Count strings with a certain condition - r

Question

I have the following dataset
#mydata
Factors Transactions
a,c 2
b 0
c 0
d,a 0
a 1
a 0
b 1
I'd like to count those factors who had transactions.For example, we had two times "a" with transaction. I can write a code to give me my desirable outcome for each variable separately. The following is for "a".
nrow (subset (mydata,mydata$Transaction > 0 & length(mydata[grep("a", mydata$Factors),] )> 0))
But I have too much variables and do not want to repeat a code for all of them. I would think there should be a way to write a code to give me the results for all of the variables. I wish to have the following out put:
#Output
a 2
b 1
c 1
d 0

Answer 1

Equivalent data.table option:
library(data.table)
setDT(df)[, .(Factors = unlist(strsplit(as.character(Factors), ","))),
by = Transactions][,.(Transactions = sum(Transactions > 0)), by = Factors]
# Factors Transactions
#1: a 2
#2: c 1
#3: b 1
#4: d 0

Answer 2

You could create a table using the unique values of the Factor column as the levels. Consider df to be your data set.
s <- strsplit(as.character(df$Factors), ",", fixed = TRUE)
table(factor(unlist(s[df$Transactions > 0]), levels = unique(unlist(s))))
#
# a c b d
# 2 1 1 0
Wrap in as.data.frame() for data frame output.
with(df, {
s <- strsplit(as.character(Factors), ",", fixed = TRUE)
f <- factor(unlist(s[Transactions > 0]), levels = unique(unlist(s)))
as.data.frame(table(Factors = f))
})
# Factors Freq
# 1 a 2
# 2 c 1
# 3 b 1
# 4 d 0

Answer 3

With tidyverse packages, assuming your data is strings/factors and numbers,
library(tidyr)
library(dplyr)
# separate factors with two elements
df %>% separate_rows(Factors) %>%
# set grouping for aggregation
group_by(Factors) %>%
# for each group, count how many transactions are greater than 0
summarise(Transactions = sum(Transactions > 0))
## # A tibble: 4 x 2
## Factors Transactions
## <chr> <int>
## 1 a 2
## 2 b 1
## 3 c 1
## 4 d 0
You could also avoid dplyr by using xtabs, though some cleaning is necessary to get to the same arrangement:
library(tidyr)
df %>% separate_rows(Factors) %>%
xtabs(Transactions > 0 ~ Factors, data = .) %>%
as.data.frame() %>%
setNames(names(df))
## Factors Transactions
## 1 a 2
## 2 b 1
## 3 c 1
## 4 d 0
A full base R equivalent:
df2 <- do.call(rbind,
Map(function(f, t){data.frame(Factors = strsplit(as.character(f), ',')[[1]],
Transactions = t)},
df$Factors, df$Transactions))
df3 <- as.data.frame(xtabs(Transactions > 0 ~ Factors, data = df2))
names(df3) <- names(df)
df3
## Factors Transactions
## 1 a 2
## 2 b 1
## 3 c 1
## 4 d 0

Answer 4

We can use cSplit from splitstackshape to split the 'Factors' into 'long' format and grouped by 'Factors' we get the sum of logical column ('Transactions > 0`).
library(splitstackshape)
cSplit(df1, "Factors", ",", "long")[, .(Transactions=sum(Transactions > 0)),.(Factors)]
# Factors Transactions
#1: a 2
#2: c 1
#3: b 1
#4: d 0
Or using base R
with(df1, table(factor(unlist(strsplit(Factors[Transactions>0], ",")),
levels = letters[1:4]) ))
# a b c d
# 2 1 1 0
data
df1 <- structure(list(Factors = c("a,c", "b", "c", "d,a", "a", "a",
"b"), Transactions = c(2L, 0L, 0L, 0L, 1L, 0L, 1L)), .Names = c("Factors",
"Transactions"), class = "data.frame", row.names = c(NA, -7L))