Add new columns and insert values in columns based on value in another column
Question
I have a R dataframe data1 as below:
prodID storeID Term Exit
1 1001 5 0
1 1002 4 1
1 1003 3 1
1 1004 5 0
2 1001 4 1
2 1002 3 1
2 1003 5 0
3 1001 4 1
3 1002 3 1
3 1003 5 0
4 1001 4 1
4 1002 3 1
5 1001 5 0
5 1002 4 1
5 1003 3 1
This is of course highly simplified format of my real data which runs into around 3 million rows. I have to do the following:
- Based on max value in
Termcolumn, insert that many columns indata1withNAvalues. Column names should beWeek1,Week2,Week3, etc - For each row, fill the new columns with 0, 1 or
NAusing these rules: 1)IfTermis 5, then insert 0 inWeek1,Week2, uptoWeek4and 1 inWeek52) IfTermis 4 then insert 0 inWeek1,Week2, andWeek3, 1 inWeek4and keepNAinWeek5. And so on....
The final output should look like:
prodID storeID Term Exit Week1 Week2 Week3 Week4 Week5
1 1001 5 0 0 0 0 0 1
1 1002 4 1 0 0 0 1 NA
1 1003 3 1 0 0 1 NA NA
1 1004 5 0 0 0 0 0 1
2 1001 4 1 0 0 0 1 NA
2 1002 3 1 0 0 1 NA NA
2 1003 5 0 0 0 0 0 1
3 1001 4 1 0 0 0 1 NA
3 1002 3 1 0 0 1 NA NA
3 1003 5 0 0 0 0 0 1
4 1001 4 1 0 0 0 1 NA
4 1002 3 1 0 0 1 NA NA
5 1001 5 0 0 0 0 0 1
5 1002 4 1 0 0 0 1 NA
5 1003 3 1 0 0 1 NA NA
This is what I tried:
variant <- c("Week1","Week2","Week3","Week4","Week5")
data1[variant] <- NA
for (i in 1:length(data1$prodID)){
data1$Week1 <- ifelse(data1$Term==1,1,0)
data1$Week2 <- ifelse(data1$Term==2,1,0)
data1$Week3 <- ifelse(data1$Term==3,1,0)
data1$Week4 <- ifelse(data1$Term==4,1,0)
data1$Week5 <- ifelse(data1$Term==5,1,0)
}
This doesn't help me populate NA in the required cells. I would like to retain the NA values because I am going to do a wide to long data transformation on the data frame later on. And I know the above approach is not feasible in my huge dataset. Any suggestions are most welcome.
3 answers
solution1
3 ACCPTED 2018-06-25 18:09:07
Here is one idea. We can create the content you need and then split the columns.
library(dplyr)
library(data.table)
library(splitstackshape)
dat2 <- dat %>%
mutate(Week = case_when(
Term == 5 ~"0,0,0,0,1",
Term == 4 ~"0,0,0,1,NA",
Term == 3 ~"0,0,1,NA,NA",
Term == 2 ~"0,1,NA,NA,NA",
Term == 1 ~"1,NA,NA,NA,NA"
)) %>%
cSplit(splitCols = "Week")
dat2
# prodID storeID Term Exit Week_1 Week_2 Week_3 Week_4 Week_5
# 1: 1 1001 5 0 0 0 0 0 1
# 2: 1 1002 4 1 0 0 0 1 NA
# 3: 1 1003 3 1 0 0 1 NA NA
# 4: 1 1004 5 0 0 0 0 0 1
# 5: 2 1001 4 1 0 0 0 1 NA
# 6: 2 1002 3 1 0 0 1 NA NA
# 7: 2 1003 5 0 0 0 0 0 1
# 8: 3 1001 4 1 0 0 0 1 NA
# 9: 3 1002 3 1 0 0 1 NA NA
# 10: 3 1003 5 0 0 0 0 0 1
# 11: 4 1001 4 1 0 0 0 1 NA
# 12: 4 1002 3 1 0 0 1 NA NA
# 13: 5 1001 5 0 0 0 0 0 1
# 14: 5 1002 4 1 0 0 0 1 NA
# 15: 5 1003 3 1 0 0 1 NA NA
Or use this tidyverse method. I like this one better than my previous one because this method does not require manually typing the week values.
