Speed up data.frame rearrangement - r

I have a data frame with coordinates ("start","end") and labels ("group"):
a <- data.frame(start=1:4, end=3:6, group=c("A","B","C","D"))
a
start end group
1 1 3 A
2 2 4 B
3 3 5 C
4 4 6 D
I want to create a new data frame in which labels are assigned to every element of the sequence on the range of coordinates:
V1 V2
1 1 A
2 2 A
3 3 A
4 2 B
5 3 B
6 4 B
7 3 C
8 4 C
9 5 C
10 4 D
11 5 D
12 6 D
The following code works but it is extremely slow with wide ranges:
df<-data.frame()
for(i in 1:dim(a)[1]){
s<-seq(a[i,1],a[i,2])
df<-rbind(df,data.frame(s,rep(a[i,3],length(s))))
}
colnames(df)<-c("V1","V2")
How can I speed this up?

You can try data.table
library(data.table)
setDT(a)[, start:end, by = group]
which gives
group V1
1: A 1
2: A 2
3: A 3
4: B 2
5: B 3
6: B 4
7: C 3
8: C 4
9: C 5
10: D 4
11: D 5
12: D 6
Obviously this would only work if you have one row per group, which it seems you have here.

If you want a very fast solution in base R, you can manually create the data.frame in two steps:
Use mapply to create a list of your ranges from "start" to "end".
Use rep + lengths to repeat the "groups" column to the expected number of rows.
The base R approach shared here won't depend on having only one row per group.
Try:
temp <- mapply(":", a[["start"]], a[["end"]], SIMPLIFY = FALSE)
data.frame(group = rep(a[["group"]], lengths(temp)),
values = unlist(temp, use.names = FALSE))
If you're doing this a lot, just put it in a function:
myFun <- function(indf) {
temp <- mapply(":", indf[["start"]], indf[["end"]], SIMPLIFY = FALSE)
data.frame(group = rep(indf[["group"]], lengths(temp)),
values = unlist(temp, use.names = FALSE))
}
Then, if you want some sample data to try it with, you can use the following as sample data:
set.seed(1)
a <- data.frame(start=1:4, end=sample(5:10, 4, TRUE), group=c("A","B","C","D"))
x <- do.call(rbind, replicate(1000, a, FALSE))
y <- do.call(rbind, replicate(100, x, FALSE))
Note that this does seem to slow down as the number of different unique values in "group" increases.
(In other words, the "data.table" approach will make the most sense in general. I'm just sharing a possible base R alternative that should be considerably faster than your existing approach.)

Related

Manipulating data.table column recursively on other column condition

I need to calculate a formula in a data frame. Each set of values across few columns have to be, lets say simplicity sake, aggregated. However, I do not want calculation across rows. I want to calculate each set with another set based on condition else where.
This is what I mean:
I have a data.table.
data = data.table(A = c("a","c","b","b","a"),
B = c(1:5),
C = c(1:5)
)
setorder(data, by=A)
> data
A B C
1: a 1 1
2: a 5 5
3: b 3 3
4: b 4 4
5: c 2 2
In column D I need to have and aggregate of values in B and C and values B and C when A is "a". As I have more than one "a", multiple aggregations are needed. From every aggregate minimum should be written in.
Here is an example.
For row 1: (1+1)+(1+1)=4, (5+5)+(1+1)=12, so 4 is minimum - D1 =4.
For row 3: (3+3)+(1+1)=8, (3+3)+(5+5)=16, D3 = 8. And so on.
This is what I expect
> data_new
A B C D
1: a 1 1 4
2: a 5 5 12
3: b 3 3 8
4: b 4 4 10
5: c 2 2 6
I tried this and run into issues.
for (i in data)data[i, D:=(min((data[i,B+C]) + (data[a=="a",(B+C)])))]
The expression below for minimum selection works fine on its own when I substitute i for a row number returning list of two numbers for min() returns proper value. Below answer is 8.
min((data[3,B+C]) + (data[A=="a",(B+C)]))
My previous attempts involved grid.expansion() and intersection(). However, with the size of my data set I ran into memory issue and Rstudio quit on me. As a side note, I need to run the calculations as I could not project the smallest outcome by "a" beforehand - it is a set of coordinates and they do not correlate with the magnitude of an answer.
Any suggestion where is my glaring issue
You can store the value of B + C where A = 'a' in a variable (val). For each row you can take minimum of B + C + val value.
library(data.table)
val <- data[A =='a', B + C]
data[, D := min(B + C + val), seq_len(nrow(data))]
data
# A B C D
#1: a 1 1 4
#2: a 5 5 12
#3: b 3 3 8
#4: b 4 4 10
#5: c 2 2 6
You can also use lapply :
data[, D := lapply(B + C, function(x) min(x + val))]
An option is also to replicate the 'a' rows after taking the min of 'B', 'C' and then do a direct + with the 'B', 'C' columns. The advantage is that, we don't have to group or loop
library(data.table)
Reduce(`+`, (data[A == 'a', .(B = min(B), C = min(C))][rep(seq_len(.N), nrow(data))] + data[, .(B, C)]))
#[1] 4 12 8 10 6
Or in a single line
data[, D := B + C + min(B[A== 'a']) + min(C[A== 'a'])]
data$D
#[1] 4 12 8 10 6

