Requirement elicitation is one of the early stages of requirement engineering and is critical in the success of any software development project. There is several elicitation methods presented in the literature: interviews, surveys, brainstorming and others; all of which have their strengths and weaknesses. However, the selection of technique is normally arbitrary as software engineers tend to choose based on their own past experiences. This paper aims at developing a new method for identifying the appropriate requirement elicitation technique based on certain characteristics of the project. The approach is based on regression analysis that captures the most important factors that determine the choice of the elicitation technique depending on the project domain. A classification and regression tree model is implemented to systematically identify the optimal technique, reducing the subjectivity associated with requirement elicitation.

Breast cancer is a prominent form of cancer that is frequently detected and is a primary cause of mortality in women. Healthcare organizations prefer to utilize ultrasound imaging as the main method for detecting breast cancer due to its superior safety compared to other imaging techniques. Previous research mainly focused on handcrafted engineering, which is less accurate, time-consuming, and costly. However, we proposed a novel ensemble transfer framework for the early detection of breast cancer via ultrasound imaging. Three powerful transfer learning models such as VGG-16, EfficientNet-B2, and DenseNet-121 are utilized in the development of the ensemble framework. By employing transfer learning methodologies, the framework achieves enhanced computational efficacy in comparison to traditional deep learning approaches. The dataset for ultrasound breast imaging consists of three classes: normal, benign, and malignant. The use of cutting-edge augmentation techniques addresses the dataset's imbalance issues. The experimental findings indicate that the proposed framework demonstrates an accuracy rate of 99.74% in identifying breast cancer on the testing dataset, and it attained a testing loss of 0.013. The proposed framework has shown superior performance in comparison to existing studies.