Ndtyjky

I have two dataframes, one called "students.short", generate by:

students.short <- data.frame(shoesize=c(38,39,38,38,39,38,37,36),

 population=c("kuopio","kuopio","kuopio","tampere",

 "tampere","tampere","tampere","tampere"))



students.short



  shoesize population

1       38     kuopio

2       39     kuopio

3       38     kuopio

4       38     kuopio

5       39    tampere

6       38    tampere

7       37    tampere

8       36    tampere

and the other called "students.tall":

students.tall <- data.frame(shoesize=c(44,42,43,43,42,44,43,43),

 population=c("kuopio","kuopio","kuopio","kuopio",

 "tampere","tampere","tampere","tampere"))



students.tall



  shoesize population

1       44     kuopio

2       42     kuopio

3       43     kuopio

4       43     kuopio

5       42    tampere

6       44    tampere

7       43    tampere

8       43    tampere

and I need to create a crosstab between the population (kuopio or tampere) and the means of the shoesize of each dataframes like

                       kuopio   tampere



studenst.short          38.3       37.6



studenst.tall             43         43

I can't find a clean or easy way to do that, any idea or any help, please?

In one go, using data.table

• first, create a named list of the data.tables (using setDT() )


• then, bind the lists together (using rbindlist(), using the names as an id (idcol = TRUE).


• last, dcast to wide format, summarising with mean of the value.var;
  shoesize
  

code

library( data.table )



dcast( rbindlist( list( students.short = setDT( students.short ), 

                        students.tall = setDT( students.tall ) ),

                  idcol = TRUE ),

       .id ~ population, 

       value.var = "shoesize", 

       fun = mean )



#               .id   kuopio tampere

# 1: students.short 38.33333    37.6

# 2:  students.tall 43.00000    43.0

Here is a dplyr driven answer. We basically bind the two data frames first using the .id argument to differentiate between the data frames. We then group_by the .id and population and calculate the mean, i.e.

library(dplyr)



bind_rows(df1, df2, .id = 'group') %>% 

       group_by(group, population) %>% 

       summarise(new = mean(shoesize))

which gives,

# A tibble: 4 x 3

# Groups:   group [?]

  group population   new

  <chr> <fct>      <dbl>

1 1     kuopio      38.3

2 1     tampere     37.6

3 2     kuopio      43  

4 2     tampere     43

Combine your data frames using rbind() first:

df <- rbind(studnets.short, students.tall)

df$height_cat <- rep(c("short", "tall"), # create categorical height variable

   c(nrow(students.short), nrow(students.tall)))

Then use tapply(). For this mock data frame, it works like this:

df <- data.frame(size = round(rnorm(30, 39, 2)),

                 pop = sample(c("kuopio", "tampere"), 30, replace = T),

                 height = sample(c("short", "tall"), 30, replace = T))

tapply(df$size, INDEX = df[c(3, 2)], mean, na.rm=T)

# df[c(3, 2)] refers to height and pop columns of df respectively



       pop

height  kuopio  tampere

  short     39 39.57143

  tall      41 39.22222