How to build data science platforms – Part 4: Database scalability and business models

How to Build Analytics Platforms – Part 3: Customizable Workflows and Dashboards

How to Build Data Science Platforms – Part 2: Intelligent User and Role Concept

What does a modern analytics platform need to offer companies real added value?

Why is the administration of user and role rights a factor not to be underestimated when using analytics platforms? In the previous article, we showed how important an intuitive user interface and an open user group concept are for the company-wide use of data science. Now we are developing the idea further – without intelligent user and role management, this concept simply cannot work.

Rights and roles as the first step to new functions!

As a rule, platforms for the use of data science have an almost incomprehensible database with which the respective analyses work and from which reports are generated then. Analytics platforms need the ability to assign individual rights at user and group level in order to process the database efficiently. Only in this way, it is possible to use the capabilities in a targeted manner and at the same time guarantee the security of the database. Not every individual needs complete access. In addition, it is not helpful to equip a single user group with all the extensive admin rights.

It would be even more intelligent if user and role rights would refer to individual components, such as filters or result views. In this way, a single view can display different information without having to provide a separate view for each group. Analytics projects can therefore be implemented more quickly, as they can be reused and extended by adding new components and simultaneously giving new users access via the respective roles without stopping ongoing operations. Thinking one step further again, additional security precautions can be built in when different roles on a platform work together on a project. Role X could then set and customize the analysis. Role Y could see the analysis script, but could not influence it, however, it could process the results accordingly. This point becomes even more important in the area of data discovery, i.e. the recognition of patterns and correlations.

Another important point in the context of role concepts is meaningful integration into existing structures. Ideally, the corresponding analytics platforms are also technically structured in such a way that they function as a supplement to an existing security concept of the database. This means that they can be easily inserted into the cycle of authentification, authorization and authentication. It shows that the “user and role concept” point, intelligently implemented, is the basis for other important factors in the use of analytics platforms.

CONCLUSION: Different user groups contribute to using data science as profitably and company-wide as possible. The next logical step is an intelligent concept consisting of granular and gradual authorizations on the role and component level and thus becomes the basis for other functions.

Outlook

In the next part: How important are customizable workflows and flexible, reusable dashboards?

How to build analytics platforms – Part 1: UI & Teams

Version control – The uncomplicated work on a joint project

Package management: Using repositories in production systems

Ansible: Infrastructure as code (IaC)

Kubernetes: Horizontal scaling of data science applications in the cloud

R, Python & Julia in data science: A comparison

Easy data access: The advantages of a unique database connection with ODBC and DBI