"Not Join" in R - r

I am looking for a quick way to do 'not join' (i.e. keep rows that didn't merge, or inverse of inner join). The way I've been doing is to use data.table for X and Y, then set key. For example:
require(data.table)
X <- data.table(category = c('A','B','C','D'), val1 = c(0.2,0.3,0.8,0.7))
Y <- data.table(category = c('B','C','D','E'), val2 = c(2,3,5,7))
XY <- merge(X,Y,by='category')
> XY
category val1 val2
1: B 0.3 2
2: C 0.8 3
3: D 0.7 5
But I need the inverse of this, so I have to do:
XY_All <- merge(X,Y,by='category',all=TRUE)
setkey(XY,category)
setkey(XY_All,category)
notXY <- XY_All[!XY] #data.table not join (finally)
> notXY
category val1 val2
1: A 0.2 NA
2: E NA 7
I feel like this is quite long winded (especially from data.frame). Am I missing something?
EDIT: I got this after thinking more about not joins
X <- data.table(category = c('A','B','C','D'), val1 = c(0.2,0.3,0.8,0.7),key = "category")
Y <- data.table(category = c('B','C','D','E'), val2 = c(2,3,5,7), key = "category")
notXY <- merge(X[!Y],Y[!X],all=TRUE)
But WheresTheAnyKey's answer below is clearer. One last hurdle is the presetting data.table keys, it'd be nice not to have to do that.
EDIT: To clarify, the accepted solution is:
merge(anti_join(X, Y, by = 'category'),anti_join(Y, X, by = 'category'), by = 'category', all = TRUE)

require(dplyr)
rbind_list(anti_join(X, Y), anti_join(Y, X))
EDIT:
Since someone asked for some explanation, here's what is happening:
The first anti_join() function returns rows from X that have no matching row in Y with the match determined by what the join is joining by. The second does the reverse. rbind_list() just takes the results of its inputs and makes them into a single tbl with all the observations from each of its inputs, replacing missing variable data with NA.

setkey(X,category)
setkey(Y,category)
rbind(X[!Y], Y[!X], fill = TRUE)

You can make it more concise like this:
X <- data.table(category = c('A','B','C','D'), val1 = c(0.2,0.3,0.8,0.7),key = "category")
Y <- data.table(category = c('B','C','D','E'), val2 = c(2,3,5,7), key = "category")
notXY <- merge(X,Y,all = TRUE)[!merge(X,Y)]

Try this.
First, merge with "all" set to "TRUE". Then take out all complete cases:
XY_All <- merge(X,Y,by='category',all=TRUE)
notXY <- XY_All[!complete.cases(XY_All),]

require(dplyr)
notXY = merge(X[!X$category %in% Y$category,], Y[!Y$category %in% X$category,],by =
"category",all = TRUE)
One way to look at an Anti-Join is that you need observations from X not in Y and observations from Y not in X concatenated together. This can be achieved in one step as shown above.

