There is a problem with incorrect performance placement in the research of self-directed learning efficacy among Chinese online learners, which is an important area to address. Standard Neural Network techniques provide inadequate results when applied to the problem of Chinese online learners' learning performance. Research on the effectiveness of self-learning among Chinese online learners is therefore evaluated and a data mining technique-based study is presented. First, we employ data set theory to identify the factors that have an impact, and then we divide the indicators according to the requirements of the study of self-directed learning effectiveness so that we can reduce the interference factors. After that, we use data set theory to build a data mining techniques study of self-directed learning effectiveness scheme, and we look closely at the results. Data mining approaches beat conventional Neural network methods in a MATLAB simulation when it comes to studying the efficacy of self-directed learning, accuracy, and the time required to influence variables. This is true under certain assessment circumstances.