Skip to content

Computing means of multiple variables based on a condition

5 messages · KMNanus, William Dunlap, Jeff Newmiller +1 more

KMNanus
# 
I have a large dataset, a sample of which is:

a<- c(?A?, ?B?,?A?, ?B?,?A?, ?B?,?A?, ?B?,?A?, ?B?)
b <-c(15, 35, 20,  99, 75, 64, 33, 78, 45, 20)
c<- c( 111, 234, 456, 876, 246, 662, 345, 480, 512, 179)
d<- c(1.1, 3.2, 14.2, 8.7, 12.5, 5.9, 8.3, 6.0, 2.9, 9.3) 

df <- data.frame(a,b,c,d)

I?m trying to construct a data frame that shows the means of c & b based on the condition of d and grouped by a.

I want to create the data frame below, then use ggplot2 to create a line plot of b at various conditions of d.

I can compute the grouped means (d>=2, d>=4, etc.) one at a time using dplyr but haven?t figured out how to put them all together or put them in one data frame.

I?d rather not use a loop and am relatively new to R.  Is there a way i can use tapply and set it to the conditions above so that I can create the df below?


        condition    mean(b)     mean(c)    
A        d>=2          ____         _____
B        d>=2          ____         _____
A        d>=4          ____         _____
B        d>=4         ____         _____
A        d>=6         ____         _____
B       d>=6         ____         _____



Ken
kmnanus at gmail.com
914-450-0816 (tel)
347-730-4813 (fax)
William Dunlap
# 
Just to be clear, do you really want your 'condition' groups to be be
subsets
of one another?  Most (all?) of the *ply functions assume you want
non-overlapping groups so they do a split-summarize-combine sequence.
You would have to replace the split part of that.

Bill Dunlap
TIBCO Software
wdunlap tibco.com
On Wed, May 25, 2016 at 3:37 PM, KMNanus <kmnanus at gmail.com> wrote:

            

      
      
      
    

        

      
      
    

        

  
  


  


      

    

    

      
      

        


  

        

      KMNanus
    

    

      
      #
    

  


  

      
These will be overlapping subgroups from the same data frame.  For example, d<=2 will have length=9, d<=4 will have length=7, etc.


Ken
kmnanus at gmail.com
914-450-0816 (tel)
347-730-4813 (fax)

            

      
      
      
    

        

      
      
    

  
  


  


      

    

    

      
      

        


  

        

      Jeff Newmiller
    

    

      
      #
    

  


  

      
Thank you for including some sample data, but I have to ask that you 
please invest some time in learning how to edit your code in a text editor 
and to post in plain text. The quote marks in your example were "curly", 
which R does not understand. There are other ways in which HTML email 
leads to corruption on this mailing list as well, so you will save 
everyone numerous headaches by investing this time sooner rather than 
later.

The type of operation you are looking for is referred to as an "outer 
join" in SQL nomenclature, and it is intrinsically slow because the only 
way to accomplish it is computationally equivalent to a for loop that 
successively applies each minimum "d" value to your whole data set.

Having said that, you can accomplish this in the "dplyr" syntax instead of 
using a for loop, if that makes you happy, but it is not really any 
"better" than a for loop (and some people might consider it misleading 
to drape a for loop in such fancy syntax):

DF <- data.frame( a = c( "A", "B", "A", "B", "A", "B", "A", "B", "A", "B" )
                 , b = c( 15, 35, 20,  99, 75, 64, 33, 78, 45, 20 )
                 , c = c( 111, 234, 456, 876, 246, 662, 345, 480, 512, 179 )
                 , d = c( 1.1, 3.2, 14.2, 8.7, 12.5, 5.9, 8.3, 6.0, 2.9, 9.3 )
                 , stringsAsFactors = FALSE
                 )
passes <- data.frame( dmin = c( 2, 4, 6 ) )

library(dplyr)

DF2 <- (   passes
        %>% rowwise
        %>% do({ # run once for each row in "passes"
             dmin <- .$dmin # dot here refers to row of
                            # "passes" data frame
             (   DF
             %>% filter( d >= dmin )
             %>% group_by( a )
             %>% summarise( meanb = mean( b )
                          , meanc = mean( c )
                          )
             %>% mutate( condition = paste0( "d>=", dmin ) )
             )
            })
        %>% select( a, condition, meanb, meanc )
        %>% as.data.frame
        )

        
On Wed, 25 May 2016, KMNanus wrote:

        

            

      
      
      
    

        

      
      
    

            

      
      
      
    

        

      
      
    

        
---------------------------------------------------------------------------
Jeff Newmiller                        The     .....       .....  Go Live...
DCN:<jdnewmil at dcn.davis.ca.us>        Basics: ##.#.       ##.#.  Live Go...
                                       Live:   OO#.. Dead: OO#..  Playing
Research Engineer (Solar/Batteries            O.O#.       #.O#.  with
/Software/Embedded Controllers)               .OO#.       .OO#.  rocks...1k

  
  


  


      

    

    

      
      

        


  

        

      Thierry Onkelinx
    

    

      
      #
    

  


  

      
Another option would be to convert the data into a long format and add
columns for each condition.

library(dplyr)
library(tidyr)
DF %>%
  gather(key = "key", value = "value", -a, -d) %>%
  mutate(
    "d>=2" = ifelse(d >= 2, value, NA),
    "d>=4" = ifelse(d >= 4, value, NA),
    "d>=6" = ifelse(d >= 6, value, NA)
  ) %>%
  select(-d, -value) %>%
  gather(key = "condition", value = "value", -a, -key, na.rm = TRUE) %>%
  group_by(a, key, condition) %>%
  summarise(mean = mean(value)) %>%
  spread(key = key, value = mean) %>%
  arrange(condition, a)


ir. Thierry Onkelinx
Instituut voor natuur- en bosonderzoek / Research Institute for Nature and
Forest
team Biometrie & Kwaliteitszorg / team Biometrics & Quality Assurance
Kliniekstraat 25
1070 Anderlecht
Belgium

To call in the statistician after the experiment is done may be no more
than asking him to perform a post-mortem examination: he may be able to say
what the experiment died of. ~ Sir Ronald Aylmer Fisher
The plural of anecdote is not data. ~ Roger Brinner
The combination of some data and an aching desire for an answer does not
ensure that a reasonable answer can be extracted from a given body of data.
~ John Tukey

2016-05-26 8:34 GMT+02:00 Jeff Newmiller <jdnewmil at dcn.davis.ca.us>: