passing strings as arguments in ddply

Question

set.seed(1)
df <-data.frame(category=rep(LETTERS[1:5],each=10),superregion=sample(c("EMEA","LATAM","AMER","APAC"),100,replace=T),country=sample(c("Country1","Country2","Country3","Country4","Country5","Country6","Country7","Country8"),100,replace=T),market=sample(c("Market1","Market2","Market3","Market4","Market5","Market6","Market7","Market8","Market9","Market10","Market11","Market12"),100,replace=T),hospitalID=sample(seq(1,50,1),100,replace=T),IndicatorFlag=sample(seq(0,1,1),100,replace=T))

I'm trying to create a SummaryTab grouped by level and a binaryIndicator, where level is an argument that could be country or market. I hence want to combine the following examples into one, by making level an argument.

 SummaryTab1 = ddply(df, .(market,IndicatorFlag), summarize, counts=length(unique(hospitalID)))
 SummaryTab1 = ddply(df, .(country,IndicatorFlag), summarize, counts=length(unique(hospitalID)))

With level as an argument, I tried the following:

 level<-c("market")
 string<-paste(level,"IndicatorFlag",sep=" , ")
 SummaryTab1 = ddply(df,.(string) , summarize, counts=length(unique(hospitalID)))

which just gives a string

I also tried this

   SummaryTab1 =as.formula(paste0("ddply(df,.(",level,",IndicatorFlag),summarize, counts=length(unique(hospitalID)))"))

Any suggestions how to do this ?

What I'm trying to do is group by the level and the IndicatorFlag

In terms of what Miha suggested, I'm trying to do this (Neither works):

library(dplyr)

SumTab<-df %>% group_by_(my_level,IndicatorFlag) %>% summarise(counts = length(unique(hospitalID)))

SumTab<-ddply(df, .(my_level[2],IndicatorFlag), summarize, counts=length(unique(hospitalID)))

Miha Trošt · Accepted Answer · 2015-03-12T14:43:35.707

Is this what you are after...

library(plyr)
library(dplyr)

Data

set.seed(1)

df <-data.frame(category=rep(LETTERS[1:5],each=10),
                superregion=sample(c("EMEA","LATAM","AMER","APAC"),100,replace=T),
                country=sample(c("Country1","Country2","Country3","Country4","Country5","Country6","Country7","Country8"),100,replace=T),
                market=sample(c("Market1","Market2","Market3","Market4","Market5","Market6","Market7","Market8","Market9","Market10","Market11","Market12"),100,replace=T),
                hospitalID=sample(seq(1,50,1),100,replace=T),IndicatorFlag=sample(seq(0,1,1),100,replace=T))

Solution

dplyr

Solution updated with suggestions given by docendo discimus in comment section below.

# your level of choice
my_level <- "market"

# code
df %>%
  group_by_(my_level) %>%
  summarise(counts = n_distinct(hospitalID))

     market counts
1   Market1      5
2  Market10      4
3  Market11      3
4  Market12      9
5   Market2     12
6   Market3     10
7   Market4     12
8   Market5      8
9   Market6      9
10  Market7      7
11  Market8      4
12  Market9      5

# multiple levels
my_level <- c("market", "country", "IndicatorFlag")

  df %>%
  group_by_(.dots = my_level[c(1, 3)]) %>%
  summarise(counts = n_distinct(hospitalID))

     market IndicatorFlag counts
1   Market1             0      1
2   Market1             1      5
3  Market10             0      2
4  Market10             1      3
5  Market11             1      3
6  Market12             0      7
7  Market12             1      3
8   Market2             0      4
9   Market2             1     10
10  Market3             0      7
..      ...           ...    ...

 # using all levels

  df %>%
  group_by_(.dots = my_level) %>%
  summarise(counts = n_distinct(hospitalID))

     market  country IndicatorFlag counts
1   Market1 Country1             1      1
2   Market1 Country2             1      1
3   Market1 Country3             1      2
4   Market1 Country6             0      1
5   Market1 Country8             1      1
6  Market10 Country2             1      1
7  Market10 Country3             0      1
8  Market10 Country3             1      1
9  Market10 Country5             1      1
10 Market10 Country7             0      1
..      ...      ...           ...    ...

plyr

ddply(df, c(my_level[1], my_level[3]), 
      summarize, 
      counts = n_distinct(hospitalID)) %>%
  head(.)

    market IndicatorFlag counts
1  Market1             0      1
2  Market1             1      5
3 Market10             0      2
4 Market10             1      3
5 Market11             1      3
6 Market12             0      7

Hi Miha, no, I pass in a string called level which can take the value "market" or "country". The value of this argument should then be passed into the ddply command and aggregate at whatever the level is, i.e. country or market. Does that make sense? — TinaW, Mar 12 '15 at 08:50
Almost, I have edited my post to use your code. The problem is that I cannot group by level and IndicatorFlag, which I need to. I tried: 1) SumTab<-df %>% group_by_(my_level,IndicatorFlag) %>% summarise(counts = length(unique(hospitalID))) OR 2) SumTab<-ddply(df, .(my_level[2],IndicatorFlag), summarize, counts=length(unique(hospitalID))) - Neither works. — TinaW, Mar 12 '15 at 10:22
You do not use ddply correctly. Look at documentation on how to use grouping term. — Miha Trošt, Mar 12 '15 at 12:14
What is wrong? Here's another example using the same grouping syntax".(groupvar1,groupvar2)". "http://stackoverflow.com/questions/22239745/r-ddply-with-multiple-variables — TinaW, Mar 12 '15 at 12:56
Have you tried the above solution again? Does it work or not? — Miha Trošt, Mar 12 '15 at 12:58
I see now...you should reference IndicatorFlag as string like this: "IndicatorFlag" — Miha Trošt, Mar 12 '15 at 13:05
Miha, I'd like to upvote your answer but I think you should include a dplyr version with the `.dots = ` argument. Why? because without it, `group_by_` will only use the first element of a length > 1 character vector of grouping variables. (Check out `df %>% group_by_(my_level[1:2]) %>% groups`) and the version you supplied is a bit redundant because you have to call the same vector multiple times (`df %>% group_by_(my_level[1], my_level[3]) %>% groups`) instead of indexing it right away (`df %>% group_by_(.dots = my_level[c(1,3)]) %>% groups`). — talat, Mar 12 '15 at 13:22
Plus, if you wanted to group by all elements in the vector, you just need `df %>% group_by_(.dots = my_level) %>% groups` instead of `df %>% group_by_(my_level[1], my_level[2], my_level[3]) %>% groups` — talat, Mar 12 '15 at 13:23
Ok, I will update my answer later on today. Thanks for suggestions. — Miha Trošt, Mar 12 '15 at 13:26

passing strings as arguments in ddply

1 Answers1

Data

Solution

dplyr

plyr