Related

Replacing marks with grades dynamically

I have two dataframes:
marks<-data.frame("student" =c("stud1","stud2","stud3") ,"sub1" =c(25,75,43), "sub2" = c(43,99,45),"sub3" = c(32,53,45), stringsAsFactors = FALSE)
grades<-data.frame("grade" =c("F","B","A") ,"sub1" =c(50,75,85), "sub2" =c(35,75,85)),"sub3" =c(32,75,85), stringsAsFactors = FALSE)
(sample in image format as well)
I need to compare each mark in Marks df and get corresponding grades from Grades df. The grade definition is different for different subjects.
I have tried using lapply function and cut:
(marks<-sapply(marks, function(x) cut(x,
breaks=c(0,50,75,85),
labels=c("F","B","A"),include.lowest = TRUE, right = TRUE,na.rm = TRUE))
This works without difficulty if grade boundaries are fixed. But when they change dynamically (based on subject column) I could not do this.
The expected output is:
gradedmarks<-data.frame("student" =c("stud1","stud2","stud3") ,"sub1" =c("F","B","F"), "sub2" = c("F","A","F"),"sub3" = c("F","F","F"), stringsAsFactors = FALSE)
Any quick way to achieve this in R?
Please note this is NOT the duplicate of this(Looping through multiple if_else statements). This is about using cut function with dynamic values depending on the column.
Using a data.table non-equi join:
library(data.table)
setDT(marks)
setDT(grades)
#reshape to long format
marks <- melt(marks, id.vars = "student")
grades <- melt(grades, id.vars = "grade")
#non.equi join
marks[grades, grade := i.grade, on = c("variable", "value >= value")]
#fill "F" for low marks
marks[is.na(grade), grade := "F"]
#reshape to wide format
dcast(marks, student ~ variable, value.var = "grade")
# student sub1 sub2 sub3
#1 stud1 F F F
#2 stud2 B A F
#3 stud3 F F F
Study the data.table documentation to understand data.table syntax. There are some excellent vignettes.
You can use Map to change the grades dynamically :
marks[-1] <- Map(function(x, y) cut(x, breaks=c(0, y),
labels=grades$grade,include.lowest = TRUE, right = TRUE),
marks[-1], grades[-1])
You might need to adjust settings of cut function based on your requirement. Also this requires subjects to be in same order in marks and grades dataframe.

Is there a way to replace rows in one dataframe with another in R?

I'm trying to figure out how to replace rows in one dataframe with another by matching the values of one of the columns. Both dataframes have the same column names.
Ex:
df1 <- data.frame(x = c(1,2,3,4), y = c("a", "b", "c", "d"))
df2 <- data.frame(x = c(1,2), y = c("f", "g"))
Is there a way to replace the rows of df1 with the same row in df2 where they share the same x variable? It would look like this.
data.frame(x = c(1,2,3,4), y = c("f","g","c","d")
I've been working on this for a while and this is the closest I've gotten -
df1[which(df1$x %in% df2$x),]$y <- df2[which(df1$x %in% df2$x),]$y
But it just replaces the values with NA.
Does anyone know how to do this?
We can use match. :
inds <- match(df1$x, df2$x)
df1$y[!is.na(inds)] <- df2$y[na.omit(inds)]
df1
# x y
#1 1 f
#2 2 g
#3 3 c
#4 4 d
First off, well done in producing a nice reproducible example that's directly copy-pastable. That always helps, specially with an example of expected output. Nice one!
You have several options, but lets look at why your solution doesn't quite work:
First of all, I tried copy-pasting your last line into a new session and got the dreaded factor-error:
Warning message:
In `[<-.factor`(`*tmp*`, iseq, value = 1:2) :
invalid factor level, NA generated
If we look at your data frames df1 and df2 with the str function, you will see that they do not contain text but factors. These are not text - in short they represent categorical data (male vs. female, scores A, B, C, D, and F, etc.) and are really integers that have a text as label. So that could be your issue.
Running your code gives a warning because you are trying to import new factors (labels) into df1 that don't exist. And R doesn't know what to do with them, so it just inserts NA-values.
As r2evens answered, he used the stringsAsFactors to disable using strings as Factors - you can even go as far as disabling it on a session-wide basis using options(stringsAsFactors=FALSE) (and I've heard it will be disabled as default in forthcoming R4.0 - yay!).
After disabling stringsAsFactors, your code works - or does it? Try this on for size:
df2 <- df2[c(2,1),]
df1[which(df1$x %in% df2$x),]$y <- df2[which(df1$x %in% df2$x),]$y
What's in df1 now? Not quite right anymore.
In the first line, I swapped the two rows in df2 and lo and behold, the replaced values in df1 were swapped. Why is that?
Let's deconstruct your statement df2[which(df1$x %in% df2$x),]$y
Call df1$x %in% df2$x returns a logical vector (boolean) of which elements in df1$x are found ind df2 - i.e. the first two and not the second two. But it doesn't relate which positions in the first vector corresponds to which in the second.
Calling which(df1$x %in% df2$x) then reduces the logical vector to which indices were TRUE. Again, we do not now which elements correspond to which.
For solutions, I would recommend r2evans, as it doesn't rely on extra packages (although data.table or dplyr are two powerful packages to get to know).
In his solution, he uses merge to perform a "full join" which matches rows based on the value, rather than - well, what you did. With transform, he assigns new variables within the context of the data.frame returned from the merge function called in the first argument.
I think what you need here is a "merge" or "join" operation.
(I add stringsAsFactors=FALSE to the frames so that the merging and later work is without any issue, as factors can be disruptive sometimes.)
Base R:
df1 <- data.frame(x = c(1,2,3,4), y = c("a", "b", "c", "d"), stringsAsFactors = FALSE)
# df2 <- data.frame(x = c(1,2), y = c("f", "g"), stringsAsFactors = FALSE)
merge(df1, df2, by = "x", all = TRUE)
# x y.x y.y
# 1 1 a f
# 2 2 b g
# 3 3 c <NA>
# 4 4 d <NA>
transform(merge(df1, df2, by = "x", all = TRUE), y = ifelse(is.na(y.y), y.x, y.y))
# x y.x y.y y
# 1 1 a f f
# 2 2 b g g
# 3 3 c <NA> c
# 4 4 d <NA> d
transform(merge(df1, df2, by = "x", all = TRUE), y = ifelse(is.na(y.y), y.x, y.y), y.x = NULL, y.y = NULL)
# x y
# 1 1 f
# 2 2 g
# 3 3 c
# 4 4 d
Dplyr:
library(dplyr)
full_join(df1, df2, by = "x") %>%
mutate(y = coalesce(y.y, y.x)) %>%
select(-y.x, -y.y)
# x y
# 1 1 f
# 2 2 g
# 3 3 c
# 4 4 d
A join option with data.table where we join on the 'x' column, assign the values of 'y' in second dataset (i.y) to the first one with :=
library(data.table)
setDT(df1)[df2, y := i.y, on = .(x)]
NOTE: It is better to use stringsAsFactors = FALSE (in R 4.0.0 - it is by default though) or else we need to have all the levels common in both datasets

