predict_race() occasionally returns NaN

[mrhumanzee]

I am imputing the race of individuals in a relatively large dataset (100,000+ observations) using first name, surname, and county. For around 40 observations, the function predict_race() returned NaN. I don't understand the implementation of wru well enough to understand why this occurred, but I have provided code below that allows the problem to be reproduced. The 'John Smith' observation was included to show that this problem does not necessarily affect every observation in the data frame. Also, changing the combination of first name and surname for a given observation may affect whether NaN is returned.

df1= data.frame(first=c("T-Yvette", "Loddy", "Bitta", "Rayka", "John"),
           surname=c("Soutiere", "Tolzmann", "Mostofi", "Belova", "Smith"),
           state=c("AK", "MN", "NY", "NY", "NY"),
           county=c("110", "053", "047", "061", "061"))

df1=predict_race(voter.file = df1, census.geo = "county",
                                   census.key = yourkey,
                                   impute.missing = F, names.to.use="surname, first")
df1


df2= data.frame(first=c("T-Yvette", "Loddy", "Bitta", "Rayka", "John"),
                surname=c("Smith", "Tolzmann", "Mostofi", "Belova", "Smith"),
                state=c("AK", "MN", "NY", "NY", "NY"),
                county=c("110", "053", "047", "061", "061"))

df2=predict_race(voter.file = df2, census.geo = "county",
                 census.key = yourkey,
                 impute.missing = F, names.to.use="surname, first")
df2