Using a lookup table in R with continous values - r

Question

I have a lookup table in R that I am trying to figure out how to implement. The challenge for me is that it involves continuous values or ranges of data. If the value falls inbetween I'd like it to pick the right value.
I want to use the two continuous 'GRADE', 'SAT' variables plus the categorical 'TYPE' value to assign a 'GROUP' value. This big block of code looks intimidating but these are tiny tiny tables.
Any advice is appreciated!!!!
#lookup table code for recreating dataframe
structure(list(Type = structure(c(1L, 2L, 1L, 1L), .Label = c("A",
"B"), class = "factor"), min_grade = c(93L, 85L, 93L, 80L), max_grade = c(100L,
93L, 100L, 92L), min_sat = c(600L, 700L, 400L, 600L), max_sat = c(800L,
800L, 599L, 800L), Group = structure(c(1L, 1L, 2L, 3L), .Label = c("A",
"B", "C"), class = "factor")), .Names = c("Type", "min_grade",
"max_grade", "min_sat", "max_sat", "Group"), class = "data.frame", row.names = c(NA,
-4L))
#example ----- desired value is in the 'GROUP' column so this would be NULL before I used the lookup table
structure(list(Name = structure(c(3L, 1L, 2L, 4L), .Label = c("Jack",
"James", "John", "Jordan"), class = "factor"), Grade = c(95L,
95L, 92L, 93L), Sat = c(701L, 500L, 800L, 800L), Type = structure(c(1L,
1L, 1L, 2L), .Label = c("A", "B"), class = "factor"), Group = structure(c(1L,
2L, 3L, 1L), .Label = c("A", "B", "C"), class = "factor")), .Names = c("Name",
"Grade", "Sat", "Type", "Group"), class = "data.frame", row.names = c(NA,
-4L))

Answer 1

how abt this?
ltab <- structure(list(Type = structure(c(1L, 2L, 1L, 1L), .Label = c("A",
"B"), class = "factor"), min_grade = c(93L, 85L, 93L, 80L), max_grade = c(100L,
93L, 100L, 92L), min_sat = c(600L, 700L, 400L, 600L), max_sat = c(800L,
800L, 599L, 800L), Group = structure(c(1L, 1L, 2L, 3L), .Label = c("A",
"B", "C"), class = "factor")), .Names = c("Type", "min_grade",
"max_grade", "min_sat", "max_sat", "Group"), class = "data.frame", row.names = c(NA,
-4L))
dat <- structure(list(Name = structure(c(3L, 1L, 2L, 4L), .Label = c("Jack",
"James", "John", "Jordan"), class = "factor"), Grade = c(95L,
95L, 92L, 93L), Sat = c(701L, 500L, 800L, 800L), Type = structure(c(1L,
1L, 1L, 2L), .Label = c("A", "B"), class = "factor")), .Names = c("Name",
"Grade", "Sat", "Type"), class = "data.frame", row.names = c(NA,
-4L))
library(plyr)
mdat <- adply(merge(dat, ltab, by="Type", all=T), 1, function(x) {
c(FallsIn=x$Grade > x$min_grade & x$Grade <= x$max_grade & x$Sat > x$min_sat & x$Sat <= x$max_sat)
})
mdat[mdat$FallsIn,]
thinking about generalizing, are there going to be more continuous variables that you need to check?
EDIT: could not edit OP post so taking OP's comment into account is how I would tackle an example of "categorizing multidimensional continuous random variables"
so that these keywords will flag up in future searches
breaks <- list(Var1=c(0, 0.25, 1),
Var2=c(0, 0.5, 1),
Var3=c(0, 0.25, 0.75, 1))
#generate this on the fly
genIntv <- function(x) {
ret <- paste0("(", x[1:(length(x)-1)],", ",x[2:length(x)], "]")
names(ret) <- 1:(length(x)-1)
ret
}
lookupTbl <- data.frame(expand.grid(lapply(breaks, genIntv), stringsAsFactors=F),
Group=LETTERS[1:12])
lookupTbl2 <- data.frame(expand.grid(lapply(breaks, function(x) 1:(length(x)-1)), stringsAsFactors=F),
Group=LETTERS[1:12])
#data set
dat <- data.frame(Var1=c(0.1, 0.76), Var2=c(0.5, 0.75), Var3=c(0.25,0.9))
binDat <- do.call(cbind, setNames(lapply(1:ncol(dat), function(k)
.bincode(dat[,k], breaks[[k]], T, T)),colnames(dat)))
merge(binDat, lookupTbl2, all.x=T, all.y=F)
good to learn if someone else has better approaches

Answer 2

If you have small data, a full join should be fine.
library(dplyr)
result =
example %>%
select(-Type) %>%
full_join(look_up) %>%
filter(min_grade < Grade & Grade <= max_grade &
min_sat < Sat & Sat <= max_sat)