I have a dataframe that I would like to group in both directions, first rowise and columnwise after. The first part worked well, but I am stuck with the second one. I would appreciate any help or advice for a solution that does both steps at the same time.
This is the dataframe:
df1 <- data.frame(
ID = c(rep(1,5),rep(2,5)),
ID2 = rep(c("A","B","C","D","E"),2),
A = rnorm(10,20,1),
B = rnorm(10,50,1),
C = rnorm(10,10,1),
D = rnorm(10,15,1),
E = rnorm(10,5,1)
)
This is the second dataframe, which holds the "recipe" for grouping:
df2 <- data.frame (
Group_1 = c("B","C"),
Group_2 = c("D","A"),
Group_3 = ("E"), stringsAsFactors = FALSE)
Rowise grouping:
df1_grouped<-bind_cols(df1[1:2], map_df(df2, ~rowSums(df1[unique(.x)])))
Now i would like to apply the same grouping to the ID2 column and sum the values in the other columns. My idea was to mutate a another column (e.g. "group", which contains the name of the final group of ID2. After this i can use group_by() and summarise() to calculate the sum for each. However, I can't figure out an automated way to do it
bind_cols(df1_grouped,
#add group label
data.frame(
group = rep(c("Group_2","Group_1","Group_1","Group_2","Group_3"),2))) %>%
#remove temporary label column and make ID a character column
mutate(ID2=group,
ID=as.character(ID))%>%
select(-group) %>%
#summarise
group_by(ID,ID2)%>%
summarise_if(is.numeric, sum, na.rm = TRUE)
This is the final table I need, but I had to manually assign the groups, which is impossible for big datasets
I will offer such a solution
library(tidyverse)
set.seed(1)
df1 <- data.frame(
ID = c(rep(1,5),rep(2,5)),
ID2 = rep(c("A","B","C","D","E"),2),
A = rnorm(10,20,1),
B = rnorm(10,50,1),
C = rnorm(10,10,1),
D = rnorm(10,15,1),
E = rnorm(10,5,1)
)
df2 <- data.frame (
Group_1 = c("B","C"),
Group_2 = c("D","A"),
Group_3 = ("E"), stringsAsFactors = FALSE)
df2 <- df2 %>% pivot_longer(everything())
df1 %>%
pivot_longer(-c(ID, ID2)) %>%
mutate(gr_r = df2$name[match(ID2, table = df2$value)],
gr_c = df2$name[match(name, table = df2$value)]) %>%
arrange(ID, gr_r, gr_c) %>%
pivot_wider(c(ID, gr_r), names_from = gr_c, values_from = value, values_fn = list(value = sum))
Related
Probably a tidyr or a reshape2 question.
I have this type of a data.frame:
df <- data.frame(group = c("A","B","C","D"),
id1 = c("AV14D","BV29",NA,NA),
id2 = c(NA,"BD1",NA,NA),
id3 = c("AJ31","BJ1",NA,NA),
n = c(2,4,NA,NA),
stringsAsFactors = F)
which has a row with 3 IDs and a count for each group.
I want to transform it to a single row data.frame with #groups x 4 (id1,id2,id3,n) columns, where each group has 4 columns: <group>_id1, <group>_id2, <group>_id3, and <group>_n and the corresponding values.
So the resulting data.frame will be:
data.frame(A_id1 = "AV14D",A_id2 = NA,A_id3 = "AJ31",A_n = 2,
B_id1 = "BV29",B_id2 = "BD1",B_id3 = "BJ1",B_n = 4,
C_id1 = NA,C_id2 = NA,C_id3 = NA,C_n = NA,
D_id1 = NA,D_id2 = NA,D_id3 = NA,D_n = NA,
stringsAsFactors = F)
We may use pivot_wider after creating a row index column
library(dplyr)
library(tidyr)
library(stringr)
out1 <- df %>%
mutate(rn = 1) %>%
pivot_wider(names_from = group, values_from = id1:n,
names_glue = "{group}_{.value}") %>%
select(order(str_remove(names(.), "_.*")), -rn) %>%
type.convert(as.is = TRUE)
-checking with OP's output
> all.equal(out, out1, check.attributes = FALSE)
[1] TRUE
My question is about performing a calculation between each pair of groups in a data.frame, I'd like it to be more vectorized.