How can I create a function to generate new variables based on values in different dataframe in R

I would like to create a function like this (obviously not proper code):
forEach ID in DATAFRAME1 look at each row with ID in DATAFRAME2 {
if DATAFRAME2$VARIABLE1 = something {
DATAFRAME1$VARIABLE1 = TRUE;
DATAFRAME1$VARIABLE2 = DATAFRAME2$VARIABLE2
}
}
In plain text, I've got a list of individuals and a database with mixed information on these
individuals. Let's say DATAFRAME2 contains informations on books read c(id, title, author, date). I want to create a new variable in DATAFRAME1 with a boolean of if the individual has read a specific book (VARIABLE1 above) and the date they first read it (VARIABLE2above). Also adding a third variable with number of times read would be interesting but not neccesary.
I haven't really done this in R before, mostly doing basic statistics and basic wrangling with dplyr. I guess I could use dplyr and join but this feels like a better approach. Any help to get me started would be much appreciated.
The following function does what the question asks for. Its arguments are
DF1 and DF2 have an obvious meaning;
var1 and var2 are VARIABLE1 and VARIABLE2 in the question;
value is the value of something.
The test data is at the end.
fun <- function(DF1, DF2, ID = 'ID', var1, var2, value){
DF1[[var1]] <- NA
DF1[[var2]] <- NA
k <- DF2[[var1]] == value
for(id in df1[[ID]]){
i <- DF1[[ID]] == id
j <- DF2[[ID]] == id
if(any(j & k)){
DF1[[var1]][i] <- TRUE
DF1[[var2]][i] <- DF2[[var2]][j & k]
}
}
DF1
}
fun(df1, df2, value = 4, var1 = 'X', var2 = 'Y')
# ID X Y
#1 a NA NA
#2 d TRUE 19
Test data.
set.seed(1234)
df1 <- data.frame(ID = c("a", "d"))
df2 <- data.frame(ID = rep(letters[1:5], 4),
X = sample(20, 20, TRUE),
Y = sample(20))

