The general data science process

Data science projects have a general process that the majority of well-run projects follow. Let's outline the overall data science approach to a project to ensure that we have a shared understanding of the approach. The structure of the team is irrelevant to this process. Any data science team will execute a project process for most data science-related projects that are similar to the following list of steps:

1. Project ideation
2. Engagement with project sponsors and subject matter experts
3. Project charter initiation
4. Project charter refinement
5. Project management
6. Convening team meetings
7. Obtaining internal and external data
8. Testing various analytical techniques
9. Building analytical models
10. Designing the user interface (UI) and user experience (UX)
11. Presenting interim results
12. Discussing the level of success or failure in the modeling process
13. Planning for the testing of models and applications in the user workflows and daily processes
14. Planning for the production implementation of models and applications
15. Presenting the models and applications to end users
16. Receiving and acting upon suggested changes to the data, models, and applications from users, sponsors, and subject matter experts
17. Turning the final products over to the information technology team to integrate into the technology architecture
18. Turning over the final products to the end users for use in their daily workflows
19. Setting up the rhythm of communication for the on-going maintenance of models and applications
20. Beginning the next phase of work on the models or applications or moving on to the next project

The recommended process flow above works well for data science projects that are managed in the traditional waterfall project management methodology, but the process also works well in the agile project management methodology. Recently, we have been managing projects with the agile approach and the results have been reliable and repeatable. Either approach works well. My experience has been that the project management methodology used depends more on the training of the project management team and the culture of the company rather than any factor related to a data science project.