python工具包：random的注意事項

問題

我們知道在由於random包中生成的是僞隨機數，因此通過設置seed可以固定住隨機結果。

但有一次使用時發現設置完seed之後結果依然改變：

代碼如下：

 19     random.seed(args.seed)
 20
 21     # read all filenames in list files
 22     with open(args.file, 'r') as f:
 23     ¦   filenames = f.readlines()
 24     filenames = [item.replace('\n', '') for item in filenames]
 25     filenames.sort()
 26     random.shuffle(filenames)
 27     print('total filenames: {}'.format(len(filenames)))
 28     print('==== filenames[:4]', filenames[:4])
 29
 30     # select patient, if None, all patients are selected
 31     patient_ID = [filename[:4] for filename in filenames]
 32     set_patient_ID = set(patient_ID)
 33     print('total {} patients'.format(len(set_patient_ID)))
 34     selected_patient_ID = list(set_patient_ID)
 35     #selected_patient_ID.sort()
 36     selected_patient_ID = selected_patient_ID[:args.patient_num]
 37     #selected_patient_ID = list(set_patient_ID)[:args.patient_num]
 38     print('select {} pathents'.format(len(selected_patient_ID)))
 39     print('selected pathent IDs:\n', selected_patient_ID)
 40     print('-'*100)
 41     print('===== selected_patient_ID[:4]', selected_patient_ID[:4])

解決方法
我們發現28行每次輸出固定，但是41行每次輸出不固定。

查找原因，發現是第32行每次將list改爲set之後，會被自動排序，但還是沒理解爲什麼會導致出現了隨機。

解決方法是增加第35行代碼即可，如下：

 19     random.seed(args.seed)
 20
 21     # read all filenames in list files
 22     with open(args.file, 'r') as f:
 23     ¦   filenames = f.readlines()
 24     filenames = [item.replace('\n', '') for item in filenames]
 25     filenames.sort()
 26     random.shuffle(filenames)
 27     print('total filenames: {}'.format(len(filenames)))
 28     print('==== filenames[:4]', filenames[:4])
 29
 30     # select patient, if None, all patients are selected
 31     patient_ID = [filename[:4] for filename in filenames]
 32     set_patient_ID = set(patient_ID)
 33     print('total {} patients'.format(len(set_patient_ID)))
 34     selected_patient_ID = list(set_patient_ID)
 35     selected_patient_ID.sort()
 36     selected_patient_ID = selected_patient_ID[:args.patient_num]
 37     #selected_patient_ID = list(set_patient_ID)[:args.patient_num]
 38     print('select {} pathents'.format(len(selected_patient_ID)))
 39     print('selected pathent IDs:\n', selected_patient_ID)
 40     print('-'*100)
 41     print('===== selected_patient_ID[:4]', selected_patient_ID[:4])