function (x,y), both x and y to vary - r

Question

I have a data frame consisting of about 22 fields, some system ids and some measurements, such as
bsystemid dcesystemid lengthdecimal heightquantity
2218 58 22 263
2219 58 22 197
2220 58 22 241
What I want:
1 . loop through a list of field ids
2 . define a function to test for a condition
3 . such that both x and y can vary
Where does the y variable definition belong, for varying both x and y? Other different structures?
This code block works for a single field and value of y:
varlist4<-names(brg) [c(6)]
f1<-(function(x,y) count(brg[,x]<y) )
lapply(varlist4, f1, y=c(7.5))
This code block executes, but the counts are off:
varlist4<-names(brg) [c(6,8,10,12)]
f1<-(function(x,y) count(brg[,x]<y) )
lapply(varlist4, f1, y=c(7.5,130,150,0))
For example,
varlist4<-names(brg) [c(6)]
f1<-(function(x,y) count(brg[,x]<y) )
lapply(varlist4, f1, y=c(7.5))
returns (correctly),
x freq
1 FALSE 9490
2 TRUE 309
3 NA 41
whereas the multiple x,y block of code above returns this for the first case,
x freq
1 FALSE 4828
2 TRUE 4971
3 NA 41
Thanks for any comments.
Update:
What I would like is to automate counting of occurances of values in specified fields in a df, meeting some condition. The conditions are numeric constants or text strings, one for each field. For example, I might want to count occurances meeting the condition >360 in field1, >0 in field2, etc. What I thus mean by allowing x and y to vary is reading x and y vectors with the field names and corresponding conditions into a looping structure.
I'd like to automate this task because it involves around 30 tables, each with up to 50 or so fields. And I'll need to do it twice, scanning once for values exceeding a maximum and once for values less than a minimum. Better still might be loading the conditions into a table and referencing that in the loop. That may be the next step but I'd like to understand this piece first.
This working example
t1<-18:29
t2<-c(76.1,77,78.1,78.2,78.8,79.7,79.9,81.1,81.2,81.8,82.8,83.5)
t3<-c(1.2,-0.2,-0.3,1.2, 2.2,0.4,0.6,0.4,-0.8,-0.1,5.0,3.1)
t<-data.frame(v1=t1,v2=t2,v3=t3)
varlist<-names(t) [c(1)]
f1<-(function(x,y) count(t[,x]>y) )
lapply(varlist, f1, y=c(27))
illustrates the correct answer for the first field, returning
x freq
1 FALSE 10
2 TRUE 2
But if I add in other fields and the corresponding conditions (the y's) I get something different for the first case:
varlist<-names(t) [c(1,2,3)]
f1<-(function(x,y) count(t[,x]>y) )
lapply(varlist, f1, y=c(27,83,3))
x freq
1 FALSE 8
2 TRUE 4
[[2]]
x freq
1 FALSE 1
2 TRUE 11
[[3]]
x freq
1 FALSE 11
2 TRUE 1
My sense is I'm not going about structuring the y part correctly.
Thanks for any comments.

Answer 1

You can use mapply. Let's create some data:
set.seed(123) # to get exactly the same results
brg = data.frame(x = rnorm(100), y=rnorm(100), z=rnorm(100))
brg$x[c(10, 15)] = NA # some NAs
brg$y[c(12, 21)] = NA # more NAs
Then you need to define the function to do the job. The function .f1 counts the data, and ensure there are always three levels (TRUE, FALSE, NA). Then, f1 uses .f1 in an mapply context to be able to vary x and y. Finally, some improvements in the output (changing the names of the columns).
f1 = function(x, y, data) {
.f1 = function(x, y, data) {
out = factor(data[, x] < y,
levels=c("TRUE", "FALSE", NA), exclude=NULL)
return(table(out))
}
out = mapply(.f1, x, y, MoreArgs = list(data = data)) # check ?mapply
colnames(out) = paste0(x, "<", y) # more clear names for the output
return(out)
}
Finally, the test:
varlist = names(brg)
threshold = c(0, 1, 1000)
f1(x=varlist, y=threshold, data=brg)
And you should get
x<0 y<1 z<1000
TRUE 46 87 100
FALSE 52 11 0
<NA> 2 2 0