Company identification method and device based on multi-dimensional holographic features and logistic regression

The invention discloses a company identification method and device based on multi-dimensional holographic features and logistic regression. Comprising the following steps: S1, carrying out risk labeling and preprocessing on an enterprise data set; s2, extracting features required by modeling from the enterprise data set; s3, training a logistic regression model LR by using the extracted features and the labeling result; and S4, judging whether the enterprise is an empty shell company or not by using the trained logistic regression model LR. The method has the beneficial effects that the multi-dimensional holographic features covering all aspects of the enterprise are collected, the full view of the enterprise can be fully reflected, the biases and unfairness of modeling by adopting a single feature in a traditional method are overcome, and the application range is wide; a logistic regression method with self-learning ability, robustness and stability is adopted, limitation of a rule model and subjectivity of