library(dplyr)
library(tidyr)
library(purrr)
dat2 <- dat %>%
mutate(Week = map2(1, Term, `:`)) %>%
unnest() %>%
group_by(prodID, Term) %>%
mutate(Week_Value = as.integer(Week == max(Week)),
Week = paste0("Week", Week)) %>%
spread(Week, Week_Value) %>%
ungroup()
dat2
# # A tibble: 15 x 9
# prodID storeID Term Exit Week1 Week2 Week3 Week4 Week5
# <int> <int> <int> <int> <int> <int> <int> <int> <int>
# 1 1 1001 5 0 0 0 0 0 1
# 2 1 1002 4 1 0 0 0 1 NA
# 3 1 1003 3 1 0 0 1 NA NA
# 4 1 1004 5 0 0 0 0 0 1
# 5 2 1001 4 1 0 0 0 1 NA
# 6 2 1002 3 1 0 0 1 NA NA
# 7 2 1003 5 0 0 0 0 0 1
# 8 3 1001 4 1 0 0 0 1 NA
# 9 3 1002 3 1 0 0 1 NA NA
# 10 3 1003 5 0 0 0 0 0 1
# 11 4 1001 4 1 0 0 0 1 NA
# 12 4 1002 3 1 0 0 1 NA NA
# 13 5 1001 5 0 0 0 0 0 1
# 14 5 1002 4 1 0 0 0 1 NA
# 15 5 1003 3 1 0 0 1 NA NA
UPDATE
We can use str_pad from the stringr package to pad 0 before spread the week column to sort the column name.
library(tidyverse)
dat2 <- dat %>%
mutate(Week = map2(1, Term, `:`)) %>%
unnest() %>%
group_by(prodID, Term) %>%
mutate(Week_Value = as.integer(Week == max(Week)),
Week = paste0("Week", str_pad(Week, width = 3, pad = "0"))) %>%
spread(Week, Week_Value) %>%
ungroup()
dat2
# # A tibble: 15 x 9
# prodID storeID Term Exit Week001 Week002 Week003 Week004 Week005
# <int> <int> <int> <int> <int> <int> <int> <int> <int>
# 1 1 1001 5 0 0 0 0 0 1
# 2 1 1002 4 1 0 0 0 1 NA
# 3 1 1003 3 1 0 0 1 NA NA
# 4 1 1004 5 0 0 0 0 0 1
# 5 2 1001 4 1 0 0 0 1 NA
# 6 2 1002 3 1 0 0 1 NA NA
# 7 2 1003 5 0 0 0 0 0 1
# 8 3 1001 4 1 0 0 0 1 NA
# 9 3 1002 3 1 0 0 1 NA NA
# 10 3 1003 5 0 0 0 0 0 1
# 11 4 1001 4 1 0 0 0 1 NA
# 12 4 1002 3 1 0 0 1 NA NA
# 13 5 1001 5 0 0 0 0 0 1
# 14 5 1002 4 1 0 0 0 1 NA
# 15 5 1003 3 1 0 0 1 NA NA
DATA
dat <- read.table(text = "prodID storeID Term Exit
1 1001 5 0
1 1002 4 1
1 1003 3 1
1 1004 5 0
2 1001 4 1
2 1002 3 1
2 1003 5 0
3 1001 4 1
3 1002 3 1
3 1003 5 0
4 1001 4 1
4 1002 3 1
5 1001 5 0
5 1002 4 1
5 1003 3 1",
header = TRUE)
solution2
2 2018-06-25 18:27:35
Here is one option with base R where we loop through the 'Term', tabulate to get a 0s and 1 for each element, append NA at the end with length<- and rbind the list elements to create the columns of interest
dat[paste0("Week", 1:5)] <- do.call(rbind, lapply(dat$Term,
function(x) `length<-`(tabulate(x), max(dat$Term))))
dat
# prodID storeID Term Exit Week1 Week2 Week3 Week4 Week5
#1 1 1001 5 0 0 0 0 0 1
#2 1 1002 4 1 0 0 0 1 NA
#3 1 1003 3 1 0 0 1 NA NA
#4 1 1004 5 0 0 0 0 0 1
#5 2 1001 4 1 0 0 0 1 NA
#6 2 1002 3 1 0 0 1 NA NA
#7 2 1003 5 0 0 0 0 0 1
#8 3 1001 4 1 0 0 0 1 NA
#9 3 1002 3 1 0 0 1 NA NA
#10 3 1003 5 0 0 0 0 0 1
#11 4 1001 4 1 0 0 0 1 NA
#12 4 1002 3 1 0 0 1 NA NA
#13 5 1001 5 0 0 0 0 0 1
#14 5 1002 4 1 0 0 0 1 NA
#15 5 1003 3 1 0 0 1 NA NA
Or using the similar approach with tidyverse
library(tidyverse)
dat %>%
mutate(Week = map(Term, ~
tabulate(.x) %>%
as.list %>%
