R modifying a data.frame in a for loop

Question

coming from Java I am trying to learn R (and some statistics) I am trying to reproduce the following table

from Jonathan Gillard: A First Course in Statistical Inference This table shows the possible results of two draws with replacement from a piggybank piggybank <- c(5, 10, 10, 20, 50, 50)

With the following code I encounter some unexpected (at least for a Java programmer) behavior.

The first seven lines basically give me what I want, except for (5, 10) and (10, 5) should be aggregated in one category. I have thought of using a set for this, but the sets library seems to mess up the first 7 lines.

    library(dplyr)
rm(list=ls())
piggybank <- c(5, 10, 10, 20, 50, 50)
draws <- expand.grid(d1=piggybank, d2=piggybank)
draws <- draws %>% rowwise() %>% mutate(sum=sum(c(d1,d2)), var=var(c(d1,d2)), mean=mean(c(d1,d2)))
draws <- draws %>% group_by(d1, d2, var, mean, sum) %>% summarise(n=n())
draws <- draws %>% ungroup() %>% mutate(P=n/sum(n))
nr <- nrow(draws)
aggdraws <- data.frame(x1x2=character(0), var=numeric(0), mean=numeric(0), sum=numeric(0), n=numeric(0))
str(aggdraws)
local(
  for (i in 1:nr) {
    newrow <<- data.frame(x1x2=character(1), var=numeric(1), mean=numeric(1), sum=numeric(1), n=numeric(10))
    newrow$n <- draws[i, ]$n
    newrow$var <- draws[i, ]$var
    newrow$mean <- draws[i, ]$mean
    newrow$sum <- draws[i, ]$mean
    
    newrow$x1x2 <- paste(min(draws[i, ]$d1, draws[i, ]$d2), max(draws[i, ]$d1, draws[i, ]$d2))
    
    #print(aggdraws)
    if (nrow(aggdraws) > 0) {
      for(j in 1:nrow(aggdraws)) {
        print(paste(aggdraws[j,]$x1x2, newrow$x1x2))
        if(aggdraws[j,]$x1x2 == newrow$x1x2) {
          aggdraws[j,]$n <- aggdraws[j,]$n +newrow$n
        } else {
          aggdraws[nrow(aggdraws)+1, ] <- newrow
        }
      }
    } else {
      aggdraws[nrow(aggdraws)+1, ] <- newrow
    }
  }
)

newrow <<- data.frame(x1x2=character(1), var=numeric(1), mean=numeric(1), sum=numeric(1), n=numeric(10)) creates a data.frame with 10 observations of 5 variables. Why? I need a dataframe with 1 observation.

newrow seems not to be local to the for loop, it is filled with a row in each iteration. I need a new instance in every iteration

Probably because of this behavior if(aggdraws[j,]$x1x2 == newrow$x1x2) never evaluates to TRUE

Any help would be greatly appreciated. Is there a good book or other source which points out the pitfalls of R for a programmer coming from Java or another object-oriented language?

Thanks,

Hans

Hi there. I'm not really sure about the exact problem. A good start with R is the book R for Data Science, for more advanced readingn Advanced R. — Martin Gal
– Martin Gal, Commented May 22, 2021 at 16:54
newrow <<- data.frame(x1x2=character(1), var=numeric(1), mean=numeric(1), sum=numeric(1), n=numeric(10)) --> n = numeric(10) creates a vector of length 10. Replace it with numeric(1) for a vector with length 1. Probably that's the source of your strange behaviour. — Martin Gal
– Martin Gal, Commented May 22, 2021 at 17:16

Martin Gal · Accepted Answer · 2021-05-22 17:18:25Z

That's not an direct answer to your question. I took a look at your code and optimized the first part for dplyr:

draws <- expand.grid(d1=piggybank, d2=piggybank) %>% 
  rowwise() %>%
  mutate(d1_new = min(d1, d2),
         d2_new = max(d1, d2)) %>%
  select(d1 = d1_new, d2 = d2_new) %>%
  mutate(sum = sum(d1, d2), 
         var = var(c(d1, d2)), 
         mean = mean(d1, d2))%>% 
  group_by(d1, d2, var, mean, sum) %>% 
  summarise(n = n(), .groups="drop") %>%
  mutate(P = n/sum(n))

returns

# A tibble: 10 x 7
      d1    d2    var  mean   sum     n      P
   <dbl> <dbl>  <dbl> <dbl> <dbl> <int>  <dbl>
 1     5     5    0       5    10     1 0.0278
 2     5    10   12.5     5    15     4 0.111 
 3     5    20  112.      5    25     2 0.0556
 4     5    50 1012.      5    55     4 0.111 
 5    10    10    0      10    20     4 0.111 
 6    10    20   50      10    30     4 0.111 
 7    10    50  800      10    60     8 0.222 
 8    20    20    0      20    40     1 0.0278
 9    20    50  450      20    70     4 0.111 
10    50    50    0      50   100     4 0.111

which is pretty much your table from Jonathan Gillard: A First Course in Statistical Inference.

Collectives™ on Stack Overflow

R modifying a data.frame in a for loop

1 Answer 1

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Your Answer

Sign up or log in

Post as a guest

Related