Group a data.table using a column which is list

I have a really big problem and looping through the data.table to do what I want is too slow, so I am trying to get around looping. Let assume I have a data.table as follows:
a <- data.table(i = c(1,2,3), j = c(2,2,6), k = list(c("a","b"),c("a","c"),c("b")))
> a
i j k
1: 1 2 a,b
2: 2 2 a,c
3: 3 6 b
And I want to group based on the values in k. So something like this:
a[, sum(j), by = k]
right now I am getting the following error:
Error in `[.data.table`(a, , sum(i), by = k) :
The items in the 'by' or 'keyby' list are length (2,2,1). Each must be same length as rows in x or number of rows returned by i (3).
The answer I am looking for is to group first all the rows having "a" in column k and calculate sum(j) and then all rows having "b" and so on. So the desired answer would be:
k V1
a 4
b 8
c 2
Any hint how to do it efficiently? I cant melt the column K by repeating the rows since the size of the data.table would be too big for my case.
I think this might work:
a[, .(k = unlist(k)), by=.(i,j)][,sum(j),by=k]
k V1
1: a 4
2: b 8
3: c 2
If we are using tidyr, a compact option would be
library(tidyr)
unnest(a, k)[, sum(j) ,k]
# k V1
#1: a 4
#2: b 8
#3: c 2
Or using the dplyr/tidyr pipes
unnest(a, k) %>%
group_by(k) %>%
summarise(V1 = sum(j))
# k V1
# <chr> <dbl>
#1 a 4
#2 b 8
#3 c 2
Since by-group operations can be slow, I'd consider...
dat = a[rep(1:.N, lengths(k)), c(.SD, .(k = unlist(a$k))), .SDcols=setdiff(names(a), "k")]
i j k
1: 1 2 a
2: 1 2 b
3: 2 2 a
4: 2 2 c
5: 3 6 b
We're repeating rows of cols i:j to match the unlisted k. The data should be kept in this format instead of using a list column, probably. From there, as in #MikeyMike's answer, we can dat[, sum(j), by=k].
In data.table 1.9.7+, we can similarly do
dat = a[, c(.SD[rep(.I, lengths(k))], .(k = unlist(k))), .SDcols=i:j]

data.table version of split and repeat

I'm trying to convert some code to use data.table. In this situation, I need to create a graph structure from columns in a data.frame/data.table where rows have information containing the id and depth in the tree. My normal approach is a split/apply/combine, so I feel like it should be possible using by and some expression in data.table but I can't get it.
Here is an example,
## A data.table like this with ids and levels
dat <- data.table(level = rep(1:4, times=2^(0:3)), id = 1:15)
## my normal way, not using data table would involve a split and rep
levs <- split(dat$id, dat$level)
nodes <- unlist(mapply(function(a,b) rep(a, length.out=b), head(levs, -1L),
tail(lengths(levs), -1L)), use.names = FALSE)
## Desired result
res <- cbind(nodes, dat$id[-1L])
## To visualize
library(igraph)
plot(graph_from_edgelist(cbind(nodes, dat$id[-1L])), layout=layout.reingold.tilford,
asp=0.6)
Edit
I think the problem I'm having is when I do a by=level I need information from two levels to get the proper repeat lenght.
Here's another way of getting your nodes column:
dat[, .N, by = .(level = level - 1)][
dat, on = 'level', nomatch = 0][
, .(nodes = rep(id, length.out = N[1])), by = level]
# level nodes
# 1: 1 1
# 2: 1 1
# 3: 2 2
# 4: 2 3
# 5: 2 2
# 6: 2 3
# 7: 3 4
# 8: 3 5
# 9: 3 6
#10: 3 7
#11: 3 4
#12: 3 5
#13: 3 6
#14: 3 7

Using merge command in r for merging depending upon column values

So, I have several dataframes like this
1 2 a
2 3 b
3 4 c
4 5 d
3 5 e
......
1 2 j
2 3 i
3 4 t
3 5 r
.......
2 3 t
2 4 g
6 7 i
8 9 t
......
What I want is, I want to merge all of these files into one single file showing the values of third column for each pair of values in columns 1 and columns 2 and 0 if that pair is not present.
So, the output for this will be, since, there are three files (there are more)
1 2 aj0
2 3 bit
3 4 ct0
4 5 d00
3 5 er0
6 7 00i
8 9 00t
......
What I did was combine all my text .txt files in a single list.
Then,
L <- lapply(seq_along(L), function(i) {
L[[i]][, paste0('DF', i)] <- 1
L[[i]]
})
Which will indicate the presence of a value when we will be merging them.
I don't know how to proceed further. Any inputs will be great. Thanks!
Here is one way to do it with Reduce
# function to generate dummy data
gen_data<- function(){
data.frame(
x = 1:3,
y = 2:4,
z = sample(LETTERS, 3, replace = TRUE)
)
}
# generate list of data frames to merge
L <- lapply(1:3, function(x) gen_data())
# function to merge by x and y and concatenate z
f <- function(x, y){
d <- merge(x, y, by = c('x', 'y'), all = TRUE)
# set merged column to zero if no match is found
d[['z.x']] = ifelse(is.na(d[['z.x']]), 0, d[['z.x']])
d[['z.y']] = ifelse(is.na(d[['z.y']]), 0, d[['z.y']])
d$z <- paste0(d[['z.x']], d[['z.y']])
d['z.x'] <- d['z.y'] <- NULL
return(d)
}
# merge data frames
Reduce(f, L)

How do I take subsets of a data frame according to a grouping in R?

