Contextual factors have been identified as greatly influencing students' reading performance. However, the collaborative influence of key contextual factors on students' reading performance is still elusive and warrants further exploration. Based on Walberg's educational productivity theory and Bronfenbrenner's ecological system theory, emphasizing that learning in humans can only be understood by considering the influence of multiple factors combined into a unit or system, the current study sought to identify the optimal factor set of key contextual factors that collaboratively influences fourth-grade students' reading performance. In this study, data from 183,428 students from 61 countries/regions were extracted from the progress in international reading literacy study 2016 dataset. First, a support vector machine (SVM) was adopted to classify the contextual factors influencing high-performing (students whose reading score is above 550) and low-performing (students whose reading score is below 475) students. Second, SVM recursive feature elimination (SVM-RFE) was applied to identify the key contextual factors capable of differentiating the two student cohorts. The findings indicate that 20 key contextual factors selected from 106 contextual factors at the student, family and school levels collectively differentiate high- and low-performing students, providing implications for future teaching and learning on elementary school students' reading performance.