Various models for the data science workflow, each differing slight according to the field of application, have an important role. In addition, data scientists have various tools they can leverage. To launch AI initiatives, wealth and asset managers need to understand what tools and what skills need to be combined.
In this report, we discuss the Data Science workflow, which comprises the major steps of proposing business goals, specifying data requirements, collecting and retrieving data, exploring data, cleaning and transforming data, sampling data, modeling, evaluating and testing, deploying, and monitoring the applications. This workflow is amenable to automation, a topic increasingly referred to as DataOps. DataOps is a sibling of DevOps for data science, which aims to allow more effective industrialization of data science in practice, through:
- Improved repeatability of findings.
- Reduced time to identifying actionable insights.
- Decreased time to impact.
