JSON file with requisition data, same as what's served by the API
- useful keys for Full Availability indicator
  - period
    - start date
    - end date
  - facility location
  - stockout days
  - product ID
  - product name
  - commodity type
  - geographic zone, e.g. district
  - [latitude, longitude]
JSON data with mapping from geographic zone to region
- maybe shape files for geographic zone and region
JSON data with mapping from product to commodity
- use this to find the commodity from the product ID
JSON file stock card data, same as what's served by the API
- useful keys for Full Availability indicator
  - date
  - facility location
  - number of stock on hand
    - when this hits zero we do not have full availability for the reporting period

Possible responsibilities for OpenLMIS core team to support the reporting work:

...

OpenLMIS is also adding HL7 FHIR facility/location master lists. Reporting stack could use that when it becomes available.

Question: How often does the data change and how frequently does the report need to be refreshed?

In terms of what is practical, real time is practical.
- There may be situations where we need to "split the columns" and re-process the historical data to re-run a query with a new grouping. Those changes will not happen real-time.
In order to demo the system, we would need to run/refresh the data during the live demo event.

Question: Where does Tableau fit in this picture?

It is sort of like Superset. You could point Tableau at this. Query Hadoop through Tableau, or index Hadoop through Druid. Then Tableau could query Druid. Tableau can use HIVE to provide a JDBC or ODBC SQL way to query the Hadoop data.

Image Added

Mobile

Two different diagrams from earlier morning conversations about Mobile (black drawing from Mary Jo) and Reporting (blue drawing from Peter):

...

Versions Compared