number and data distribution pattern. For the comparative test of class data, it is also called the independence test.
The significance test of distribution pattern difference
The distribution pattern test has been described earlier, and the square test of low-measurement data will be described in the next article. The following focuses on the distribution pattern significance test method for high-measurement data for non-normal distribution.
Nonparametric testing of two associated samples
For two associated samples that do not meet the normal distribution, if they are analyzed for significant differences, their differences can not be compared by mean, usually by comparing their distribution patterns.
The three requirements of the data series: 1, the sample data comes from different perspectives of the same population, or multiple measurements of the same sample; |
|