I have a data.frame that has a consists of the following columns: Location , Sample , Var1, and Var2. I'd like to find the closet match for each Sample for each pair of Locations for both Var1 and Var2.
I can accomplish this for one pair of locations as such:
df0 <- data.frame(Location = rep(c("A", "B", "C"), each =30),
Sample = rep(c(1:30), times =3),
Var1 = sample(1:25, 90, replace =T),
Var2 = sample(1:25, 90, replace=T))
df00 <- data.frame(Location = rep(c("A", "B", "C"), each =30),
Sample = rep(c(31:60), times =3),
Var1 = sample(1:100, 90, replace =T),
Var2 = sample(1:100, 90, replace=T))
df000 <- rbind(df0, df00)
df <- sample_n(df000, 100) # data
dfl <- df %>% gather(VAR, value, 3:4)
df1 <- dfl %>% filter(Location == "A")
df2 <- dfl %>% filter(Location == "B")
df3 <- merge(df1, df2, by = c("VAR"), all.x = TRUE, allow.cartesian=TRUE)
df3 <- df3 %>% mutate(DIFF = abs(value.x-value.y))
result <- df3 %>% group_by(VAR, Sample.x) %>% top_n(-1, DIFF)
I tried other possibilities such as using dplyr::spread but could not avoid the "Error: Duplicate identifiers for rows" or columns half filled with NA.
Is there a more clean and automated way to do this for each possible group pair? I'd like to avoid the manual subset and merge routine for each pair.
One option would be to create the pairwise combination of 'Location' with combn and then do the other steps as in the OP's code
library(tidyverse)
df %>%
# get the unique elements of Location
distinct(Location) %>%
# pull the column as a vector
pull %>%
# it is factor, so convert it to character
as.character %>%
# get the pairwise combinations in a list
combn(m = 2, simplify = FALSE) %>%
# loop through the list with map and do the full_join
# with the long format data df1
map(~ full_join(df1 %>%
filter(Location == first(.x)),
df1 %>%
filter(Location == last(.x)), by = "VAR") %>%
# create a column of absolute difference
mutate(DIFF = abs(value.x - value.y)) %>%
# grouped by VAR, Sample.x
group_by(VAR, Sample.x) %>%
# apply the top_n with wt as DIFF
top_n(-1, DIFF))
Also, as the OP mentioned about automatically picking up instead of doing double filter (not clear about the expected output though)
df %>%
distinct(Location) %>%
pull %>%
as.character %>%
combn(m = 2, simplify = FALSE) %>%
map(~ df1 %>%
# change here i.e. filter both the Locations
filter(Location %in% .x) %>%
# spread it to wide format
spread(Location, value, fill = 0) %>%
# create the DIFF column by taking the differene
mutate(DIFF = abs(!! rlang::sym(first(.x)) -
!! rlang::sym(last(.x)))) %>%
group_by(VAR, Sample) %>%
top_n(-1, DIFF))
I'm Brazilian, sorry about my english!
I would like to know if there is an function implemented in some R package to filter first "n" rows and group the remaining into an "Other" row and summarise the column.
