Case_when in mutate for weighted averages - creating unexpected NAs for some cases that meet a condition but working for all others - conditional-statements

Question

This is my first post! I have a dataset in long form where i have a column called "indicator" with 51 levels and then a column with numeric values and then a series of columns with identifying and weighting variables. I am using a long sequence of case_when conditions inside mutate to weight different levels of indicator by different other columns. The code is mostly working, but there are a few conditions where it is creating NAs for some of the levels of indicator that meet the condition, but not others. So this means it's not a problem in the line of the code, but something in the underlying data and I can't figure it out. I have tried separating out the distinct conditions into separate mutate functions but that didn't help. All help would be so much appreciated!
# Weighted means # Grouping by indicator only - this is where I get errors
df4 <- df3 %>%
group_by(indicator) %>%
mutate(globalmean = case_when(
((grepl("production", mean_weighting) & grepl("Emissions intensity, beef", indicator)) ~ weighted.mean(value, prod_beef, na.rm = TRUE)),
((grepl("production", mean_weighting) & grepl("Emissions intensity, cereals (excl rice)", indicator)) ~ weighted.mean(value, prod_cerealsnorice, na.rm = TRUE)),
((grepl("production", mean_weighting) & grepl("Emissions intensity, chicken", indicator)) ~ weighted.mean(value, prod_chickenmeat, na.rm = TRUE)),
((grepl("production", mean_weighting) & grepl("Emissions intensity, eggs", indicator)) ~ weighted.mean(value, prod_eggs, na.rm = TRUE)),
((grepl("production", mean_weighting) & grepl("Emissions intensity, milk", indicator)) ~ weighted.mean(value, prod_cowmilk, na.rm = TRUE)),
((grepl("production", mean_weighting) & grepl("Emissions intensity, pork", indicator)) ~ weighted.mean(value, prod_pork, na.rm = TRUE)),
((grepl("production", mean_weighting) & grepl("Emissions intensity, rice", indicator)) ~ weighted.mean(value, prod_rice, na.rm = TRUE)),
((grepl("areaharvested", mean_weighting) & grepl("Yield, cereals", indicator)) ~ weighted.mean(value, areaharvested_cereals, na.rm = TRUE)),
((grepl("areaharvested", mean_weighting) & grepl("Yield, citrus", indicator)) ~ weighted.mean(value, areaharvested_citrus, na.rm = TRUE)),
((grepl("areaharvested", mean_weighting) & grepl("Yield, fruit", indicator)) ~ weighted.mean(value, areaharvested_fruit, na.rm = TRUE)),
((grepl("areaharvested", mean_weighting) & grepl("Yield, pulses", indicator)) ~ weighted.mean(value, areaharvested_pulses, na.rm = TRUE)),
((grepl("areaharvested", mean_weighting) & grepl("Yield, roots and tubers", indicator)) ~ weighted.mean(value, areaharvested_roottuber, na.rm = TRUE)),
((grepl("areaharvested", mean_weighting) & grepl("Yield, treenuts", indicator)) ~ weighted.mean(value, areaharvested_treenuts, na.rm = TRUE)),
((grepl("areaharvested", mean_weighting) & grepl("Yield, vegetables", indicator)) ~ weighted.mean(value, areaharvested_vegetables, na.rm = TRUE)),
((grepl("producinganimals", mean_weighting) & grepl("Yield, beef", indicator)) ~ weighted.mean(value, producinganimals_beef, na.rm = TRUE)),
((grepl("producinganimals", mean_weighting) & grepl("Yield, chicken", indicator)) ~ weighted.mean(value, producinganimals_chickenmeat, na.rm = TRUE)),
((grepl("producinganimals", mean_weighting) & grepl("Yield, milk", indicator)) ~ weighted.mean(value, producinganimals_cowmilk, na.rm = TRUE)),
((grepl("producinganimals", mean_weighting) & grepl("Yield, eggs", indicator)) ~ weighted.mean(value, producinganimals_eggs, na.rm = TRUE)),
((grepl("producinganimals", mean_weighting) & grepl("Yield, pork", indicator)) ~ weighted.mean(value, producinganimals_pork, na.rm = TRUE)),
(grepl("pop_u", mean_weighting) ~ weighted.mean(value, pop_u, na.rm = TRUE)),
(grepl("GDP", mean_weighting) ~ weighted.mean(value, GDP, na.rm = TRUE)), # one of two indicator levels meeting this condition works, the other results in NA
(grepl("unweighted", mean_weighting) ~ mean(value, na.rm = TRUE)),
(grepl("cropland", mean_weighting) ~ weighted.mean(value, croplandcov, na.rm = TRUE)),
(grepl("totalpop", mean_weighting) ~ weighted.mean(value, totalpop, na.rm = TRUE)), # Almost all of the indicator levels meeting this condition work, 3 result in NA
(grepl("landarea", mean_weighting) ~ weighted.mean(value, landarea, na.rm = TRUE)),
TRUE ~ mean(value)
)
) %>%
## Tried this manual overwriting the NA indicator levels as a workaround but it doesn't fix the problem:
mutate(globalmean = case_when(
indicator == "safeh20" ~ weighted.mean(value, totalpop, na.rm = TRUE),
