Python 实战常用操作（2）

Booststrap随机采样

在随机森林中或者对大样本进行评估时经常用到 booststrap随机采样方法
Bootstrap是现代统计学较为流行的一种统计方法，在小样本时效果很好。其步骤主要是从样本中有放回重复采样，样本数可自行决定。在随机森林中，一般会用到1000次。比如样本a如下：
a=[‘2020-05-14_070053.JPG’, ‘2020-05-14_074724.JPG’, ‘2020-05-14_074819.JPG’,
‘2020-05-14_075839.JPG’, ‘2020-05-14_080441.JPG’, ‘2020-05-14_083346.JPG’,
‘2020-05-14_083415.JPG’, ‘2020-05-14_083605.JPG’, ‘2020-05-14_083609.JPG’,
‘2020-05-14_084512.JPG’]

files_sample = get_subsample(files_special,30/len(files_special))

files_sample = [‘2020-05-15_110400.JPG’, ‘2020-05-18_074556.JPG’,
‘2020-05-18_102641.JPG’, ‘2020-05-14_145305.JPG’, ‘2020-05-18_084513.JPG’, ‘2020-05-18_085349.JPG’,
‘2020-05-18_084513.JPG’, ‘2020-05-18_140645.JPG’, ‘2020-05-14_145119.JPG’, ‘2020-05-18_131855.JPG’,
‘2020-05-18_090435.JPG’, ‘2020-05-15_130147.JPG’, ‘2020-05-14_150905.JPG’, ‘2020-05-18_103104.JPG’,
‘2020-05-18_094705.JPG’, ‘2020-05-18_122045.JPG’, ‘2020-05-14_102612.JPG’, ‘2020-05-18_082003.JPG’,
‘2020-05-17_142413.JPG’, ‘2020-05-18_090737.JPG’, ‘2020-05-14_134729.JPG’, ‘2020-05-18_122140.JPG’,
‘2020-05-14_155347.JPG’, ‘2020-05-18_131609.JPG’, ‘2020-05-15_134928.JPG’, ‘2020-05-14_150548.JPG’,
‘2020-05-15_091451.JPG’, ‘2020-05-14_114317.JPG’, ‘2020-05-14_133010.JPG’, ‘2020-05-18_103126.JPG’]

可见，从维度为20的总样本中可抽取了30个样本，足见其为重复采样。