Forecasting the software engineering model’s effort estimation using constructive cost estimation models

Machine learning models are increasingly being developed to solve various problems in academic and corporate settings. However, data scientists encounter multiple challenges during model development that require software engineering methodologies to overcome. In many cases, developers working on machine learning projects may need to fully realize the benefits of adhering to the steps outlined in the Software Engineering Development Lifecycle. It is essential to recognize that the development processes for machine learning systems differ from those of traditional software systems. Therefore, it is crucial to investigate how software engineering principles can be adapted or applied to the machine learning workflow. This paper explores such practices and problems from a software engineering perspective. To achieve this, we utilized the COCOMO II Model to predict a software engineering model’s effort estimation and effort estimation incorporating machine learning hybrid techniques. Specifically, we employed a Genetic Algorithm-based Support Vector Regressor, which yielded the highest PRED value for effort estimation using the COCOMO II Model. Overall, this study provides insights into the application of software engineering methodologies for building machine learning models and highlights the importance of incorporating these practices to optimize the development process and achieve more accurate effort estimation.