DEVELOPMENT OF MODEL METRICS FOR INDIVIDUALS AND PAIR PROGRAMMERS AMONG SOFTWARE DEVELOPERS IN AN AGILE ENVIRONMENT

Abstract

In this work, maintainability as a function of time to correct codes was examined among various categories of software developers. Deliberate errors, ranging from two to ten, were introduced into sets of agile codes written in python programming language and given to 100 programmers each in the groups of Individual Junior, Individual Expert, Random, Expert pairs, junior pairs and Junior Expert pairs. The time spent to correct the errors was analysed using regression model for prediction. Bivariate correlation was used to check the relationships between the number of bugs in projects and the time spent to correct the errors. The correlation between the number of bugs and time of debugging was highly significant, strong and positive. This revealed that the time spent in correcting system software errors increased significantly as the number of bugs increased. Linear, logarithmic, inverse, quadratic, cubic and exponential regression models were used to generate metrics with time spent on error as dependent variable and number of bugs as independent variable for each of the pair and individual programmers. On the average, cubic model gave the highest R2 value of 0.639 in comparison to other models. Therefore, Cubic model gave the best fit as it explains the patterns of the relationship between the dependent and independent variable most appropriately.