ansaurus

Question

R: Count number of entries in a row based on external criteria

Answer 1

+2 A:

If the data.frame is called m and vector v, simple

rowSums(m<=v)

should do (this works because m is represented by R as a vector glued of following columns, and v will be recycled). Still, first ensure that all dates are POSIXcts or Dates; see this question for info about the conversion itself.

mbq 2010-08-30 09:27:50

or all dates are `Date`'s

Marek 2010-08-30 09:42:34

@Marek thanks, updated.

mbq 2010-08-30 09:44:46

This doesn't work with dates. Underneath your solution there is a conversion to "double", which doesn't work on a dataframe of dates.

Joris Meys 2010-08-30 10:30:44

@Joris It doesn't work with `data.frame`. With `matrix` of `Date`'s (as in `dput` results) everything works.

Marek 2010-08-30 11:08:08

@Marek The matrix is not a matrix of Date's on my computer. It's a matrix of type "character". And then off course the code works, but doesn't give you the correct result. You'll have to explain me how to get a matrix of class Date. With as.Date, you end up with a vector. With matrix or as.matrix, you end up with class numeric. See also http://stackoverflow.com/questions/3599851/how-to-transform-a-dataframe-of-characters-to-the-respective-dates

Joris Meys 2010-08-30 11:24:05

@Joris I answer your question about `data.frame`. In case of matrix you have to use two step `Out <- as.Date(Data,format="%d %B %Y");dim(Out)<-dim(Data)`

Marek 2010-08-30 11:35:39

@Marek: OK, thx. After the whole conversion thing, the rowSums solution works.

Joris Meys 2010-08-30 11:39:49

@Joris I'll check it out; for now I'll just point to your question.

mbq 2010-08-30 12:32:25

Answer 2

A:

First thing : You really have to transform everything to Dates, and that can be a bit tricky. I read in the matrix as Data, and the vector as vect. Then :

vect <- as.Date(vect,format="%d/%m/%Y")

# Due to the apart nature of the Date class, the normal apply-solutions 
# don't give the result you're looking for.
Data <- as.data.frame(Data)
for (i in 1:ncol(Data)){
    Data[,i] <- as.Date(Data[,i],format="%d %B %Y")
}
> apply(Data,2,"<=",vect)
      Rep1  Rep2  Rep3  Rep4
[1,] FALSE FALSE FALSE FALSE
[2,]  TRUE  TRUE FALSE FALSE
[3,]  TRUE  TRUE  TRUE FALSE
[4,]  TRUE  TRUE  TRUE FALSE
[5,]  TRUE FALSE FALSE FALSE
[6,]  TRUE  TRUE  TRUE FALSE

> rowSums(apply(Data,2,"<=",vect))
[1] 0 2 3 3 1 3

Joris Meys 2010-08-30 10:25:14

For people with different than English locale I recommend look at example section in `strptime` how to deal with names of months (cause `%B` transform month name in local language).

Marek 2010-08-30 11:12:38

Thanks Joris, this did the trick :)

Thomas Jensen 2010-08-31 10:49:34

ansaurus

tags:

views:

answers:

R: Count number of entries in a row based on external criteria

related questions