Create column names based on "by" argument the data.table way

Say I have the following data.table
dt <- data.table(var = c("a", "b"), val = c(1, 2))
Now I want to add two new columns to dt, named a, and b with the respective values (1, 2). I can do this with a loop, but I want to do it the data.table way.
The result would be a data.table like this:
dt.res <- data.table(var = c("a", "b"), val = c(1, 2), #old vars
a = c(1, NA), b = c(NA, 2)) # newly created vars
So far I came up with something like this
dt[, c(xx) := val, by = var]
where xx would be a data.table-command similar to .N which addresses the value of the by-group.
Thanks for the help!
Appendix: The for-loop way
The non-data.table-way with a for-loop instead of a by-argument would look something like this:
for (varname in dt$var){
dt[var == varname, c(varname) := val]
}
Based on the example showed, we can use dcast from the data.table to convert the long format to wide, and join with the original dataset on the 'val' column.
library(data.table)#v1.9.6+
dt[dcast(dt, val~var, value.var='val'), on='val']
# var val a b
#1: a 1 1 NA
#2: b 2 NA 2
Or as #CathG mentioned in the comments, for previous versions either merge or set the key column and then join.
merge(dt, dcast.data.table(dt, val~var, value.var='val'))

Merge two data frames while keeping the original row order

I want to merge two data frames keeping the original row order of one of them (df.2 in the example below).
Here are some sample data (all values from class column are defined in both data frames):
df.1 <- data.frame(class = c(1, 2, 3), prob = c(0.5, 0.7, 0.3))
df.2 <- data.frame(object = c('A', 'B', 'D', 'F', 'C'), class = c(2, 1, 2, 3, 1))
If I do:
merge(df.2, df.1)
Output is:
class object prob
1 1 B 0.5
2 1 C 0.5
3 2 A 0.7
4 2 D 0.7
5 3 F 0.3
If I add sort = FALSE:
merge(df.2, df.1, sort = F)
Result is:
class object prob
1 2 A 0.7
2 2 D 0.7
3 1 B 0.5
4 1 C 0.5
5 3 F 0.3
But what I would like is:
class object prob
1 2 A 0.7
2 1 B 0.5
3 2 D 0.7
4 3 F 0.3
5 1 C 0.5
You just need to create a variable which gives the row number in df.2. Then, once you have merged your data, you sort the new data set according to this variable. Here is an example :
df.1<-data.frame(class=c(1,2,3), prob=c(0.5,0.7,0.3))
df.2<-data.frame(object=c('A','B','D','F','C'), class=c(2,1,2,3,1))
df.2$id <- 1:nrow(df.2)
out <- merge(df.2,df.1, by = "class")
out[order(out$id), ]
Check out the join function in the plyr package. It's like merge, but it allows you to keep the row order of one of the data sets. Overall, it's more flexible than merge.
Using your example data, we would use join like this:
> join(df.2,df.1)
Joining by: class
object class prob
1 A 2 0.7
2 B 1 0.5
3 D 2 0.7
4 F 3 0.3
5 C 1 0.5
Here are a couple of links describing fixes to the merge function for keeping the row order:
http://www.r-statistics.com/2012/01/merging-two-data-frame-objects-while-preserving-the-rows-order/
http://r.789695.n4.nabble.com/patching-merge-to-allow-the-user-to-keep-the-order-of-one-of-the-two-data-frame-objects-merged-td4296561.html
You can also check out the inner_join function in Hadley's dplyr package (next iteration of plyr). It preserves the row order of the first data set. The minor difference to your desired solution is that it also preserves the original column order of the first data set. So it does not necessarily put the column we used for merging at the first position.
Using your example above, the inner_join result looks like this:
inner_join(df.2,df.1)
Joining by: "class"
