Doing a ranged lookup with multiple variables in a matrix in R - r

Question

I feel like I have a bit of a complicated problem (or at least for me it is!).
I have a table of prices which will need to be read from a csv which will look exactly like this:
V1 <- c("","Destination","Spain","Spain","Spain","Portugal","Portugal","Portugal","Italy","Italy","Italy")
V2 <- c("","Min_Duration",rep(c(1,3,6),3))
V3 <- c("","Max_Duration",rep(c(2,5,10),3))
V4 <- c("Full-board","Level_1",runif(9,100,200))
V5 <- c("Full-board","Level_2",runif(9,201,500))
V6 <- c("Full-board","Level_3",runif(9,501,1000))
V7 <- c("Half-board","Level_1",runif(9,100,200))
V8 <- c("Half-board","Level_2",runif(9,201,500))
V9 <- c("Half-board","Level_3",runif(9,501,1000))
Lookup_matrix <- as.data.frame(cbind(V1,V2,V3,V4,V5,V6,V7,V8))
The prices in the above table will of course come out a bit strange as they're completely random - but we can ignore that...
I also have a table like this:
Destination <- c("Spain", "Italy", "Portugal")
Duration <- c(2,4,8)
Level <- c(1,3,3)
Board <- c("Half-board","Half-board","Full-board")
Price <- "Empty"
Price_matrix <- as.data.frame(cbind(Destination,Duration,Level,Board,Price))
My question is - how do I populate the 'Price' column of the price matrix with the corresponding prices that can be found in the lookup matrix? Please note that the duration variable of the price matrix will have to fit into a range found between the 'Min_Duration' and 'Max_Duration' columns in the lookup matrix.
In Excel I would use an Index,Match formula. But I'm stumped with R.
Thanks in advance,
Dan

Answer 1

Here is a tidyverse possibility
First, please note that I rename your input objects; both Price_matrix and Lookup_matrix are data.frames (not matrices).
df1 <- Price_matrix
df2 <- Lookup_matrix
Next we need to fix the column names of df2 = Lookup_matrix.
# Fix column names
colnames(df2) <- gsub("^_", "", apply(df2[1:2, ], 2, paste0, collapse = "_"))
df2 <- df2[-(1:2), ]
We now basically do a left join of df1 and df2; in order for df2 to be in a suitable format we spread data from wide to long, extract Price values for every Board and Level, and expand entries from Min_Duration to Max_Duration. Then we join by Destination, Duration, Level and Board.
Note that in your example, Destination = Italy has no Level = 3 entry in Lookup_matrix; we therefore get Price = NA for this entry.
library(tidyverse)
left_join(
df1 %>%
mutate_if(is.factor, as.character) %>%
select(-Price),
df2 %>%
mutate_if(is.factor, as.character) %>%
gather(key, Price, -Destination, -Min_Duration, -Max_Duration) %>%
separate(key, into = c("Board", "Level"), sep = "_", extra = "merge") %>%
mutate(Level = sub("Level_", "", Level)) %>%
rowwise() %>%
mutate(Duration = list(seq(as.numeric(Min_Duration), as.numeric(Max_Duration)))) %>%
unnest() %>%
select(-Min_Duration, -Max_Duration) %>%
mutate(Duration = as.character(Duration)))
#Joining, by = c("Destination", "Duration", "Level", "Board")
# Destination Duration Level Board Price
#1 Spain 2 1 Half-board 119.010942545719
#2 Italy 4 3 Half-board <NA>
#3 Portugal 8 3 Full-board 764.536124917446

Answer 2

Using datatable:
library(data.table)
nms = trimws(do.call(paste, transpose(Lookup_matrix[1:2, ])))# column names
cat(do.call(paste, c(collapse="\n", Lookup_matrix[-(1:2), ])), file = "mm.csv")
# Rewrite the data in the correct format. You do not have to.
# Just doing Lookup_matrix1 = setNames(Lookup_matrix[-(1:2),],nms) is enough
# but it will not have rectified the column classes.
Lookup_matrix1 = fread("mm.csv", col.names = nms)
melt(Lookup_matrix1, 1:3)[,
c("Board", "Level") := .(sub("[.]", "-", sub("\\.Leve.*", "", variable)), sub("\\D+", "", variable))][
Price_matrix[, -5], on=c("Destination", "Board", "Level", "Min_Duration <= Duration", "Max_Duration >= Duration")]
Destination Min_Duration Max_Duration variable value Board Level
1: Spain 2 2 Half.board.Level_1 105.2304 Half-board 1
2: Italy 4 4 <NA> NA Half-board 3
3: Portugal 8 8 Full.board.Level_3 536.5132 Full-board 3