סכומי ביניים

[ronen]

I have a big data base

I want to do the following partial sums:

 

X=matrix(1:64,8,8)

y=rep(1:2,c(4,4))

psum=by(X,y,apply,2,sum)

print(psum)

 

Is there a better/quicker way to do it?

Thank you

Ronen

 

[Jonathan Rosenblatt]

What exactly is the problem?

Long runtime? Crashing? Opaque code?

Ronen

 

--
You received this message because you are subscribed to the Google Groups "Israel R User Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email to israel-r-user-g...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

--
Jonathan Rosenblatt
www.john-ros.com

[ronen]

no problem just wondered if this is how it should be done

to optimize my work

Thank you!

 

On Wednesday, August 27, 2014 10:44:42 AM UTC+3, Jonathan Rosenblatt wrote:

What exactly is the problem?

Long runtime? Crashing? Opaque code?

On Wed, Aug 27, 2014 at 10:11 AM, ronen <rfl...@gmail.com> wrote:

I have a big data base

I want to do the following partial sums:

 

X=matrix(1:64,8,8)

y=rep(1:2,c(4,4))

psum=by(X,y,apply,2,sum)

print(psum)

 

Is there a better/quicker way to do it?

Thank you

Ronen

 

--
You received this message because you are subscribed to the Google Groups "Israel R User Group" group.

To unsubscribe from this group and stop receiving emails from it, send an email to israel-r-user-group+unsub...@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
Jonathan Rosenblatt
www.john-ros.com

[Jonathan Rosenblatt]

(a) Readability could be improved. Recall that future-self will have to understand the code. I would loop over categories of y as you will lose nothing in efficiency (by() simply wraps around tapply() which is nothing but a loop), but will gain on readability. 

(b) I think that rowSums and colSums are slightly more efficiency that apply(X,2,sum).

(c) The dplyr package is super efficient for data manipulations. You might get good performance by compounding dplyr::group_by and dplyr::summarise. It is actually a package worth knowing anyhow....

To unsubscribe from this group and stop receiving emails from it, send an email to israel-r-user-g...@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
Jonathan Rosenblatt
www.john-ros.com

[ronen]

Thank you !

I will look into this package.

Ronen

 

On Wednesday, August 27, 2014 1:42:15 PM UTC+3, Jonathan Rosenblatt wrote:

(a) Readability could be improved. Recall that future-self will have to understand the code. I would loop over categories of y as you will lose nothing in efficiency (by() simply wraps around tapply() which is nothing but a loop), but will gain on readability. 

(b) I think that rowSums and colSums are slightly more efficiency that apply(X,2,sum).

(c) The dplyr package is super efficient for data manipulations. You might get good performance by compounding dplyr::group_by and dplyr::summarise. It is actually a package worth knowing anyhow....

On Wed, Aug 27, 2014 at 11:20 AM, ronen <rfl...@gmail.com> wrote:

no problem just wondered if this is how it should be done

to optimize my work

Thank you!

 

On Wednesday, August 27, 2014 10:44:42 AM UTC+3, Jonathan Rosenblatt wrote:

What exactly is the problem?

Long runtime? Crashing? Opaque code?

On Wed, Aug 27, 2014 at 10:11 AM, ronen <rfl...@gmail.com> wrote:

I have a big data base

I want to do the following partial sums:

 

X=matrix(1:64,8,8)

y=rep(1:2,c(4,4))

psum=by(X,y,apply,2,sum)

print(psum)

 

Is there a better/quicker way to do it?

Thank you

Ronen

 

--
You received this message because you are subscribed to the Google Groups "Israel R User Group" group.

To unsubscribe from this group and stop receiving emails from it, send an email to israel-r-user-group+unsubscribe...@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

--
Jonathan Rosenblatt
www.john-ros.com

--
You received this message because you are subscribed to the Google Groups "Israel R User Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email to israel-r-user-group+unsub...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

--
Jonathan Rosenblatt
www.john-ros.com