2018-12-19 Check-in

Date

19 Dec 2018 

Attendees / Stakeholders

• Clay Crosby

• Swetha Srinath

• Timoteo Chaluco

• Felimone Amone Junior

• Dércio Duvane

• Christine Lenihan

• @Vidya Sampath

• Ben Leibert

Notes

• Antonate may have some availability to work on backend related issues. It's important to identify all of them upfront, however, so that her time can be allocated just once as a single block. This will allow us to avoid any continual and ongoing distraction from her other work. She'll be in the office tomorrow (12/20) and then again starting on 1/7.
  
  
• Speaking of holidays, Timoteo will be off all of next week but available by email. Clay is off next week, working the 2nd and 3rd of January, and then traveling on the 4th. Jason is out until January 2nd.
  
  
• There's a lot of pressure from stakeholders to introduce the system to folks. We need to demonstrate that it isn't vaporware. Swetha Srinath (Unlicensed) is thus visiting Mozambique in mid-January to help Timoteo Chaluco (Unlicensed) compile user-centric documentation and begin teaching people how to use it. It would be nice if Timoteo can send Ben Leibert Portuguese translations asap so as to ensure there's time to incorporate them into the system prior to documentation and training exercises. However, we also need Timoteo Chaluco (Unlicensed) to do quality-assurance on the system, ensuring that it's calculations are accurate. This is critically important and, as mentioned, it's important that we identify any aberrations requiring Antonate's time as soon as possible. Timoteo Chaluco (Unlicensed) will therefore prioritize verification of the system's metrics more highly than those of translations.
  
  
• Timoteo Chaluco (Unlicensed) will be the person on VR's side to officially confirm that the dashboard's metrics are accurate. (Thank you!) Clay Crosby (Unlicensed) volunteered a significant amount of his time to meet with Timoteo, answer any questions, and get him up to speed on how he might validate the dashboard's metrics. (Thank you!)
  
  
•  Clay Crosby (Unlicensed) and Timoteo Chaluco (Unlicensed) will initially focus on validating the Utilization, Breakage, and Trimestral metrics. The Coverage metrics are dependent on population metrics and currently being tweaked. Meanwhile, the stockout metrics are ones that Ben Leibert can help Timoteo Chaluco (Unlicensed) validate.
  
  
•  The best way to validate the dashboard's stockout metrics will be to compare them to those offered within Tableau. Only the stockout section pulls data from SELV - no other part of the dashboard does.
  
  
• As a general rule, Timoteo Chaluco (Unlicensed) will probably want to sample indicators at the national, district, provincial levels for 1 or 2 specific months as well as for perhaps a 6-month period.
  
  

• Because DHIS-2 provides data at the facility-level, it would be extremely helpful to have facility-level population values. There's currently an open email thread attempting to determine whether we do.
  
  

• Just as population metrics are temporal, so are the other coefficients (like target-coverage rates) used on the coverage tab. We therefore created a webpage intended to allow administrators to easily change the coefficients. It's been a while since Ben has looked at the page, though, and he's unsure whether schema changes have broken it. 
  
  

• Pleas consult the project's Docket for a list of pending work. 
   
• Overall, Clay Crosby (Unlicensed) feels comfortable that the system implementation will be ready for user-training by mid-January. It's important to note, however, that there are a number of significant user-facing problems with our version of Superset. The degree to which a Superset upgrade or downgrade will alleviate them is unclear, and they therefore represent a risk. The worst-case scenario may be that we have to describe the bugs to users as part of the training, and assure them that the defects have been reported to the Superset team.
  
  
• An inability to specify our own default time-range within the dashboard's filter seems like a Superset bug. We should nevertheless tinker with it to see whether there's a workaround.