Short Paper: An Initial Investigation of the Correlations Between the Quality of Engineering Assignments and Task-Independent Features

Abstract The potential development of automated grading brings us the opportunity to change the way of teaching. In this paper, we present our pilot study of the correlations between the scores of engineering assignments and several task-independent features. This work aims to identify features that can subsequently be used in a machine learning framework to support automated grading. We also identified some interesting patterns in this dataset. Our result shows that length features such as word count may indicate a highly structured assignments quality, and the number of grammar errors has the potential to be part of the evaluation if we use a carefully designed grammar checker. The present study is expected to contribute to our understanding of automated grading of mechanical engineering technical reports and provide insights into rubric design for assignments.

Short Paper: An Initial Investigation of the Correlations Between the Quality of Engineering Assignments and Task-Independent Features Conference Paper