Here is below an example of what I want:
library(tidyverse)
library(plotly)
library(scales)
data("lakers")
x = bind_rows(
lakers %>% count(player) %>% arrange(-n) %>% head(10),
lakers %>% count(player) %>% arrange(-n) %>% slice(11:n()) %>%
summarise(player = "Others", n = sum(n))) %>%
filter(!player == "") %>%
mutate(
player = factor(player, levels = rev(.$player)))
ggplot(x, aes(x=player, y=n))+
geom_col(fill = "DodgerBlue1", col = "DodgerBlue3")+
coord_flip()+
geom_text(aes(y=n, label = comma(n)),hjust = -.2)+
scale_y_continuous(limits = c(0, max( x$n*1.1 )))+
theme_minimal()
I need to create an ggplot like that. So I have a big query using dplyr and I don't want to repeat the query every time.
I would like some function like:
head.other(x, rows = 20, fun = sum, name = "Others")
Here is a function that I think will give you what you need:
library(tibble)
library(dplyr)
df <- data.frame(col1 = rnorm(10), col2 = rnorm(10)) # your data frame
n <- 6 # top n rows to keep
myfun <- function(df, n) {
# seperate keep rows and those to aggregate
preserve.df <- df[1:n, ]
summarise.df <- df[(n+1):nrow(df), ]
# create new df in required format
new.df <- bind_rows(preserve.df, sapply(summarise.df, sum))
# add a column to identify the rows and return
rownames(new.df) <- c(paste0("r", 1:n), "Other")
rownames_to_column(new.df)
}
myfun(df, 6)
Sorry to bother with a relatively simple question perhaps.
I have this type of dataframe:
A long list of names in the column "NAME" c(a, b, c, d, e ...) , two potential classes in the column "SURNAME" c(A, B) and a third column containing values.
I want to remove all NAMES for which at least in one of the SURNAME classes I have more than 2 "NA" in the VALUE column.
I wanted to post an example dataset but I am struggling to format it properly
I was trying to use
df <- df %>%
group_by(NAME) %>%
group_by(SURNAME) %>%
filter(!is.na(VALUE)) %>%
filter(length(VALUE)>=3)
it does not throw an error but I have the impression that something is wrong. Any suggestion? Many thanks
Let's create a dataset to work with:
set.seed(1234)
df <- data.frame(
name = sample(x=letters, size=1e3, replace=TRUE),
surname = sample(x=c("A", "B"), size=1e3, replace=TRUE),
value = sample(x=c(1:10*10,NA), size=1e3, replace=TRUE),
stringsAsFactors = FALSE
)
Here's how to do it with Base R:
# count NAs by name-surname combos (na.action arg is important!)
agg <- aggregate(value ~ name + surname, data=df, FUN=function(x) sum(is.na(x)), na.action=NULL)
# rename is count of NAs column
names(agg)[3] <- "number_of_na"
#add count of NAs back to original data
df <- merge(df, agg, by=c("name", "surname"))
# subset the original data
result <- df[df$number_of_na < 3, ]
Here's how to do it with data.table:
library(data.table)
dt <- as.data.table(df)
dt[ , number_of_na := sum(is.na(value)), by=.(name, surname)]
result <- dt[number_of_na < 3]
Here's how to do it with dplr/tidyverse:
library(dplyr) # or library(tidyverse)
result <- df %>%
group_by(name, surname) %>%
summarize(number_of_na = sum(is.na(value))) %>%
right_join(df, by=c("name", "surname")) %>%
filter(number_of_na < 3)
After grouping by 'NAME', 'SURNAME', create a column with the number of NA elements in that group and then filter out any 'NAME' that have an 'ind' greater than or equal to 3
df %>%
group_by(NAME, SURNAME) %>%
mutate(ind = sum(is.na(VALUE))) %>%
group_by(NAME) %>%
filter(!any(ind >=3)) %>%
select(-ind)
Or do an anti_join after doing the filtering by 'NAME', 'SURNAME' based on the condition
df %>%
group_by(NAME, SURNAME) %>%
filter(sum(is.na(VALUE))>=3) %>%
ungroup %>%
distinct(NAME) %>%
anti_join(df, .)
