Implementation of random forest regression for COCOMO II effort estimation

One of Project Manager early activity is to estimate time, and cost based on given scope, which can help project manager to plan schedule and used resources. Estimation is very important in project management because a bad result of estimation will result in bad management of project and may cause failure. There are methods that can be used to estimate software development effort; COCOMO II is one method that commonly used. Many researcher before have been used algorithm, such as Bat, Bee Colony, or MOPSO to increase COCOMO II estimation accuracy. However, as the technology advanced, there are a lot more options that can be used to predict software effort estimation based on COCOMO, such as machine learning. In this paper, we compare machine learning algorithm with tuning parameter method to know whether tuning parameter estimation is better than machine learning estimation or vice versa. In this paper, we use Random Forest Regression as machine learning algorithm to estimate the effort. We also compare it with another machine learning algorithm, Support Vector Regression, and Bee Colony Method as parameter tuning method. The results of experiment is evaluated by their error rate. The results show that Random Forest Regression is better than Support Vector Regression and Bee Colony Method.