Application of Random Forest Technique in Data Cleaning

Abstract

Data cleaning is an important part of data preprocessing. Its role is to ensure the accuracy of data analysis and improve model performance. Its main tasks are to identify and resolve various problems in raw data, such as missing values, outliers, and duplicate records, thereby providing strong support for subsequent data mining and modeling. In the era of big data, traditional data cleaning methods suffer from poor adaptability and low accuracy when dealing with high-dimensional, nonlinear, and massive datasets. By contrast, random forest has been widely applied in data cleaning because of its resistance to overfitting, strong ability to handle high-dimensional data, low requirement for complex preprocessing, and relatively strong interpretability. This paper reviews recent domestic and international studies on the principles, methods, and research status of random forest in the three core scenarios of data cleaning. It mainly discusses its advantages over traditional methods, its existing limitations, and future research directions. At the same time, visualized content is incorporated to improve the readability of this review and to provide references for future research and applications.