data
set.seed(24)
df <- data.frame(NAME = rep(letters[1:5], each = 20),
SURNAME = sample(LETTERS[1:4], 5 * 20, replace = TRUE),
VALUE = sample(c(NA, 1:3), 5 *20, replace = TRUE),
stringsAsFactors = FALSE)
Here is a data structure that I have:
structure(list(UDD_beta = c(1.17136554204268, 0.939587997289016
), UDD_pval = c(0, 0), UDD_R.sq = c(0.749044972637797, 0.516943886705951
), SSX_beta = c(1.05356804780772, 0.927948300464624), SSX_pval = c(0,
0), SSX_R.sq = c(0.60226298037862, 0.629111666509209), SPP_beta = c(0.675765151939885,
0.516425218613404), SPP_pval = c(0, 0), SPP_R.sq = c(0.479849538274406,
0.378266618442121), EEE_beta = c(0.690521022226874, 0.639380962824289
), EEE_pval = c(0, 0), EEE_R.sq = c(0.585610742768951, 0.676073352909597
)), .Names = c("UDD_beta", "UDD_pval", "UDD_R.sq", "SSX_beta",
"SSX_pval", "SSX_R.sq", "SPP_beta", "SPP_pval", "SPP_R.sq",
"EEE_beta", "EEE_pval", "EEE_R.sq"), row.names = c("DDK", "DDL"
), class = "data.frame")
I want to take R.sq columns and for each row find the max and the column name of the max value. Then take corresponding beta. Expected output:
Name Value
DDK UDD 1.17136554204268
DDL EEE 0.690521022226874
Sorry, the second expected value should be 0.639380962824289.
We could use max.col. Subset the columns of interest i.e. columns that have 'R.sq' using the grep, then get the column index of max value with max.col. Use that to get the column names and also the values that correspond to a particular row (row/column indexing)
i1 <- grep("R.sq", names(df1))
i2 <- max.col(df1[i1], "first")
i3 <- grep("beta", names(df1))
res <- data.frame(Names = sub("_.*", "", names(df1)[i1][i2]),
Value = df1[i3][cbind(1:nrow(df1), i2)])
row.names(res) <- row.names(df1)
sub_data <- data[grep("R.sq", colnames(data))]
colnames(sub_data) <- gsub("_R.sq", "", colnames(sub_data))
sub_data$Name <- NA
sub_data$Value <- NA
for (i in 1:nrow(sub_data)){
sub_data$Name[i] <- names(sub_data[i,])[which.max(apply(sub_data[i,], 2, max))]
sub_data$Value[i] <- max(data[grep(paste0(sub_data$Name[i], "_beta"), colnames(data))], na.rm=T)
}
sub_data[c("Name", "Value")]
# Name Value
#DDK UDD 1.171366
#DDL EEE 0.690521
You can use a tidyverse approach via gathering your df to long and filtering both R.sq vars and max value, i.e.
library(tidyverse)
df %>%
rownames_to_column('ID') %>%
gather(var, val, -ID) %>%
filter(grepl('R.sq|beta', var)) %>%
group_by(ID) %>%
mutate(max1=as.integer(val == max(val[grepl('R.sq', var)]))) %>%
group_by(ID, grp = sub('_.*', '', var)) %>%
filter(!all(max1 == 0) & grepl('beta', var)) %>%
ungroup() %>% select(-c(max1, grp))
which gives,
# A tibble: 2 x 3
ID var val
<chr> <chr> <dbl>
1 DDK UDD_beta 1.171366
2 DDL EEE_beta 0.639381
# Need ID for all possible betas and Rsq
ID <- gsub("_R.sq", "", grep("_R.sq$", names(INPUT), value = TRUE))
dummy <- function(x) {
# Find out which Rsq is largest
i <- ID[which.max(x[paste0(ID, "_R.sq")])]
# Extract beta for largest Rsq
data.frame(Name = i, Value = x[paste0(i, "_beta")])
}
do.call("rbind", apply(INPUT, 1, dummy))