Anesrake error: "Error in targetvec - dat: non-numeric argument to binary operator"

[mrlazerbeam]

First time posting - and very new user of R Studio here. I've made a reprex below of an error i keep getting.

I am trying to rake survey data to make it representative of UK population data. I have a set of responses and have found distribution data to make it representative. The raking procedure I am using is anesrake , similar to this: R anesrake error: “Error in x + weights: non-numeric argument for binary operator” Ask Question.

I have 4 columns of data with 30 rows: caseid, gender (M,F); age (18-24, 25-34, 35-44, 45-54, 55-64, 65-74, 75) and Answer (1,0). My data is saved on shared drive, havent included in the reprex.

I have loaded target distributions for age and gender, but continue to get the following error: "Error in targetvec - dat: non-numeric argument to binary operator"

I have attempted to include a reprex below.

Any assistance would really help!

Thanks

Ed

library(readxl)
library(weights)
library(anesrake)
library(plyr)
library(dplyr)
library(reshape2)
library(reprex)
rtest <- read_excel(data is not shared)
View(rtest) 
# UK Census 2019
gender  <- c(.51,.49)
age <- c(0.1,.17,.163,0.166,0.161,0.128,0.112)
# definitions of target list
targets <- list(gender, age)
#important to use same variable names of the dataset
names(targets) <- c("gender", "age")
#label levels of targets#
names(targets$gender) <- levels(rtest$gender)
names(targets$age) <- levels(rtest$age)
# change table type
rtest <- as.data.frame(rtest)
class(rtest)
#> [1] "data.frame"
#measure variance in population vs sample
anesrakefinder(targets, rtest, choosemethod = "total")
#>    gender       age 
#> 0.1533333 0.4053333
#convert to factor (unsure if this works)
rtest$gender <- as.factor(rtest$gender) 
rtest$age <- as.factor(rtest$age)
#raking procedure
raking <- anesrake(targets, rtest, rtest$caseid, cap = 5,choosemethod = "total", type = "pctlim",pctlim = 0.05)
#> Error in targetvec - dat: non-numeric argument to binary operator

^{Created on 2021-02-05 by the reprex package (v1.0.0)}

[technocrat]

@mrlazerbeam

rtest <- read_excel(data is not shared)

No representative data means the issue has to be reverse engineered, which is a deterrent to helpful answers. It doesn't have to be all the data or even your data, just enough data in the same form to reproduce the issue.

See the FAQ: How to do a minimal reproducible example reprex for beginners

[mrlazerbeam]

Thanks for the guidance - I have now included sample data, and have been able to progress with the issue further by renaming Age categories as 1 - 7, and Gender as 1 - 2

• I now get the following NA issue in the tables that are produced. Any thoughts?
• I'd also like to not have to rename categorical variables into numbers for this to work. Not sure how?

