Evaluation and Validation of Design Metrics in Machine Learning for Depression Detection

Depression is a medical syndrome that is often neglected and considered not as significant as physical illness due to lack of awareness and knowledge, but it should be treated in early stages to avoid extreme cases such as suicide. Therefore, Machine Learning Algorithms (MLA) are used to detect depression primitively to avoid extreme hazards. In-ternational Organization for Standardization (ISO) standards are essential for designing metrics in MLA as they ensure global consistency, quality assurance, and objective benchmark, enhancing credibility and reliability of the model. Although there is no proper standard design level quality metric is defined for MLA. Therefore, in this study, ISO 9126 standard is used to make as a quality standard considering functionality parameter of MLA design. These metrics are used to evaluate at the design level of MLA such as evaluation of dataset, preprocessing, MLA, and results. We have provide comparative analysis of MLA with partial functionality compliance with full ISO functionality conformance. In this study, MLA with ISO have significance compliance which is promising, where the raw data added to MLA to get the output and or the other hand all the steps are performed. The results clearly show that proposed model of MLA design metrics are out-performed provides 28% better results in terms of evaluations. Furthermore, the proposed design quality metrics can be used as the standard for the MLA domain.