indicator == "govteffect" ~ weighted.mean(value, totalpop, na.rm = TRUE),
indicator == "healthtax" ~ weighted.mean(value, totalpop, na.rm = TRUE),
indicator == "aginGDP" ~ weighted.mean(value, GDP, na.rm = TRUE),
indicator == "functionalintegrity" ~ weighted.mean(value, croplandcov, na.rm = TRUE),
indicator == "pesticides" ~ weighted.mean(value, croplandcov, na.rm = TRUE),
indicator == "sustNO2mgmt" ~ weighted.mean(value, croplandcov, na.rm = TRUE),
TRUE ~ globalmean)) %>%
ungroup()
df4 <- df4 %>% group_by(indicator) %>%
mutate(globalmedian = median(value, na.rm = TRUE),
globalmin = min(value, na.rm = TRUE),
globalmax = max(value, na.rm = TRUE),
globalp25 = quantile(value, c(.25)),
globalp75 = quantile(value, c(.75))) %>%
ungroup() %>% as.data.frame()
**# Weighted means grouping by region and indicator - otherwise all the same code and it works perfectly.**
df5 <- df4 %>% group_by(fsci_regions, indicator) %>%
mutate(regionmean = case_when(
((grepl("production", mean_weighting) & grepl("Emissions intensity, beef", indicator)) ~ weighted.mean(value, prod_beef, na.rm = TRUE)),
((grepl("production", mean_weighting) & grepl("Emissions intensity, cereals (excl rice)", indicator)) ~ weighted.mean(value, prod_cerealsnorice, na.rm = TRUE)),
((grepl("production", mean_weighting) & grepl("Emissions intensity, chicken", indicator)) ~ weighted.mean(value, prod_chickenmeat, na.rm = TRUE)),
((grepl("production", mean_weighting) & grepl("Emissions intensity, eggs", indicator)) ~ weighted.mean(value, prod_eggs, na.rm = TRUE)),
((grepl("production", mean_weighting) & grepl("Emissions intensity, milk", indicator)) ~ weighted.mean(value, prod_cowmilk, na.rm = TRUE)),
((grepl("production", mean_weighting) & grepl("Emissions intensity, pork", indicator)) ~ weighted.mean(value, prod_pork, na.rm = TRUE)),
((grepl("production", mean_weighting) & grepl("Emissions intensity, rice", indicator)) ~ weighted.mean(value, prod_rice, na.rm = TRUE)),
((grepl("areaharvested", mean_weighting) & grepl("Yield, cereals", indicator)) ~ weighted.mean(value, areaharvested_cereals, na.rm = TRUE)),
((grepl("areaharvested", mean_weighting) & grepl("Yield, citrus", indicator)) ~ weighted.mean(value, areaharvested_citrus, na.rm = TRUE)),
((grepl("areaharvested", mean_weighting) & grepl("Yield, fruit", indicator)) ~ weighted.mean(value, areaharvested_fruit, na.rm = TRUE)),
((grepl("areaharvested", mean_weighting) & grepl("Yield, pulses", indicator)) ~ weighted.mean(value, areaharvested_pulses, na.rm = TRUE)),
((grepl("areaharvested", mean_weighting) & grepl("Yield, roots and tubers", indicator)) ~ weighted.mean(value, areaharvested_roottuber, na.rm = TRUE)),
((grepl("areaharvested", mean_weighting) & grepl("Yield, treenuts", indicator)) ~ weighted.mean(value, areaharvested_treenuts, na.rm = TRUE)),
((grepl("areaharvested", mean_weighting) & grepl("Yield, vegetables", indicator)) ~ weighted.mean(value, areaharvested_vegetables, na.rm = TRUE)),
((grepl("producinganimals", mean_weighting) & grepl("Yield, beef", indicator)) ~ weighted.mean(value, producinganimals_beef, na.rm = TRUE)),
((grepl("producinganimals", mean_weighting) & grepl("Yield, chicken", indicator)) ~ weighted.mean(value, producinganimals_chickenmeat, na.rm = TRUE)),
((grepl("producinganimals", mean_weighting) & grepl("Yield, milk", indicator)) ~ weighted.mean(value, producinganimals_cowmilk, na.rm = TRUE)),
((grepl("producinganimals", mean_weighting) & grepl("Yield, eggs", indicator)) ~ weighted.mean(value, producinganimals_eggs, na.rm = TRUE)),
((grepl("producinganimals", mean_weighting) & grepl("Yield, pork", indicator)) ~ weighted.mean(value, producinganimals_pork, na.rm = TRUE)),
((grepl("unweighted", mean_weighting)) ~ mean(value, na.rm = TRUE)),
(grepl("totalpop", mean_weighting) ~ weighted.mean(value, totalpop, na.rm = TRUE)),
(grepl("landarea", mean_weighting) ~ weighted.mean(value, landarea, na.rm = TRUE)),
(grepl("cropland", mean_weighting) ~ weighted.mean(value, croplandcov, na.rm = TRUE)),
(grepl("pop_u", mean_weighting) ~ weighted.mean(value, pop_u, na.rm = TRUE)),
(TRUE ~ mean(value, na.rm = TRUE)))) %>% ungroup() %>%
group_by(fsci_regions, short_label, globalmean, globalmedian, globalmin, globalmax, globalp25,
globalp75, theme, direction) %>%
summarise(regionmean = first(regionmean))
I moved things around in the case_when ordering going from most specific to most general, before that even more lines had errors but now it's down to the 6 described above. I also tried coding more specifically the 6 that aren't working after the main chunk of code but that also didn't work (see annotation in code).
All insights would be so helpful!
Many thanks,
Kate