问题
My data looks as follows:
DF <- structure(list(No_Adjusted_Gross_Income = c(183454, 241199, 249506
), NoR_from_1_to_5000 = c(1035373, 4272260, 1124098), NoR_from_5000_to_10000 = c(319540,
4826042, 1959866)), row.names = c(NA, -3L), class = c("data.table",
"data.frame"))
val <- c(2500.5, 7500)
vn <- c("AGI_from_1_to_5000", "AGI_from_5000_to_10000")
No_Adjusted_Gross_Income NoR_from_1_to_5000 NoR_from_5000_to_10000
1: 183454 1035373 319540
2: 241199 4272260 4826042
3: 249506 1124098 1959866
I would like to create new columns, based on column 2 and 3, multiplied with the values from val, using the names from vn. I tried to do it as follows:
DF[,2:3] <- lapply(DF[,2:3], vn := val*DF[,2:3])
But this does not work..
Desired output:
DF <- setDT(DF)[, vn[1]:=val[1]*DF[,2]]
DF <- setDT(DF)[, vn[2]:=val[2]*DF[,3]]
DFout <- structure(list(No_Adjusted_Gross_Income = c(183454, 241199, 249506
), NoR_from_1_to_5000 = c(1035373, 4272260, 1124098), NoR_from_5000_to_10000 = c(319540,
4826042, 1959866), AGI_from_1_to_5000 = c(2588950186.5, 10682786130,
2810807049), AGI_from_5000_to_10000 = c(2396550000, 36195315000,
14698995000)), row.names = c(NA, -3L), class = c("data.table",
"data.frame"))
No_Adjusted_Gross_Income NoR_from_1_to_5000 NoR_from_5000_to_10000 AGI_from_1_to_5000 AGI_from_5000_to_10000
1: 183454 1035373 319540 2588950187 2396550000
2: 241199 4272260 4826042 10682786130 36195315000
3: 249506 1124098 1959866 2810807049 14698995000
回答1:
This should work.. lapply() is not needed
library( data.table )
setDT( DF )
DF[, (var) := as.data.table ( t( t( DF[, 2:3] ) * val ) ) ][]
# No_Adjusted_Gross_Income NoR_from_1_to_5000 NoR_from_5000_to_10000 AGI_from_1_to_5000 AGI_from_5000_to_10000
# 1: 183454 1035373 319540 2588950187 2396550000
# 2: 241199 4272260 4826042 10682786130 36195315000
# 3: 249506 1124098 1959866 2810807049 14698995000
回答2:
you can use apply to get your values, then use cbind if you want to combine with your original DF
t(apply(DF[,2:3],1, function(x) x*val ))
NoR_from_1_to_5000 NoR_from_5000_to_10000
[1,] 2588950187 2396550000
[2,] 10682786130 36195315000
[3,] 2810807049 14698995000
回答3:
The OP has asked in a comment for a grouping variable.
Although the accepted answer apparently does what the OP initially has asked for I would like to suggest a completey different approach where the data is stored and processed in tidy (long) format. IMHO, processing data in long format is much more straightforward and flexible (which includes aggregation & grouping).
For this, the dataset is reshaped from wide, Excel-style format to long, SQL-style format by
library(data.table)
col <- "NoR"
long <- melt(DF, measure.vars = patterns(col), value.name = col, variable.name = "range")
long[, range := stringr::str_remove(range, paste0(col, "_"))]
long
No_Adjusted_Gross_Income range NoR 1: 183454 from_1_to_5000 1035373 2: 241199 from_1_to_5000 4272260 3: 249506 from_1_to_5000 1124098 4: 183454 from_5000_to_10000 319540 5: 241199 from_5000_to_10000 4826042 6: 249506 from_5000_to_10000 1959866
In tidy (long) format there is one row for each observation and one column for each variable (see Chapter 12.2 of Hadley Wickham's book R for Data Science.
The vector of multipliers val also needs to be reshaped from wide to long format:
valDF <- long[, .(range = unique(range), val)]
valDF
range val 1: from_1_to_5000 2500.5 2: from_5000_to_10000 7500.0
Now, valDF is also in tidy format as there is one row for each range.
Finally, we can add a new column AGI to DF by an update join:
long[valDF, on = "range", AGI := val * NoR][]
No_Adjusted_Gross_Income range NoR AGI 1: 183454 from_1_to_5000 1035373 2588950187 2: 241199 from_1_to_5000 4272260 10682786130 3: 249506 from_1_to_5000 1124098 2810807049 4: 183454 from_5000_to_10000 319540 2396550000 5: 241199 from_5000_to_10000 4826042 36195315000 6: 249506 from_5000_to_10000 1959866 14698995000
If required for presentation, the dataset can be reshaped back from long to wide format:
dcast(long, No_Adjusted_Gross_Income ~ range, value.var = c("NoR", "AGI"))
No_Adjusted_Gross_Income NoR_from_1_to_5000 NoR_from_5000_to_10000 AGI_from_1_to_5000 AGI_from_5000_to_10000 1: 183454 1035373 319540 2588950187 2396550000 2: 241199 4272260 4826042 10682786130 36195315000 3: 249506 1124098 1959866 2810807049 14698995000
which reproduces OP's expected result. Note that the variable names vn are created automagically.
Aggregation and grouping can be performed while reshaping
dcast(long, No_Adjusted_Gross_Income ~ range, sum, value.var = c("NoR", "AGI"))
No_Adjusted_Gross_Income NoR_from_1_to_5000 NoR_from_5000_to_10000 AGI_from_1_to_5000 AGI_from_5000_to_10000 1: 183454 1035373 319540 2588950187 2396550000 2: 241199 4272260 4826042 10682786130 36195315000 3: 249506 1124098 1959866 2810807049 14698995000
or
dcast(long, No_Adjusted_Gross_Income ~ ., sum, value.var = c("NoR", "AGI"))
No_Adjusted_Gross_Income NoR AGI 1: 183454 1354913 4985500187 2: 241199 9098302 46878101130 3: 249506 3083964 17509802049
Alternatively, aggregation & grouping can be performed in long format:
long[, lapply(.SD, sum), .SDcols = c("NoR", "AGI"), by = No_Adjusted_Gross_Income]
No_Adjusted_Gross_Income NoR AGI 1: 183454 1354913 4985500187 2: 241199 9098302 46878101130 3: 249506 3083964 17509802049
来源：https://stackoverflow.com/questions/62004219/using-lapply-to-create-new-columns-based-on-old-columns