There is a typo error about the SMART criteria cut-off point for skewness/kurtosis in "Article called Distributions of variables and indices".
The cut-off point for "acceptable" was mentioned as "≥ 0.6 and < 0.6". Actually, it should be ">=0.4 and <0.6". You can see in the table name "The range of absolute values of skewness and kurtosis statistics that are applied by SMART (2015)" under sub-topic called "Skew and kurtosis".

I found a bug in the outliersUV() function in the NiPN data quality toolkit that can prevent calls such as:

svy[outliersUV(svy$muac), ]

from returning the correct set of records when there are NA values in the variable being tested.

I found this when using the toolkit with a dataset from MSF that had NA for all anthropometry when oedema was present. This may be a common case.

Here is an example of the problem using the example dataset in the NiPN toolkit:

svy <- read.table("rl.ex01.csv", header = TRUE, sep = ",")
head(svy)
## Test function
svy[outliersUV(svy$muac), ]

This gives:

Univariate outliers : Lower fence = 98, Upper fence = 178

Note that the values in the muac column are all outside of the fences. This is correct.

Adding a missing value:

## Add a missing value
svy$muac[1] <- NA
head(svy)
## Test function
svy[outliersUV(svy$muac), ]

gives us:

Univariate outliers : Lower fence = 98, Upper fence = 178

The listed records do not contain outliers. This is wrong.

Add sex ratio vignette to articles

• #60
• #61
• #62
• #63
• #64
• #65
• #66
• #67
• #68
• #69
• #70
• #71
• #74

I installed the package from CRAN and followed the sample code provided in the articles' session. But, those build-in datasets (example: rl.ex01) were not able to open. Below was the error message I got after running this comment "svy <- rl.ex01".
Error: object 'rl.ex01' not found

current sticker looks very dull.

I am writing to report an issue I detected with ageRatioTest() in the nipnTK R package. The issues is that the function is returning wrong p-values. I tested this with different datasets and compared the results with those returned in ENA for SMART software. Let me showcase the issue here and how I tried to identify the source of this error in your code. Lets say I take a sample data and I do:

> ageRatioTest(as.numeric(!is.na(input_data$age)), ratio = 0.85)$p

I get this (after transforming scientific notation into conventional):

[1] 0

When I aggregate at a district, for instance I get this: as you see, exact same results returned in an all locations.

## Check Age ratio ----

data |> select(district, `Age ratio (p)`) |> head()

# A tibble: 6 × 2
  district            `Age ratio (p)`
  <chr>               <chr>         
1 Addun Pastoral LZ   <0.001        
2 Bay Agropastoral LZ <0.001        
3 Beletweyne district <0.001        
4 Coastal Deah        <0.001        
5 East Golis          <0.001        
6 Elbarde             <0.001 

When I return the observed ratio, I get the this:

> ageRatioTest(as.numeric(!is.na(input_data$age)), ratio = 0.85)$observedR
[1] 12.8738
```'
and the observed proportion, I get this: 
```r
> ageRatioTest(as.numeric(!is.na(input_data$age)), ratio = 0.85)$observedP
[1] 0.9279217

I inspected the code by running AgeRatioTest in console:

function (x, ratio = 0.85) 
{
    g <- recode(x, "6:29=1; 30:59=2")
    expectedP <- ratio/(ratio + 1)
    observedP <- sum(g == 1)/sum(table(g))
    observedR <- observedP/(1 - observedP)
    X2 <- prop.test(sum(g == 1), sum(table(g)), p = expectedP)
    result <- list(expectedR = ratio, expectedP = expectedP, 
        observedR = observedR, observedP = observedP, X2 = X2$statistic, 
        df = X2$parameter, p = X2$p.value)
    class(result) <- "ageRatioTest"
    return(result)
}

I tried to address this by writing a different function using the guidance in the nipnTK website:

# Function to run age ratio test -------------------------------------------------------------------
 
age_ratio_test <- function(age, .expectedR = 0.85) {
  
  x <- ifelse(age < 30, 1, 2)
  eprop <- .expectedR / (.expectedR + 1)
  ratio <- sum(x[x == 1], na.rm = TRUE) / sum(x[x == 2], na.rm = TRUE)
  prop <- sum(x[x == 1], na.rm = TRUE) / sum(table(x))
  test <- prop.test(sum(x[x == 1], na.rm = TRUE), sum(table(x)), p = eprop)
  
  return(
    list(
      p = test$p.value,
      observedR = ratio,
      observedP = prop
    )
  )
}

