Summary

In this chapter, we highlighted the value of establishing a framework guiding the use of ML models in businesses. ML governance capability supports users in ensuring that ML models continue to deliver commercial value while meeting regulatory expectations. Also, we set controls for what different levels of stakeholders can do with ML deployments. In some industries, there is a need to seriously consider the impact of bias in any decision process. Because ML models are based on data that might have been affected by human bias, it is possible that these models will compound such bias. As such, we explored ways to mitigate ML bias during and after model development.

We also examined the effects features have on the outcome variable. Such changes could have a critical bearing on business outcomes, hence the need to monitor the performance of model outcomes in production. During this chapter, we explored ways the performance of models could be assessed over time. Importantly...