MWI 2016-20

[QLLZ]

Data pull: HASH: 10db6c83d6486774d3148785df24b3145f375bb2 config

[QLLZ]

reran the initial values data after the new updates on initial values HASH: 6772ea9da4f43ddc3da11c494a573af2ca3dd3c8

[QLLZ]

Model run: HASH: 6772ea9da4f43ddc3da11c494a573af2ca3dd3c8

[QLLZ]

Prestan report

Country data report

[eclee25]

Convergence, maps, Rhats, fits, match to GAM input look good. Some small sfrac cells seem to have large rates on lake border

Opinion: Approve

[QLLZ]

Config

Data pull:

HASH: 211913e0eb7d570ed62c48bc1ae9ecd6671a6cbc

[QLLZ]

Failed stan model run log file

[QLLZ]

Rerun on dev_u_combs_fix

HASH: e56580fa5ddab00293e31ff90139351f80ae3f6c

[QLLZ]

country data report

[eclee25]

Diags look good but annual estimates for 2016 and 2018 seem a bit high -- This is similar to what we saw in the last round of production runs and the magnitude of the overestimation seems the same as well.

Temp Approve - someone else should review the report

[javierps]

Sep 2023 Production run: convergence OK, over-estimation of cases in 2018.

Suggestion: 2nd opinion.

[eclee25]

Continued overestimation in 2016 & 2018, similar to previous runs. sd_w and rho convergence issues could potentially be improved with a no-mixture model, but I don't think this will improve the over-estimation issue.

For Discussion - Temp Approve

Action Required: Investigate subnational sum of cases vs national case counts -- Is the subnational data suggesting that the 2016 and 2018 observations should be larger?

[QLLZ]

The subnational data suggests that there are around 793 cases in 2018 and around 1920 cases in 2016. These are similar to the national data (around 800 in 2018 and 1792 in 2016).

[eclee25]

Discussed as a group and decided to Approve