Dr Krivitsky will demonstrate how to set questions that require the students to write segments of R code, which are then automatically assessed by writing test cases, also in R. Rather than requiring specific textual output, which can be prone to error, the students are asked to store their answers in R variables with specified names. (Their existence and type can be validated before the students submit their assessments for marking.) These variables are then made available to test cases, which can then check them for correctness.

It is possible to assess both the final results, such as regression outputs and predictions and implementations of algorithms. The latter can be accomplished by asking the student to store an R function that satisfies a particular specification, which can then be run for a variety of inputs to test its correctness.

This was set up in collaboration with Steven Parker and Morgan Harris.


Pavel Krivitsky

Research Area

Learning and Teaching


UNSW Sydney



Mon, 26/10/2020 - 1:00pm