In-Depth Feedback on Programming Assignments Using Pattern Recognition and Real-Time Hints

Automatic grading of students' programs is a challenge in computer science (CS) education. In the last two years we have been using a machine learning tool called 'Sense Education' in our '\Intro to CS' course. The tool provides real-time pedagogical hints while students work on solutions, and in-depth feedback on submissions. It also provides a bird's eye view of the class's current capabilities, misconceptions, and biases in approaches to solutions. We describe this effort and several test cases. Our goal is to provide teachers with tips on how to effectively use such tools.