School Climate Factors as Predictors of School Performance: A Machine Learning Approach

The importance of equity in the provision of quality education is widely recognized. However, in many countries, the structure of the education system has resulted in large disparities in the quality of education provided among schools. To ensure greater equity in education policy makers must focus on targeted school-based interventions. In this study we use Machine Learning techniques to predict the performance of secondary schools using school level factors such as school climate. We fit 5 machine learning algorithms and compared model performance using the root mean square error (RMSE) and the coefficient of determination (R²). We found that the Support Vector Regression with a linear kernel was the most effective model for predicting a school's overall academic performance. However, our focus is really on the importance of these climate factor features on school performance since this information can then be used to make informed decisions for making school improvements. School climate variables such as student academic ability on entry, bullying, school safety, student academic goals and gang activity were identified as strong predictors. We propose that these and other climate factors be periodically collected from each school and used for specific interventions.