The formula for the growth rate divides by the prior-year value, but when prior values are missing (coded as null), the calculation becomes null, which propagates the missing-data problem forward. Over time, a string of missing values can silence an entire region. Should we implement a forward-fill or interpolation strategy for missing values?
The reversion to an older methodology in 2020 was described as a "return to proven practices," but it effectively invalidates any comparisons across the 2015-2020 period. The methodology used in 2015-2019 is no longer in the codebase, making it hard to audit what changed. Should we add a note that 2015-2020 should not be compared to adjacent periods? See /u/pp_sofia_erebor/p/plot-0052.
Can we label the colour scale? Hard to read otherwise.
Decimals are inconsistent between Nigeria and Mexico.