Research on outlier detection of data based on machine learning

With the increasing magnitude of data, the accuracy of single data cannot be guaranteed. In order to improve the accuracy of model prediction data, and further for the subsequent data processing, this paper focuses on the detection of data on outliers. This paper introduces the density clustering and outlier detection methods of Isolation Forest in outlier recognition, and mainly describes the principle and process of outlier recognition using density clustering and Isolation Forest. Based on this, according to the features of data, an improved algorithm combining density clustering and Isolation Forest is proposed. Finally, through the existing common outlier detection data set, the statistical outlier recognition method, the existing machine learning algorithm and the improved algorithm proposed in this paper are compared to identify outliers, which verifies the stability of the proposed method and the effective improvement compared with the existing outlier detection.