Improving the accuracy of COCOMO II using fuzzy logic and local calibration method

Software Cost Estimation is a problem that often appears on the project of making a software. A poor estimate will result in a bad project management. To resolve this problem, a lot of the estimation model was introduced. Constructive Cost Model II (COCOMO II) is the most known and widely used model for software cost estimation. To estimate the cost of a software project, COCOMO II uses cost drivers, line of code and scale factors. However, the COCOMO II is still lacking in terms of accuracy. To improve the accuracy of COCOMO II, this research studies the influence of cost drivers and scale factors on improving the accuracy of effort estimation. Then, This Research using Fuzzy Logic and Local Calibration to improve the accuracy of COCOMO II. Fuzzy Logic with Gaussian Membership Functions (GMF) has been applied to redesign the effort multiplier. This research also implements Local Calibration to make a better accuracy of COCOMO II by using the value of the dataset as the input for the calculation of the calibration. This research gives a value effort multiplier that better suited for use in estimating and improving the method for search a new value of parameters to calculation COCOMO II that is more suitable for a category of a dataset used in this research. The Local Calibration result can be used for a future software project with similar category with the dataset used in this research. The result is a proposed model contributes on decreasing error significantly (about only 10% from MMRE COCOMO II or fuzzy model).