Informal approaches to predictive modeling: fostering data literacy and critical engagement through decision tree construction

This study investigates how secondary students construct and evaluate decision trees in an open exploratory task using CODAP. By analyzing students’ tree products and oral self-reports, we identified different data-based and context-based approaches to predictor selection and stopping criteria. Students engaged with key ideas of predictive modeling, including classification, model construction, accuracy, generalization, and interpretability. Informal exploration offered rich opportunities for subsequent teaching by enabling students to begin reconstructing central technical ideas and critical tensions of predictive modeling through diverse but meaningful reasoning approaches.