What does bootstrap mean?
Bootstrap method bootstrap is an important statistical method in non-parametric statistics for estimating the variance of statistics and then performing interval estimation.
The bootstrap method refers to using the original sample's own data to obtain new samples and statistics. It can be translated as the "bootstrap" method. Some think it can be translated as: self-sampling method. That is, the practice of generating more useful information through existing samples.
"Intuitively, on the basis of the known data, use a computer to simulate the situation when N approaches infinity, and continuously re-SAMPLING the known DATA to obtain the new data. To get the information of the original data. To put it more simply and intuitively, it is: I give you 100 data, but if you feel that 100 data cannot truly reflect the whole picture of the sample, you can re-randomly SAMPLE these 100 data 1000 times, so that you There will be 100*1000 data points. Your sample size will increase a lot."
The idea of Bootstrap is to generate a series of bootstrap pseudo-samples, each sample is the initial data with replacement sampling . Through the calculation of pseudo samples, the distribution of statistics is obtained. For example, if you want to perform bootstrap 1000 times and find the confidence interval of the mean, you can calculate the mean for each pseudo sample. This gives 1000 averages. Confidence intervals are obtained by calculating the quantiles of 1,000 average values. It has been proven that when the initial sample is large enough, bootstrap sampling can be unbiased and close to the population distribution.
The above is the detailed content of What does bootstrap mean?. For more information, please follow other related articles on the PHP Chinese website!