search

department-of-veterans-affairs / abd-vro

To get Veterans benefits in minutes, VRO software uses health evidence data to help fast track disability claims.
Other
19 stars 6 forks source link

Discovery on Steamlit vs/ PGAdmin #3235

Closed meganhicks closed 3 weeks ago

meganhicks commented 1 month ago

As the VRO Team, we want to understand which tool would be best for our Data Visibility POC to ensure we select the right solution.

Notes about the POC and the Vision for Data Visibility:

  1. We want a tool that offers minimal setup for developers to begin accessing data in RDS but can scale to other roles, such as researchers and product managers, as we continue to iterate on it.
  2. We want a tool that could eventually provide not only data querying capabilities but also data analysis features.
  3. At the very least, the tool should format data into a CSV file, but ideally, it would offer multiple ways to format data for ingestion.
  4. Lastly, we want a tool that provides a variety of self-service options to ensure VRO can continue to scale without the management and maintenance of the tool becoming overly burdensome.

AC: Using the questions below, write an analysis that is no more than two pages long. Spend no more than three business days assessing the solutions. After completing the write-up, share your results with the entire team, including the enablement team. This is not a recommendation, more an informative comparison.

Technical Feasibility:

Are there any constraints or limitations we need to be aware of? Are there existing systems or integrations that need to be considered? Do we have the necessary technical expertise and skills within the team? What are the potential technical risks and challenges we might face? Can the proposed solution scale to meet future demands? What are the long-term maintenance and support requirements? Are there any storage concerns we should consider?

Usability:

Who are the primary users of the product or system? Can those users evolve over several iterations? How intuitive is the interface for first-time users? What data analysis features exist? Are the primary functions easy to find and use without extensive training? How much time does it take for users to complete common tasks? Is the layout of the interface logical and consistent? Can users easily find their way back if they make a mistake or want to try something different? How steep is the learning curve for new users? How is data served up to users?

Effort:

Can we build a POC rather quickly? What are the estimated implementation challenges?

Cost:

What is the total initial cost of the solution? Are there any hidden or unexpected costs we should anticipate? Are there any licensing fees required to use the software or tools? If so, would our partners need licenses as well?

Viability at the VA:

Do we think this solution could continue to scale as we connect to more data services? Do we foresee the VA continuing to allow such a tool?

nelsestu commented 1 month ago

I wanted to offer a quick comment that builds on the pgadmin discussion we had during daily scrum on July 31st. PGAdmin is best suited for use for database administration and management tasks and does not include data analysis or visualization capabilities of its own. With this understanding in mind, some questions to answer with regards to PGAdmin Discovery include the following:

  1. PGAdmin includes a Query Tool that allows users to execute SQL queries against PostgreSQL databases allowing users the ability to display query results in a tabular format within the tool itself. Does this results table satisfy any user requirements by itself? What export options are available from the results table in PGAdmin?
  2. PGAdmin does not offer advanced visualization tools out of the box, for example, PGAdmin should not be confused with the likes of a dedicated BI tool like Tableau or Power BI. That being said, PGAdmin is also widely used for its integration capabilities with other tools. What other visualization and analytics tools might integrate with PGAdmin? Are there Python or other open source libraries like Matplotlib or Plotly that allow for more sophisticated data visualization and analytics?
  3. How would we set about securing the admin and management features of PGAdmin. As a postgres management tool, many of the features and functions of PGAdmin would need to be disabled.
nelsestu commented 1 month ago

Regardless of the solution that ultimately work towards, there is a user authorization element that needs to be considered. I wanted to recommend that we keep the authorization functionality in line with LHDI by once again repurposing github OAuth. Thus we need our discovery work to consider how we can incorporate github OAuth users access into the final solution.