How to fill a new dataframe column by conditional summing - r

Question

I would like build a new variable by summing values considering multiple conditions (and an extra one, see below). Here you can see R Code until something like my current issue.
# The raw dataframe
area <- c("A", "A", "B", "A", "C", "B", "A", "B", "A", "C")
varclass <- c("Z1", "Z1", "Z1", "Z2", "Z1", "Z1", "Z2", "Z1", "Z2", "Z2")
count <- c(45, 56, 2, 8, 345, 3, 98, 2, 6, 9)
df1 <- data.frame(area,
varclass,
count,
stringsAsFactors = FALSE)
df1
# See how df1 looks like...
# area varclass count
#1 A Z1 45
#2 A Z1 56
#3 B Z1 2
#4 A Z2 8
#5 C Z1 345
#6 B Z1 3
#7 A Z2 98
#8 B Z1 2
#9 A Z2 6
#10 C Z2 9
# Building the final dataframe
df2 <- data.frame(unique(df1$area),
stringsAsFactors = FALSE)
names(df2)[1] <- "area"
# See how df2 looks like...
# area
#1 A
#2 B
#3 C
# The new variable to build
df2$Z1_sum <- sum(df1[df1$varclass == "Z1" & df1$area == df2$area,]$count)
# doesn't work
# See what I hope
# area Z1_sum
#1 A 101
#2 B 7
#3 C 345
As you can see in the last line, I would like building a new variable, Z1_sum, in the df2 database. Z1_sum is the sum of count from the df1 database where varclass = "Z1" and df1$area meet the value of df2$area current row (in MS Excel, that means using a LC1 or $A2 cell id).
Please, consider the fact that I'm not looking for solutions involving to directly build df2 from df1 by using a group by condition or the dcast function... I only want a formula wich allow me to return correct values in my new column. It's my extra condition. Why? It's because I have next to build other variables with most sophisticated formulas than just a sum. By understanding how make such conditional operations, I hope move on...
Thanks for your help.
Jeff

Answer 1

Do you mean:
df2 <- setNames(
aggregate(
count ~ area,
df1[df1$varclass == "Z1", ],
sum
),
c("area", "Z1_sum")
)
df2
area Z1_sum
1 A 101
2 B 7
3 C 345
or
df2$Z1_sum <- aggregate(count ~ area, df1[df1$varclass == "Z1", ], sum)$count
Edit to address your comment.
Try with:
df2 <- aggregate(
count ~ area + varclass,
df1,
sum
)
that will give you your data in the "long" format:
df2
area varclass count
1 A Z1 101
2 B Z1 7
3 C Z1 345
4 A Z2 112
5 C Z2 9
Now you need to reshape it to the "wide" format using something like:
df2 <- xtabs(count ~ area + varclass, df2)
varclass
area Z1 Z2
A 101 112
B 7 0
C 345 9
or:
df2 <- reshape(df2, idvar = "area", timevar = "varclass", direction = "wide")
area count.Z1 count.Z2
1 A 101 112
2 B 7 NA
3 C 345 9

Answer 2

Just subset based on whichever Z* you want counted in your final sum.
df1Z1 <- df1[df1$varclass %in% c("Z1"), ]
aggregate(count ~ area, data = df1Z1, FUN = sum)
area count
1 A 101
2 B 7
3 C 345

Answer 3

You can get your desired result using dplyr:
library(dplyr)
df2 <- group_by(df1, area) %>%
filter(varclass == "Z1") %>%
summarize(Z1_sum = sum(count)) %>%
df2
#> # A tibble: 3 x 2
#> area Z1_sum
#> <chr> <dbl>
#> 1 A 101
#> 2 B 7
#> 3 C 345
The dplyr verbs should be pretty explanatory, and the %>% is the pipe operating, taking the output from one function and making it the first input to the next. group_by here groups by the column area so when we calculate the sum (in summarize) it's the sum for each area group. The filter subsets the data.