library(readxl)
library(weights)
library(anesrake)
library(plyr)
library(dplyr)
library(reshape2)
library(reprex)
library(datapasta)
#Example data#
head(rtest2,10)
#> # A tibble: 10 x 4
#>    caseid gender   age Answer
#>     <dbl>  <dbl> <dbl>  <dbl>
#>  1      1      1     1      1
#>  2      2      2     2      0
#>  3      3      1     1      1
#>  4      4      1     1      1
#>  5      5      1     4      1
#>  6      6      1     1      0
#>  7      7      2     2      0
#>  8      8      2     1      1
#>  9      9      1     1      0
#> 10     10      1     4      0
datapasta::df_paste(head(rtest2,10))
#> data.frame(
#>       caseid = c(1, 2, 3, 4, 5, 6, 7, 8, 9, 10),
#>       gender = c(1, 2, 1, 1, 1, 1, 2, 2, 1, 1),
#>          age = c(1, 2, 1, 1, 4, 1, 2, 1, 1, 4),
#>       Answer = c(1, 0, 1, 1, 1, 0, 0, 1, 0, 0)
#> )
data.frame(
      caseid = c(1, 2, 3, 4, 5, 6, 7, 8, 9, 10),
      gender = c(1, 2, 1, 1, 1, 1, 2, 2, 1, 1),
         age = c(1, 2, 1, 1, 4, 1, 2, 1, 1, 4),
      Answer = c(1, 0, 1, 1, 1, 0, 0, 1, 0, 0)
                              )#Proportions of rtest2 survey#
#>    caseid gender age Answer
#> 1       1      1   1      1
#> 2       2      2   2      0
#> 3       3      1   1      1
#> 4       4      1   1      1
#> 5       5      1   4      1
#> 6       6      1   1      0
#> 7       7      2   2      0
#> 8       8      2   1      1
#> 9       9      1   1      0
#> 10     10      1   4      0
wpct(rtest2$gender)
#>    1    2 
#> 0.65 0.35
wpct(rtest2$age)
#>     1     2     3     4     5     6     7 
#> 0.274 0.246 0.114 0.096 0.076 0.102 0.092
wpct(rtest2$Answer)
#>     0     1 
#> 0.516 0.484
# UK Census 2019
gender  <- c(.51,.49)
age <- c(0.1,.17,.163,0.166,0.161,0.128,0.112)
# definitions of target list
targets <- list(gender, age)
#important to use same variable names of the dataset
names(targets) <- c("gender", "age")
#id variable
rtest2$caseid <- 1:length(rtest2$gender)
#label levels of targets#
names(targets$gender) <- levels(rtest2$gender)
names(targets$age) <- levels(rtest2$age)
# change table type
rtest2 <- as.data.frame(rtest2)
class(rtest2)
#> [1] "data.frame"
#measure variance in population vs sample
anesrakefinder(targets, rtest2, choosemethod = "total")
#> gender    age 
#>   0.28   0.50
#raking procedure
raking <- anesrake(targets, rtest2, caseid = rtest2$caseid, verbose = FALSE, cap = 5,choosemethod = "total", type = "pctlim",pctlim = 0.05, nlim = 5, iterate = TRUE, force1 = TRUE)
#> [1] "Raking converged in 11 iterations"
raking_summary <- summary(raking)
rtest2$weight <- raking$weightvec
#to find the unique weights#
rtest2 %>% select(gender, age) %>% unique()
#>    gender age
#> 1       1   1
#> 2       2   2
#> 5       1   4
#> 8       2   1
#> 12      1   2
#> 14      2   6
#> 15      1   7
#> 16      2   7
#> 17      1   3
#> 22      1   6
#> 35      1   5
#> 39      2   5
#> 40      2   3
#> 62      2   4
wpct(rtest2$Answer)
#>     0     1 
#> 0.516 0.484
wpct(rtest2$Answer, rtest2$weight)
#>         0         1 
#> 0.5120287 0.4879713
#Raking summary statistics do not work "NA"#
raking_summary$raking.variables
#> [1] "gender" "age"
raking_summary$gender
#>       Target Unweighted N Unweighted % Wtd N Wtd % Change in % Resid. Disc.
#> <NA>    0.51           NA           NA    NA    NA          NA           NA
#> <NA>    0.49           NA           NA    NA    NA          NA           NA
#> Total   1.00            0            0     0     0           0            0
#>       Orig. Disc.
#> <NA>           NA
#> <NA>           NA
#> Total           0
raking_summary$age
#>       Target Unweighted N Unweighted % Wtd N Wtd % Change in % Resid. Disc.
#> <NA>   0.100           NA           NA    NA    NA          NA           NA
#> <NA>   0.170           NA           NA    NA    NA          NA           NA
#> <NA>   0.163           NA           NA    NA    NA          NA           NA
#> <NA>   0.166           NA           NA    NA    NA          NA           NA
#> <NA>   0.161           NA           NA    NA    NA          NA           NA
#> <NA>   0.128           NA           NA    NA    NA          NA           NA
#> <NA>   0.112           NA           NA    NA    NA          NA           NA
#> Total  1.000            0            0     0     0           0            0
#>       Orig. Disc.
#> <NA>           NA
#> <NA>           NA
#> <NA>           NA
#> <NA>           NA
#> <NA>           NA
#> <NA>           NA
#> <NA>           NA
#> Total           0

^{Created on 2021-02-08 by the reprex package (v1.0.0)}

[technocrat]

Sorry for the delay. I'm going to have to study this package. DM me if you don't hear in a reasonable time.