An empirical study of factors impacting the size of object-oriented component code documentation

In this paper, we identify a set of factors that may be used to forecast source code documentation for object oriented software components. Using field data on 152 object oriented (OO) software components and the multiple regression model, we empirically test the impact of proposed factors on the prediction of OO source code documentation. Our results indicate that the number of methods, the number of sub-classes, the number of GUI elements and the number of events handled all increase the number of lines of code for OO source code documentation. Additionally, we show that scale economies exist in OO source code documentation and use non-linear regression model Classification and Regression Tree (CART) to improve OO source code documentation forecasting accuracy.