Create a vector in a LOOP with more conditions - r

Question

I have a problem a little complicate. I have two database BD and BD2. For every row in BD I want to search in entire BD2 and obtain some info: Sum, Mean, Sd etc.
With Sum I succeeded to make it work, but for Mean, Sd, Median I want to obtain the vector and after that apply these functions.
To be clearer, here is the code. For sum it worked. But I want now to save in a vector the values that met the conditions and after calculate Mean,Sd,Median. I tried to make the vector out of the base BD but nothing work or is something trickie and I can't figure it out.
for (i in 1:dim(BD)[1])
{
for (j in 1:dim(BD2)[1])
{
if((BD$Start.Date[i]<BD2$X_TIMESTAMP[j]) && (BD2$X_TIMESTAMP[j]<BD$End.Date[i]) && (BD$Linea[i]==BD2$Linea[j]))
{
vl = BD2$X_VALUE[j]
BD$vec[i] = paste(BD$vec[i],vl,sep="")
#vect = as.numeric(BD$vec[i])
BD$Sum[i] = BD$Sum[i]+ BD2$X_VALUE[j]
#BD$Average[i] = mean(vect)
}
}
}
About data I attached a photo.This is for BD.
And this is for BD2:
I tried to create vec before the for sentence. But still it doesn't work.
The error is always the same: Error in $<-.data.frame(*tmp*, "vec", value = list(NULL, NULL, NULL, : replacement has 47 rows, data has 530

Answer 1

You may be able to use a combination of dplyr and purrr to get your summaries. You could split() your data.frame BD into a list of rows. Then you can use purrr:map() to work with each element of the list (i.e. row in your data) to find matches in BD2 and calculate some statistics. After you have your stats, you can recombine back into a single data.frame with bind_rows():
library(dplyr)
library(purrr)
# create a small data example
BD <- data.frame(Start.Date = 1:5,
End.Date = 2:6,
Linea = c(1, 1, 3, 1, 1))
BD2 <- data.frame(X_TIMESTAMP = c(1.5, 1.5, 2.5, 3.5, 4.5),
X_VALUE = c(31, 4, 60, 20, -11),
Linea = c(1,1,1,1,1))
BD %>%
split(seq(nrow(.))) %>%
map(function(BD_row_i) {
# find the timestamps that overlap in BD2
BD2_matches <- BD2 %>%
filter((BD_row_i$Start.Date < X_TIMESTAMP) &
(BD_row_i$End.Date > X_TIMESTAMP) &
(BD_row_i$Linea == Linea))
if (nrow(BD2_matches) > 0) {
# calculate some stats based on these matches
BD2_matches <- BD2_matches %>%
summarize(Sum = sum(X_VALUE),
Average = mean(X_VALUE),
Median = median(X_VALUE),
Sd = sd(X_VALUE))
} else {
# if no matches, return NAs (blank)
BD2_matches <- BD2_matches %>%
summarize(Sum = NA,
Average = NA,
Median = NA,
Sd = NA)
}
# add these statistics back to your BD_row_i data.frame
BD_row_i <- cbind.data.frame(BD_row_i, BD2_matches)
# return the updated data frame
return(BD_row_i)
}) %>%
bind_rows()
Start.Date End.Date Linea Sum Average Median Sd
1 1 2 1 35 17.5 17.5 19.09188
2 2 3 1 60 60.0 60.0 NA
3 3 4 3 NA NA NA NA
4 4 5 1 -11 -11.0 -11.0 NA
5 5 6 1 NA NA NA NA