Outlier Checks & Create/Implement Cleaning Logs

[zacka-cartercenter]

Should check for outliers

visual inspection -> flag suspicious values - > modify in log.

• [x] graph/visualize outliers
• [x] make log template

Initial checks of outlier values reveal some places where shifts from zone level to woredas level reporting was not completely accounted for

• [x] create consistent admin 2 level data set
• [x] create consistent admin 3 level data set (where possible)
• [x] redo outlier visualization
• [ ] finalize log template

[zacka-cartercenter]

bench marking reported treatment values against total population values: but some minor data cleaning steps on reported tota population values:

• total_popn_census & total_popn_projected appear to be used interchangeably - will create new col (ifelse)
• total pop missing in various cases:
  • 2020-02, 2020-03, 2020-04 : no activities - can fill in with 2020-01 values or potentially delete later
  • 2016 april-october - all blanks were reported for East Gojjam & West Gojjam - strong case for omitting these two adm2 for entire year
  • 2021-01 South Omo - Semen Ari district - new district that year, use the first value that appears after
  • 2021-09/10/12 Metekel pop is blank - take values from LF sheet... for 2021-12 Bulen pop is blank in RB tab, but missing altogether in LF tab. This seems like we should either explicitly fix values in code (i.e hardcode with case_when() or see if tidyr::fill() (grabbing last reported val) is sufficient.

[zacka-cartercenter]

New to do:

• [x] move data cleaning/aggregation into functions and then {targets} pipeline
• [ ] use new {targets} outputs to produce data quality visual rmd/html doc

Once these are complete I will close this issue.