object class prob
1 A 2 0.7
2 B 1 0.5
3 D 2 0.7
4 F 3 0.3
5 C 1 0.5
From data.table v1.9.5+, you can do:
require(data.table) # v1.9.5+
setDT(df.1)[df.2, on="class"]
The performs a join on column class by finding out matching rows in df.1 for each row in df.2 and extracting corresponding columns.
For the sake of completeness, updating in a join preserves the original row order as well. This might be an alternative to Arun's data.table answer if there are only a few columns to append:
library(data.table)
setDT(df.2)[df.1, on = "class", prob := i.prob][]
object class prob
1: A 2 0.7
2: B 1 0.5
3: D 2 0.7
4: F 3 0.3
5: C 1 0.5
Here, df.2 is right joined to df.1 and gains a new column prob which is copied from the matching rows of df.1.
The accepted answer proposes a manual way to keep order when using merge, which works most of the times but requires unnecessary manual work. This solution comes on the back of How to ddply() without sorting?, which deals with the issue of keeping order but in a split-apply-combine context:
This came up on the plyr mailing list a while back (raised by #kohske no less) and this is a solution offered by Peter Meilstrup for limited cases:
#Peter's version used a function gensym to
# create the col name, but I couldn't track down
# what package it was in.
keeping.order <- function(data, fn, ...) {
col <- ".sortColumn"
data[,col] <- 1:nrow(data)
out <- fn(data, ...)
if (!col %in% colnames(out)) stop("Ordering column not preserved by function")
out <- out[order(out[,col]),]
out[,col] <- NULL
out
}
So now you can use this generic keeping.order function to keep the original row order of a merge call:
df.1<-data.frame(class=c(1,2,3), prob=c(0.5,0.7,0.3))
df.2<-data.frame(object=c('A','B','D','F','C'), class=c(2,1,2,3,1))
keeping.order(df.2, merge, y=df.1, by = "class")
Which will yield, as requested:
> keeping.order(df.2, merge, y=df.1, by = "class")
class object id prob
3 2 A 1 0.7
1 1 B 2 0.5
4 2 D 3 0.7
5 3 F 4 0.3
2 1 C 5 0.5
So keeping.order effectively automates the approach in the accepted answer.
Thanks to #PAC , I came up with something like this:
merge_sameord = function(x, y, ...) {
UseMethod('merge_sameord')
}
merge_sameord.data.frame = function(x, y, ...) {
rstr = paste(sample(c(0:9, letters, LETTERS), 12, replace=TRUE), collapse='')
x[, rstr] = 1:nrow(x)
res = merge(x, y, all.x=TRUE, sort=FALSE, ...)
res = res[order(res[, rstr]), ]
res[, rstr] = NULL
res
}
This assumes that you want to preserve the order the first data frame, and the merged data frame will have the same number of rows as the first data frame. It will give you the clean data frame without extra columns.
In this specific case you could us factor for a compact base solution:
df.2$prob = factor(df.2$class,labels=df.1$prob)
df.2
# object class prob
# 1 A 2 0.7
# 2 B 1 0.5
# 3 D 2 0.7
# 4 F 3 0.3
# 5 C 1 0.5
Not a general solution however, it works if:
You have a lookup table containing unique values
You want to update a table, not create a new one
the lookup table is sorted by the merging column
The lookup table doesn't have extra levels
You want a left_join
If you're fine with factors
1 is not negotiable, for the rest we can do:
df.3 <- df.2 # deal with 2.
df.1b <- df.1[order(df.1$class),] # deal with 3
df.1b <- df.1b[df.1$class %in% df.2$class,] # deal with 4.
df.3$prob = factor(df.3$class,labels=df.1b$prob)
df.3 <- df3[!is.na(df.3$prob),] # deal with 5. if you want an `inner join`
df.3$prob <- as.numeric(as.character(df.3$prob)) # deal with 6.
For package developers
