Method and system for carrying out feature analysis on abnormal points based on isolated forest algorithm

• Main Authors: YIN QIAN'AN, LIU SHENG, TAO JINGLONG, WANG QIFAN, YU XIANZHE, LIANG SHUYUN, ZHOU XIAOYONG, MA YING, WEI GUOFU
• Format: Patent
• Language: Chinese; English
• Published: 26.02.2021
• Subjects: CALCULATING; COMPUTING; COUNTING; HANDLING RECORD CARRIERS; PHYSICS; PRESENTATION OF DATA; RECOGNITION OF DATA; RECORD CARRIERS

The invention provides a method and system for carrying out abnormal feature analysis on abnormal points based on an isolated forest algorithm, and belongs to the technical field of data security. Themethod comprises the following steps: for each data point, obtaining a segmentation feature sequence used by a branch path of the data point on each tree; the importance of each feature is calculatedaccording to the positions and the frequency of the features appearing on the feature sequence, and the features are sorted; and selecting an abnormal point according to the abnormal probability output by the isolated forest, and outputting TOP features of the abnormal point according to the sorted feature sequence. According to the method, the abnormal points are deeply explained based on the idea of an isolated forest algorithm, the importance of each feature on the feature sequence is calculated, then the features are sorted according to the importance, and finally the TOP features of theabnormal points are output,