Statistics pack?

[hansaaa]

Hi,

is anyone aware of any GOOD statistics packs/libraries in c++?

If possible, I would need them to be QUICK, as I have a lot of data, and don't want generation of the statistics to be TOO long. And of course, I would like it to be reliable, i.e. to produce reliable information.

The statistics I need to produce, at least, are:

Mean

Standard Error

Median

Mode

Standard Deviation

Sample/Population Variance

Kurtosis

Skewness

Range

Minimum value

Maximum Value

Sum of values

Count of values (how many are there)

Largest (n), i.e. n-th largest value

smallest (n), i.e. n-th smallest value

Confidence Level for any % value


If anyone has ANY suggestions, that would be awesome!

Thanks!

[hansaaa]

It seems that R can be used from C++:

http://stackoverflow.com/questions/746364/best-library-for-statistics-in-c

Is anyone aware of any resources that can explain how to use R from C++?

[KRAkatau]

Well, R is certainly a way to go in statistics. However, I've never used it from C++ (and frankly, didn't use it much in itself - I just know it's good for stats stuff :))

[hansaaa]

Yep, I haven't used it much, so far, myself, either. I have been able to do most of the stuff I have needed, so far, in Excel, or, other Spreadsheet programs, and, I found it A LOT more straight-forward, for what I wanted to do.

But, If, I can use R, basically, programmatically, from within code (in this case, C++), that Does seem like a lot of stuff that can be done "at once", since R seems to be an ok statistics "pack". Again, however, I am not sure, yet, how to use R from C++.

I have found some information. There is some stuff, called "Rcpp", but, from what I can tell, it seems to have a lot of configuration required, and, I was hoping there would be something more straight-forward.

Again, If ANYONE has ANY info on using R, from within C++, or other programming idioms, I am glad for any pointers, or additional help.

Also, if anyone has any info about other "statistics" packs, I am also happy to have that Information!

Thanks!

C

[hansaaa]

To keep this active ;)

I have found:

http://www.stat.harvard.edu/ccr2005/index.html.


Which I can look through. This doesn't REALLY seem to be, "public domain", But, I will look through it anyway, now ;)

Still, if anyone has any other ideas/pointers, etc. I will still be very happy for ANY further input.

I believe, others, who want to do "statistical stuff", can, or COULD, also draw some benefit from this!

Thanks!

C ! :)

[hansaaa]

hmmm.... anyone?

if not, I guess, I have to implement the methods myself :(

[KRAkatau]

Doesn't any of the packages mentioned in the stackoverflow topic you linked suit you? I think implementing yourself is not a good idea - it will just take a lot of time and you need a lot of tests to make sure it's correct.

[hansaaa]

well, I don't like the ideas at stackoverflow that much, as they usually are fairly complicated. I have already started with creating my own implementation, and, at least for the functions I am needing, I THINK I should be able to manage the testing!

However, if anyone has any future suggestions, I would still be very interested in that!

Thanks!

hansaaa :)