0

I have this dataframe:

x = data.frame("city" = c("Madrid","Berlin","Florence","Madrid"), 
"Visits" = c(100,200,80,38), "Date" = c(1,2,3,4))

From that sample, I would like to obtain 3 dataframe (one for each city) with all the values containing that city and named as that city without the column city, so it would result in the following:

Madrid = data.frame("Visits" = c(100,38), "Date" = c(1,4))
Berlin = data.frame("Visits" = c(200), "Date" = c(2)
Florence = data.frame("Visits" = c(80), "Date" = c(3))

I asked the same question in pandas: How to create dataframes iterating over a set? but I cannot find something similar to dictionary comprehension.

I have managed to get the unique list and to get the values of a city:

cities = unique(select(x,city))
for (i in cities){
  dplyr::filter(x,city == i)}

I know that loops should not be used in R and apply is more efficient, but I dont know how could I do that using apply. I am open to other data structures (and not dataframes for each city) as long as I am able to access them easily to pass other functions (auto.arima for instance). Thank you very much

Javier Lopez Tomas
  • 2,072
  • 3
  • 19
  • 41
  • `split(x, x$city)` gives a named list of the 3 data frames. – Aurèle Nov 23 '18 at 12:17
  • There is no strict rule that loops shouldn't be used - speed depends on many other factors and also is often should not be the primary concern. – s_baldur Nov 23 '18 at 12:29
  • Split doesn't create new dataframes, although I have read that it should. I get no error, I just get the print in the console – Javier Lopez Tomas Nov 23 '18 at 13:40
  • @JavierLópezTomás You could do `list_dfs <- split(x, x$city)` to assign the result of split to a variable, and then access an individual data frame with e.g. `list_dfs$Madrid` – Aurèle Nov 27 '18 at 08:56

1 Answers1

1

You could do 

list_dfs <- split(x, x$city)

to assign the result of split to a variable, and then access an individual data frame with e.g. list_dfs$Madrid.

If you're more comfortable having the data frames as individual variables in your Global environment (see https://stackoverflow.com/a/9726880/6197649), you could do 

list2env(split(x, x$city), envir = .GlobalEnv)

but that is not the recommended "R way of doing things". It's usually better to have similarly structured objects in a single list.

Aurèle
  • 12,545
  • 1
  • 31
  • 49