An Enhanced COCOMO II Model to Function Point Analysis to Estimate Effort, Duration, and Productivity using a Modified Backfiring Algorithm in Software Development
Accurate and reliable software effort and cost of the software are very important for project managers in planning and scheduling the software in the software development industry. While backfiring is useful and simple, there is a high margin of error in converting SLOC data into function points. Comparing two most widely used software cost estimation models Function Point Analysis and COCOMO II model can yield different result and error. Modification of the backfiring algorithm is essential for improving the accuracy and reliability of software development effort estimation. This study presented a hybrid model combining Function Point Analysis (FPA) and the COCOMO II framework to enhance estimation accuracy. The study involves developing a user-friendly interface where users input project metrics, and the system computes Unadjusted Function Points (UFP), Complexity Adjustment Factors (CAF), Adjusted Function Points (AFP), and convert Source Lines of Code (SLOC) to FP using a modified Backfiring Algorithm and vice versa. The proposed model's performance has been evaluated using datasets from Turkish, Industry, and NASA COCOMO dataset, demonstrating its robustness and generalizability. The results indicate that the hybrid model significantly improves effort estimation accuracy compared to traditional methods. Performance metrics, including Mean Absolute Error (MAE) and Mean Magnitude Relative Error (MMRE), show that the proposed model provides estimates closer to actual effort, improving accuracy by 9.5%.. Cross validation confirms the significance of these improvements. The findings suggest that the Enhanced model can better inform project managers, reducing cost overruns and improving project timelines. Limitations of the study include the diversity and size of the datasets; which future research could address by incorporating more varied data. Additionally, integrating advanced machine learning techniques could further enhance the model's predictive capabilities. This dissertation contributes to the field of software engineering by offering a practical and improved approach to software effort estimation, supporting more accurate and efficient project management.
Download Paper