AUTOMATIC PROGRAMMING ASSESSMENT

Software metrics have been used extensively to provide quantitative measures of software characteristics. This paper aims at evaluating the relevance of using software metrics as means of assessing students' performance in programming. The study focusses on the use of four basic software metrics which are combined to form a single assessment score. The four metrics are respectively those which measure programming skill, complexity, programming style and programming efficiency. Measurements suggested that the lines of code metric is a good candidate for measuring programming skill. McCabe's cyclomatic complexity metrics have been adopted for measuring program complexity. Program execution times are used as the measuring yardsticks for programming efficiency. To facilitate automatic assessment, a program analyzer has been constructed which can provide measures of all the relevant software metrics together with the appropriate assessment scores. The tool was tested with sample assignments of Pascal programs and good distribution of marks has been obtained.