Processed data is fair and appropriate in samples, size and distribution so inferences are meaningful and representative, and personal and sensitive data receive extra care with collection and use minimised under PD(P)O guidance. The organisation strengthens data security and protection compliance, conducts a Privacy Impact Assessment on the necessity of collecting data, pseudonymises or fully anonymises data, removes illegal or harmful information such as false or biased content, masks data where anonymisation is infeasible, processes sensitive data in a single secure work area, identifies outliers in source and modelling datasets, defines and tests for fairness at group and individual level, fixes unbalanced classes, and compares training and test data distributions once the model is tuned.
The graph holds this control, the 0 it maps to, and the evidence behind each claim, over MCP and REST.