Correlation coefficient

classic Classic list List threaded Threaded
3 messages Options
Reply | Threaded
Open this post in threaded view
|

Correlation coefficient

Bobic, Tamara
Hi all,

I've recently installed this library to help me with some basic statistics in my project.
I wanted to calculate the correlation coefficient for 2 arrays, which I see I can do with "PearsonsCorrelation".
However, I realized that I should be using a special variant of Pearson if one of the variables is nominal - Point Biserial correlation coefficient.

Could you tell me if this is somehow supported in the library and if not, if there is a similar Java library which can calculate the Point Biserial coefficient.

Thanks for your help!

Best regards,
--
Tamara Bobic,
PhD Student

Hasso-Plattner-Institut für Softwaresystemtechnik GmbH
Prof.-Dr.-Helmert-Str. 2-3
D-14482 Potsdam
Germany

Amtsgericht Potsdam, HRB 12184
Geschäftsführung: Prof. Dr. Christoph Meinel

Phone:   +49 (0)331-5509-569
Fax:       +49 (0)331-5509-325


Email: [hidden email]
Web:   http://www.hpi.de
Reply | Threaded
Open this post in threaded view
|

Re: Correlation coefficient

Ted Dunning
On Mon, Aug 18, 2014 at 1:15 AM, Bobic, Tamara <[hidden email]> wrote:

> However, I realized that I should be using a special variant of Pearson if
> one of the variables is nominal - Point Biserial correlation coefficient.
>

Doesn't this just work if you build a surrogate for the dichotomous
variable with values 0 and 1?

(I would think that there are generally better options for this like
Kolmogorov-Smirnov tests)
Reply | Threaded
Open this post in threaded view
|

RE: Correlation coefficient

Bobic, Tamara
Hi Ted,

thanks for the responce, I already tried that (assigning 1 and 0 instead of true/false), I just was not sure if it's the right way to go.

Best,
Tamara

________________________________________
From: Ted Dunning [[hidden email]]
Sent: Monday, August 18, 2014 6:22 PM
To: Commons Users List
Subject: Re: Correlation coefficient

On Mon, Aug 18, 2014 at 1:15 AM, Bobic, Tamara <[hidden email]> wrote:

> However, I realized that I should be using a special variant of Pearson if
> one of the variables is nominal - Point Biserial correlation coefficient.
>

Doesn't this just work if you build a surrogate for the dichotomous
variable with values 0 and 1?

(I would think that there are generally better options for this like
Kolmogorov-Smirnov tests)

---------------------------------------------------------------------
To unsubscribe, e-mail: [hidden email]
For additional commands, e-mail: [hidden email]