What is the bootstrap algorithm?
Bootstrapping algorithm refers to the use of limited sample data through multiple repeated samplings , re-establish a new sample that is representative of the parent sample distribution. The application of bootstrapping is based on many statistical assumptions, so the accuracy of sampling will affect whether the assumptions are established or not.
In statistics, bootstrapping can refer to all experiments that rely on reset random sampling. Bootstrapping can be used to calculate the accuracy of sample estimates. For a sampling, we can only calculate one value of a certain statistic (such as the mean), and cannot know the distribution of the mean statistic. But through the bootstrap method (bootstrap method) we can simulate the approximate distribution of the mean statistic. With distribution, many things can be done (for example, you can use the results you derived to infer the actual overall situation).
The implementation of the bootstrapping method is very simple. Assume that the sample size is n:
There is sampling with replacement in the original sample, and it is drawn n times. Each time a new sample is drawn, the operation is repeated to form many new samples, through which a distribution of the sample can be calculated. The number of new samples is usually 1000-10000. If the computational cost is small or the accuracy requirements are relatively high, increase the number of new samples.
Advantages: Simple and easy to operate.
Disadvantages: The application of bootstrapping is based on many statistical assumptions, so whether the assumptions are established or not will affect the accuracy of sampling.
The above is the detailed content of What is the bootstrap algorithm?. For more information, please follow other related articles on the PHP Chinese website!