As a package developer, you want to be dependent on as few other packages as possible. Especially tidyverse functions, that change way too often for package developers IMHO.
To be able to make use of the join functions of the dplyr package without importing dplyr, below is a quick implementation. It keeps the original sorting (as requested by OP) and does not move the joining column to the front (which is another annoying thing of merge()).
left_join <- function(x, y, ...) {
merge_exec(x = x, y = y, all.x = TRUE, ...)
}
right_join <- function(x, y, ...) {
merge_exec(x = x, y = y, all.y = TRUE, ...)
}
inner_join <- function(x, y, ...) {
merge_exec(x = x, y = y, all = TRUE, ...)
}
full_join <- function(x, y, ...) {
merge_exec(x = x, y = y, ...)
}
# workhorse:
merge_exec <- function(x, y, ...) {
# set index
x$join_id_ <- 1:nrow(x)
# do the join
joined <- merge(x = x, y = y, sort = FALSE, ...)
# get suffices (yes, I prefer this over suffixes)
if ("suffixes" %in% names(list(...))) {
suffixes <- list(...)$suffixes
} else {
suffixes <- c("", "")
}
# get columns names in right order, so the 'by' column won't be forced first
cols <- unique(c(colnames(x),
paste0(colnames(x), suffixes[1]),
colnames(y),
paste0(colnames(y), suffixes[2])))
# get the original row and column index
joined[order(joined$join_id),
cols[cols %in% colnames(joined) & cols != "join_id_"]]
}
The highest rated answer does not produce what the Original Poster would like, i.e., "class" in column 1. If OP would allow switching column order in df.2, then here is a possible base R non-merge one-line answer:
df.1 <- data.frame(class = c(1, 2, 3), prob = c(0.5, 0.7, 0.3))
df.2 <- data.frame(class = c(2, 1, 2, 3, 1), object = c('A', 'B', 'D', 'F', 'C'))
cbind(df.2, df.1[match(df.2$class, df.1$class), -1, drop = FALSE])
I happen to like the information portrayed in the row.names. A complete one-liner that exactly duplicates the OP's desired outcome is
data.frame(cbind(df.2, df.1[match(df.2$class, df.1$class), -1, drop = FALSE]),
row.names = NULL)
I agree with https://stackoverflow.com/users/4575331/ms-berends that the fewer dependencies of a package developer on another package (or "verse") the better because development paths frequently diverge over time.
Note: The one-liner above does not work when there are duplicates in df.1$class. This can be overcome sans merge with 'outer' and a loop, or more generally with Ms Berend's clever post-merge rescrambling code.
There are several uses cases in which a simple subset will do:
# Use the key variable as row.names
row.names(df.1) = df.1$key
# Sort df.1 so that it's rows match df.2
df.3 = df.1[df.2$key, ]
# Create a data.frame with cariables from df.1 and (the sorted) df.2
df.4 = cbind(df.1, df.3)
This code will preserve df.2 and it's order and add only matching data from df.1
If only one variable is to be added, the cbind() ist not required:
row.names(df.1) = df.1$key
df.2$data = df.1[df.2$key, "data"]
I had the same problem with it but I simply used a dummy vector c(1:5) applied to a new column 'num'
df.2 <- data.frame(object = c('A', 'B', 'D', 'F', 'C'), class = c(2, 1, 2, 3, 1))
df.2$num <- c(1:5) # This range you can order in the last step.
dfm <- merge(df.2, df.1) # merged
dfm <- dfm[order(dfm$num),] # ascending order
There may be a more efficient way in base. This would be fairly simple to make into a function.
varorder <- names(mydata) # --- Merge
mydata <- merge(mydata, otherData, by="commonVar")
restOfvars <- names(mydata[!(names(mydata) %in% varorder)])
mydata[c(varorder,restOfvars)]

Resources