Updated coerce_fct_na_explicit

This function was previously returning NULL when the factor had no missing values. I've modified it to return the input if it has no missing values.

Description

The gg_miss_fct function wasn't working for a dataset I was using. I investigated and found the case being that the code attempts to reencode all NAs in the fct argument using the in-built function coerce_fct_na_explicit. However, coerce_fct_no_explicit has no default return value, so if your factor fails the If check, the function returns NULL and thus gg_miss_fct fails later on.

I've resolved this by simply adding an Else statement to provide a default return value, being the input factor.

Example

Example below that fails using the current Master branch.

devtools::install_github("njtierney/naniar")
library(naniar)

# Works as expected
gg_miss_fct(x = riskfactors, fct = marital)

# have 245 observations of 34 values
dim(riskfactors)
# Have 1 missing value in marital column
sum(is.na(riskfactors$marital))
# Removing this row gives us 244 x 34
riskfactors_nomiss <- riskfactors[!is.na(riskfactors$marital), ]
dim(riskfactors_nomiss)
# And now the function doesn't work
gg_miss_fct(x = riskfactors_nomiss, fct = marital)

Running this example under my fork works:

# Install my fork
devtools::install_github("stulacy/naniar")
library(naniar)

# Default method still works
gg_miss_fct(x = riskfactors, fct = marital)

# And now the function works correctly when no missing in marital column
riskfactors_nomiss <- riskfactors[!is.na(riskfactors$marital), ]
gg_miss_fct(x = riskfactors_nomiss, fct = marital)

Tests

I haven't run any automated testing. Let me know if this will be needed.

NEWS + DESCRIPTION

This is a bug fix so I don't believe the DESCRIPTION or NEWS.md need updating.

njtierney / naniar