How to make this simple function more efficient?

Question

I have data on wages and about 95% of them are given in hourly format, however some of them are given as an annual salary. So I made a function to convert the annual salaries to hourly, however it takes 1 min 40 sec to run, when my dataset is 43000 rows x 12 columns (which I didnt think would be too big) so I did not think it would take this long.

I am curious if there is a better way to do this than the current function I have created. I am new with dplyr and tidyverse so ideally an answer using those capabilities.

Here is some sample data:

NOC4  Region Region_Name Wage_2012 Wage_2013 Wage_2014   
0011  ER10   National    28.1      65000     NA       
0011  ER1010 Northern    NA        30.5      18       
0011  ER1020 Southern    42.3      72000     22       
0011  ER1030 Eastern     12        NA        45500    
0011  ER1040 Western     8         NA        99000    
0011  ER10   National    NA        65000     NA  

Here is what it should look like after the function:

NOC4  Region Region_Name Wage_2012 Wage_2013 Wage_2014   
0011  ER10   National    28.1      33.33     NA       
0011  ER1010 Northern    NA        30.5      18       
0011  ER1020 Southern    42.3      36.92     22       
0011  ER1030 Eastern     12        NA        23.33    
0011  ER1040 Western     8         NA        50.77    
0011  ER10   National    NA        33.33     NA  

Here is the function:

year_to_hour <- function(dataset, salary, startcol){
  # where "startcol" should be the first column containing the numeric
  # values that you are trying to convert. 
  for(i in startcol:ncol(dataset)){

    for(j in 1:nrow(dataset)){

      if(is.na(dataset[j, i])){

        j = j+1

      }else if(as.numeric(dataset[j, i]) >= as.numeric(salary)){

        dataset[j, i] = dataset[j, i]/1950
      }
      else{

        dataset[j, i] = dataset[j, i]

      }

    }

  }

  return(as_tibble(dataset))

}

converted <- year_to_hour(wage_data_messy, 1000, 4)

Answer 1

R will work much faster if you let it handle the loops under the hood through "vectorized" code.

http://www.noamross.net/blog/2014/4/16/vectorization-in-r--why.html

Here's an approach using dplyr :

library(dplyr)
salary <- 1000
df %>%
  mutate_at(vars(Wage_2012:Wage_2014),          # For these columns...
            ~ . / if_else(. > salary, 1950, 1)) # Divide by 1950 if > salary

Answer 2

Using dplyr I would use mutate_if

salary <- 1000
df %>% mutate_if(is.numeric, ~ifelse(. > salary, ./1950, .))