PolyPy: A Web-Platform for Generating Quasi-Random Python']Python Code and Gaining Insights on Student Learning

This Innovative Practice Full paper introduces PolyPy, a freely available web-platform that allows instructors to automate the creation of many different coding question instances from a single template. Interactive web platforms offer a way of engaging students with practice material while also providing them with real-time feedback. However, many of these platforms depend on manually pre-setting a static pool of questions and answers, which limits 1) the number of unique coding challenges presented to students, 2) the ability of instructors to dynamically customize challenges, and 3) the ability to recognize and adapt to individual student needs. PolyPy bypasses these limitations by providing a flexible and easy-to-use framework that lets instructors define question templates. These templates contain segments that can exhibit one of many pre-set values; we refer to these segments as "randomizable elements." The question instances are generated (from the templates) on the -fly every time a student requests a new question via the student interface. To verify the correctness of students' answers, PolyPy interprets these question instances and compares their output to the students' submissions in real-time. All students' attempts, both correct and incorrect, are stored and can be accessed by instructors; this information can be used to gain insight on the strengths and weaknesses of student comprehension. This paper presents the PolyPy system architecture and describes the student and instructor interfaces. It reports promising preliminary results of students' use of PolyPy along with feedback obtained regarding their learning experience.