Understanding individual students more efficiently is an essential step for educational institutions to improve their services. Constantly tracking students’ learning situations is an important but challenging task. This step is becoming easier thanks to the advances in educational technology. In fact, educational data can be gathered and analyzed more easily due to the implementation and development of multiple educational software systems. It is in this context that we conducted a study to predict the students’ learning experience using a questionnaire. After each lesson, we give the students a questionnaire containing 5 predefined questions relative to their learning activities following the PCN method. Then, students provide their freely-written comments answering these questions. We use these comments to predict their learning experience in an approach to provide meaningful insights to the professor for better interventions. We built models that predict the students’ learning experience using textual data. We investigated 2 approaches to building the models. The baseline approach is to treat all comments of all the questions together. In the second approach, we feed the context of the question to the prediction model using a simple padding technique. Experimental results show that including the context of the question provided a significant improvement in the models prediction performances, attaining an F1-score of 0.74.