Analysis and Prediction Model Construction of Educational Achievement Based on Data Mining

Based on the data of middle schools in a certain area, this paper explores the complex relationship behind students' grades through data preprocessing, feature selection and machine learning technology, and establishes a prediction model. Data preprocessing technology solves the problems of missing values, abnormal values and repeated values in the original data, and improves the data quality. Feature selection algorithm is based on mutual information (MI) to evaluate the dependence between features and students' grades, which reduces the data dimension. As a machine learning prediction model, random forest (RF) algorithm improves the accuracy and stability of the model because of its integrated learning characteristics. The empirical analysis shows that the $\mathbf{R}$ score of the model is 0.85, and the distribution of the predicted results is close to the actual results, which shows that the model has high prediction accuracy. In addition, the model reveals the key factors affecting students' performance through the score of feature importance, which provides the direction for teaching improvement.