When applying this function to the same data, it returns a different statistics than that in nipnTK::ageRatioTest and the statistics are nearly equal those from the ENA plausibility (just some differences in decimal points). If I run the same test as above I get this:

> age_ratio_test(input_data$age, .expectedR = 0.85)$p
[1] 0.06905227
 
## Check Age ratio ----
 
# A tibble: 6 × 2
  district            `Age ratio (p)`
  <chr>               <chr>          
1 Addun Pastoral LZ   0.011          
2 Bay Agropastoral LZ 0.035          
3 Beletweyne district 0.369          
4 Coastal Deah        0.265          
5 East Golis          0.713          
6 Elbarde             0.300 

When I looked into the differences I spotted two things:

Not use of na.rm = TRUE

in my function (age_ratio_test), I use the arguments na.rm = TRUE in the objects "ratio", "prop" and "test", while in yours you don't remove NA's. With this, in the script, even when trying to control for NA's using this approach (as an example) ageRatioTest(as.numeric(!is.na(input_data$age)), ratio = 0.85)$p, I still doesn't work. I tested my function with this same approach and it returned exactly what your function is returning - incorrect numbers:

> age_ratio_test(as.numeric(!is.na(input_data$age)), .expectedR = 0.85)$p
[1] 0

Use of recode(): you are using recode(). With this, even when adding na.rm = TRUE in the sum() operations, it does not work. It throws this error:

## Function ----
ART <- function (x, ratio = 0.85) 
{
  g <- recode(x, "6:29=1; 30:59=2")
  expectedP <- ratio/(ratio + 1)
  observedP <- sum(g == 1, na.rm = TRUE)/sum(table(g))
  observedR <- observedP/(1 - observedP)
  X2 <- prop.test(sum(g == 1, na.rm = TRUE), sum(table(g)), p = expectedP)
  result <- list(expectedR = ratio, expectedP = expectedP, 
                 observedR = observedR, observedP = observedP, X2 = X2$statistic, 
                 df = X2$parameter, p = X2$p.value)
  class(result) <- "ageRatioTest"
  return(result)
}
 
## Implementation ----
> ART(input_data$age, ratio = 0.85)$p

Error in prop.test(sum(g == 1, na.rm = TRUE), sum(table(g)), p = expectedP) : 
  elements of 'n' must be positive
In addition: Warning message:
Unreplaced values treated as NA as `.x` is not compatible.
Please specify replacements exhaustively or supply `.default`. 

But changing that to ifelse() makes it works :

## Function ----
ART <- function (x, ratio = 0.85) 
{
  g <- ifelse(x < 30, 1, 2)
  expectedP <- ratio/(ratio + 1)
  observedP <- sum(g == 1, na.rm = TRUE)/sum(table(g))
  observedR <- observedP/(1 - observedP)
  X2 <- prop.test(sum(g == 1, na.rm = TRUE), sum(table(g)), p = expectedP)
  result <- list(expectedR = ratio, expectedP = expectedP, 
                 observedR = observedR, observedP = observedP, X2 = X2$statistic, 
                 df = X2$parameter, p = X2$p.value)
  class(result) <- "ageRatioTest"
  return(result)
}
 
## Implementation ----
> ART(input_data$age, ratio = 0.85)$p
[1] 0.06905227

I hope you will find this useful.

I noticed that current DESCRIPTION doesn't have a testthat config.

Thanks,

\dontrun{} should only be used if the example really cannot be executed
(e.g. because of missing additional software, missing API keys, ...) by
the user. That's why wrapping examples in \dontrun{} adds the comment
("# Not run:") as a warning for the user.
Does not seem necessary.
Please unwrap the example if it is executable in < 5 sec, or replace
\dontrun{} with \donttest{}.

Please make sure that you do not change the user's options, par or
working directory. If you really have to do so within functions, please
ensure with an immediate call of on.exit() that the settings are reset
when the function is exited. e.g.:
...
oldpar <- par(no.readonly = TRUE) # code line i
on.exit(par(oldpar)) # code line i + 1
...
par(mfrow=c(2,2)) # somewhere after
...
e.g.: ageChildren.R
If you're not familiar with the function, please check ?on.exit. This
function makes it possible to restore options before exiting a function
even if the function breaks. Therefore it needs to be called immediately
after the option change within a function.

Please do not set a seed to a specific number within a function. e.g.:
greensIndex.R

this seems to be because of how the badges in README are identified. It should show this in README:

<!-- badges: start -->
<!-- badges: end -->

for pkgdown to recognise the badge chunk

see https://github.com/walkerke/idbr

see https://github.com/tomaszaba/muac-quality-checks/issues/21