comp.lang.idl-pvwave archive
Messages from Usenet group comp.lang.idl-pvwave, compiled by Paulo Penteado

Home » Public Forums » archive » regression with 95% confidence interval
Show: Today's Messages :: Show Polls :: Message Navigator
E-mail to friend 
Return to the default flat view Create a new topic Submit Reply
Re: regression with 95% confidence interval [message #85279 is a reply to message #85275] Mon, 22 July 2013 08:11 Go to previous messageGo to previous message
Fabzi is currently offline  Fabzi
Messages: 305
Registered: July 2010
Senior Member
On 07/22/2013 03:35 PM, Kenneth Bowman wrote:
I e-mailed you a routine that will compute confidence limits on the
> slope and intercept.

I am interested in your method too ;-). Could you mail it to me too?

Until now I did following to test significance (on the F-test):

; Prepare data
r = RANDOMNUMBERGENERATOR()
nx = 12
x = FINDGEN(nx)+1

b = 2. ; intercept
sl = 0.5 ; slope
ra = r->GetRandomNumbers(nx) ; some random error
y = sl * x + b + ra

; Now the regression
Np = N_ELEMENTS(y) ; number of points
Nv = 1 ; number of variables (1 for simple reg)
Df = Np - Nv - 1 ; Degrees of freedom

r = regress(x, y, FTEST=ftest)

p_e = 2 * (1 - t_pdf(sqrt(ftest), Df)) ; error probability
if p_e lt 0.01 then Print, 'Significant at the 99% interval'
if p_e lt 0.05 then Print, 'Significant at the 95% interval'



I am always wondering what people mean when they write "significant". On
the slope? OR according to the f-Test?
[Message index]
 
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Read Message
Previous Topic: display gray image with NaN
Next Topic: P-value, DF, Variance

-=] Back to Top [=-
[ Syndicate this forum (XML) ] [ RSS ] [ PDF ]

Current Time: Wed Oct 08 16:11:58 PDT 2025

Total time taken to generate the page: 0.00442 seconds