An Noncompliance Identification Method for False Labeling Outage Based on ReliefF Feature Selection and Random Forest Algorithm.

SUN Xinyuan

Accurately identifying whether the enterprise shutdown mark is compliant is an effective means to effectively combat the falsification of automatic monitoring data. As a representative model of machine learning, random forest has the advantages of high recognition accuracy and strong model generalization ability, but its computing speed and processing efficiency are not high. In order to solve such problems, a random forest automatic monitoring shutdown mark non-compliance model based on ReliefF feature selection is constructed, which is based on the principle of selecting the optimal feature subset and assigning corresponding weights through the ReliefF algorithm, so as to accelerate the calculation speed, reduce the amount of calculation and improve the processing efficiency. In order to verify the rationality and accuracy of the model, 100 non-compliant enterprises with shutdown marks verified in 2024 were selected for verification, and the model identified 98 non-compliant labeling enterprises, with an accuracy rate of 98% and a high accuracy rate, the model verified the rationality of machine learning application in off-site supervision. The composite model combined by preprocessing algorithm and machine learning algorithm has many advantages such as high recognition accuracy, fast calculation speed, and strong model generalization ability, and can be widely used in the field of environmental law enforcement to increase productivity quickly.

RANDOM forest algorithms; FEATURE selection; FOREST monitoring; PROBLEM solving; LAW enforcement

Environmental Science & Technology (10036504), 2024, Vol 47, Issue 11, p229

1003-6504

Academic Journal

10.19672/j.cnki.1003-6504.1040.24.338