Software design phase risk estimation using extreme learning machine

In the field of software engineering, the software development process is used to assist the software development team and keep control during the development of software. The software industry is expanding rapidly to meet the rising demand for software, and as a result of this, software development risks and budgets are increasing. Software development is a complex process, and risks may arise when endeavoring to develop reliable and unvarnished software. Risk assessment models have been developed by many researchers, but they have not been able to deal with risks because the risks always change, and it is difficult to know exactly what they are. Due to its dynamism and uncertainty, it is difficult to estimate the risks accurately. To overcome this issue, there is a need for machine learning (ML) techniques that can evaluate and manage the risks associated with software development. The use of ML is gradually gaining popularity across many domains. This research proposes a machine learning-based technique for estimating software development risks during the software design phase. The research uses the Extreme Learning Machine (ELM) approach to perform a software design phase risk estimation (SDPRE). Evaluate the effectiveness of the proposed framework by comparing its risk estimation performance with other machine learning methods. Risk factors that impact risk assessment during SDPRE have been identified. The Kaggle dataset has been used to estimate the software design phase during the early stages of software development. The ELM model for software design-phase risk estimation achieved the best performance among the evaluated machine learning models. This research shows that the proposed ELM model in the SDPRE framework achieved the highest estimation accuracy and superior performance compared with K -Nearest Neighbors (KNN), Linear Regression (LR), Support Vector Machine (SVM), and Multilayer Perceptron (MLP). This research will provide a structured, learning-oriented model capable of producing accurate results with the highest level of consistency when estimating the risks in the software design phase.