I have an aggregation problem which I cannot figure out how to perform efficiently in R.
Say I have the following data:
group1 <- c("a","b","a","a","b","c","c","c","c",
"c","a","a","a","b","b","b","b")
group2 <- c(1,2,3,4,1,3,5,6,5,4,1,2,3,4,3,2,1)
value <- c("apple","pear","orange","apple",
"banana","durian","lemon","lime",
"raspberry","durian","peach","nectarine",
"banana","lemon","guava","blackberry","grape")
df <- data.frame(group1,group2,value)
I am interested in sampling from the data frame df such that I randomly pick only a single row from each combination of factors group1 and group2.
As you can see, the results of table(df$group1,df$group2)
1 2 3 4 5 6
a 2 1 2 1 0 0
b 2 2 1 1 0 0
c 0 0 1 1 2 1
shows that some combinations are seen more than once, while others are never seen. For those that are seen more than once (e.g., group1="a" and group2=3), I want to randomly pick only one of the corresponding rows and return a new data frame that has only that subset of rows. That way, each possible combination of the grouping factors is represented by only a single row in the data frame.
One important aspect here is that my actual data sets can contain anywhere from 500,000 rows to >2,000,000 rows, so it is important to be mindful of performance.
I am relatively new at R, so I have been having trouble figuring out how to generate this structure correctly. One attempt looked like this (using the plyr package):
choice <- function(x,label) {
cbind(x[sample(1:nrow(x),1),],data.frame(state=label))
}
df <- ddply(df[,c("group1","group2","value")],
.(group1,group2),
pick_junc,
label="test")
Note that in this case, I am also adding an extra column to the data frame called "label" which is specified as an extra argument to the ddply function. However, I killed this after about 20 min.
In other cases, I have tried using aggregate or by or tapply, but I never know exactly what the specified function is getting, what it should return, or what to do with the result (especially for by).
I am trying to switch from python to R for exploratory data analysis, but this type of aggregation is crucial for me. In python, I can perform these operations very rapidly, but it is inconvenient as I have to generate a separate script/data structure for each different type of aggregation I want to perform.
I want to love R, so please help! Thanks!
Uri
Here is the plyr solution
set.seed(1234)
ddply(df, .(group1, group2), summarize,
value = value[sample(length(value), 1)])
This gives us
group1 group2 value
1 a 1 apple
2 a 2 nectarine
3 a 3 banana
4 a 4 apple
5 b 1 grape
6 b 2 blackberry
7 b 3 guava
8 b 4 lemon
9 c 3 durian
10 c 4 durian
11 c 5 raspberry
12 c 6 lime
EDIT. With a data frame that big, you are better off using data.table
library(data.table)
dt = data.table(df)
dt[,list(value = value[sample(length(value), 1)]),'group1, group2']
EDIT 2: Performance Comparison: Data Table is ~ 15 X faster
group1 = sample(letters, 1000000, replace = T)
group2 = sample(LETTERS, 1000000, replace = T)
value = runif(1000000, 0, 1)
df = data.frame(group1, group2, value)
dt = data.table(df)
f1_dtab = function() {
dt[,list(value = value[sample(length(value), 1)]),'group1, group2']
}
f2_plyr = function() {ddply(df, .(group1, group2), summarize, value =
value[sample(length(value), 1)])
}
f3_by = function() {do.call(rbind,by(df,list(grp1 = df$group1,grp2 = df$group2),
FUN = function(x){x[sample(nrow(x),1),]}))
}
library(rbenchmark)
benchmark(f1_dtab(), f2_plyr(), f3_by(), replications = 10)
test replications elapsed relative
f1_dtab() 10 4.764 1.00000
f2_plyr() 10 68.261 14.32851
f3_by() 10 67.369 14.14127
One more way:
with(df, tapply(value, list( group1, group2), length))
1 2 3 4 5 6
a 2 1 2 1 NA NA
b 2 2 1 1 NA NA
c NA NA 1 1 2 1
# Now use tapply to sample withing groups
# `resample` fn is from the sample help page:
# Avoids an error with sample when only one value in a group.
resample <- function(x, ...) x[sample.int(length(x), ...)]
#Create a row index
df$idx <- 1:NROW(df)
rowidxs <- with(df, unique( c( # the `c` function will make a matrix into a vector
tapply(idx, list( group1, group2),
function (x) resample(x, 1) ))))
rowidxs
# [1] 1 5 NA 12 16 NA 3 15 6 4 14 10 NA NA 7 NA NA 8
df[rowidxs[!is.na(rowidxs)] , ]

Resources