Since unsupervised machine learning systems are widely used in many real-world applications, assessing the appropriateness of these systems and validating their implementations with respect to individual users' requirements and specific application scenarios$\,/\,$contexts are indisputably two important tasks.Such assessment and validation tasks, however, are fairly challenging due to the absence of a priori knowledge of the data.
FDA regulations such as GMP, GLP and GCP and quality standards such as ISO17025 require analytical methods to be validated before and during routine use.
During the validation phase the requirements are evaluated against a question “Do the requirement specify the right product? We check with the stakeholders whether the requirements specify the product/service/change they really want.
After the stakeholders approve the requirements and commit that these requirements are what need to be delivered, they are base lined and form a kind of contract for the rest of the project.
While reading the document, the reviewer tries to answer the questions and find defects in the document.
Each reviewer applies only a single scenario/looks for one fault class only.