set_names(paste0("Week", seq_along(.))) %>%
as_tibble)) %>%
unnest
# prodID storeID Term Exit Week1 Week2 Week3 Week4 Week5
#1 1 1001 5 0 0 0 0 0 1
#2 1 1002 4 1 0 0 0 1 NA
#3 1 1003 3 1 0 0 1 NA NA
#4 1 1004 5 0 0 0 0 0 1
#5 2 1001 4 1 0 0 0 1 NA
#6 2 1002 3 1 0 0 1 NA NA
#7 2 1003 5 0 0 0 0 0 1
#8 3 1001 4 1 0 0 0 1 NA
#9 3 1002 3 1 0 0 1 NA NA
#10 3 1003 5 0 0 0 0 0 1
#11 4 1001 4 1 0 0 0 1 NA
#12 4 1002 3 1 0 0 1 NA NA
#13 5 1001 5 0 0 0 0 0 1
#14 5 1002 4 1 0 0 0 1 NA
#15 5 1003 3 1 0 0 1 NA NA
solution3
2 2018-06-25 18:31:20
An option using dplyr::mutate_at and case_when can be based on finding subscript integer in column name using quo_name(quo(.)) and then checking if column number is more/equal/less than value of Term .
# First add additional columns based on maximum value of Term
df[,paste("Week", 1:max(df$Term), sep="")] <- NA
library(dplyr)
df %>% mutate_at(vars(starts_with("Week")), funs(case_when(
as.integer(sub(".*(\\d+)","\\1",quo_name(quo(.)))) < Term ~ 0L,
as.integer(sub(".*(\\d+)","\\1",quo_name(quo(.)))) == Term ~ 1L,
TRUE ~ NA_integer_
)))
# prodID storeID Term Exit Week1 Week2 Week3 Week4 Week5
# 1 1 1001 5 0 0 0 0 0 1
# 2 1 1002 4 1 0 0 0 1 NA
# 3 1 1003 3 1 0 0 1 NA NA
# 4 1 1004 5 0 0 0 0 0 1
# 5 2 1001 4 1 0 0 0 1 NA
# 6 2 1002 3 1 0 0 1 NA NA
# 7 2 1003 5 0 0 0 0 0 1
# 8 3 1001 4 1 0 0 0 1 NA
# 9 3 1002 3 1 0 0 1 NA NA
# 10 3 1003 5 0 0 0 0 0 1
# 11 4 1001 4 1 0 0 0 1 NA
# 12 4 1002 3 1 0 0 1 NA NA
# 13 5 1001 5 0 0 0 0 0 1
# 14 5 1002 4 1 0 0 0 1 NA
# 15 5 1003 3 1 0 0 1 NA NA
Data:
df <- read.table(text="
prodID storeID Term Exit
1 1001 5 0
1 1002 4 1
1 1003 3 1
1 1004 5 0
2 1001 4 1
2 1002 3 1
2 1003 5 0
3 1001 4 1
3 1002 3 1
3 1003 5 0
4 1001 4 1
4 1002 3 1
5 1001 5 0
5 1002 4 1
5 1003 3 1",
header = TRUE, stringsAsFactors = FALSE)
The technical post webpages of this site follow the CC BY-SA 4.0 protocol. If you need to reprint, please indicate the site URL or the original address.Any question please contact:yoyou2525@163.com.
Related Question
Add new columns based on values in other columns
New DF column based on the value of another column. Possible new values obtained by checking the name of pre-existing columns with highest value
Add a new column to a dataframe based on multiple columns from another dataframe
Collapse a dataframe, creating new columns with name being the unique values of another column, and value based on the value of another column? In R
Add values to a new column based on math calculations on three columns R
Add new data.frame column based on values in other columns
New column value based on matching previous values in different columns
Attribute value to new column based on values in similarly called columns
Count or sum the values of 2 columns based on the value of another column in the same dataframe
Add new columns to data frame for each unique value in another column
Related Tags