Re: How to set alpha with robustfit function

Eddy, I probably don't understand your question, since t-statistic
computations don't require alpha. Instead they produce a p-value.

Here's an illustration of how I can get confidence intervals from a
combination of robustfit and nlparci. Maybe this will help.

-- Tom

% Fit by least squares, get confidence intervals two ways
x = (1:10)';
randn('state',0);
y = 10 - 2*x + randn(10,1); y(10) = 0;
X = [ones(10,1) x];
[bls1,bci1,resid] = regress(y,X);
bci2 = nlparci(bls1,resid,'jacobian',X);
bci1,bci2

% Use robustfit to do least squares, get confidence intervals from nlparci
[bls2,st] = robustfit(x,y,@(r) ones(size(r)),1);
bci3 = nlparci(bls2,st.resid,'cov',st.covb);
bls1,bls2
bci3

% Use robustfit to do a robust fit, get confidence intervals from nlparci
[brob,st] = robustfit(x,y);
bcirob1 = nlparci(brob,st.resid,'cov',st.covb);
brob,bcirob1

% Change alpha to get 90% confidence intervals
bcirob2 = nlparci(brob,st.resid,'cov',st.covb,'alpha',0.1)



"Eddy Letsoalo" <eddy.letsoalo@xxxxxxxx> wrote in message
news:ef58f81.1@xxxxxxxxxxxxxxxxxxxxxxxxxx
Hi Tom,

Thanks for your tip. However, I wonder whether you have understood my
question very well.

In fact, I am using both the regress and robustfit functions. With
regress, I can enter alpha as an argument but the outcome does not
display t-values rather confidence intervals. I then use robustfit
function and get more detailed results with t-stats. However, alpha
is not an explicit argument.

Is there a way to set say alpha = 0.1, 0.01, etc. and get
ready-computed t values?

Thanks,
Eddy

Eddy Letsoalo wrote:


Dear MATLAB users,

Can somebody advise me how to set a confidence interval (alpha)
with
robustfit function other than the 0.05 level of significance.

Thanks in advance


.

Relevant Pages

  • Re: Alpha, beta, power, CL, CI
    ... Alpha is the probability that our test would conclude that T is better than S even though this is not the case. ... Beta and power are two ways of talking about the same thing. ... They are all about how likely it is that, given fixed alpha, the number of patients N we will test the treatments on and assuming a scenario under which T actually is better than S, our test will actually conclude that T is better than S. That probability is called power, while beta is simply 1-power. ... Confidence intervals corresponding to a certain confidence level are loosely speaking* based on the following idea: Given outcomes to our experiment, we want to quantify how precise our estimates of e.g. for the difference in response rates between the two treatments are, for this purpose we want to specify a range about which we are confident that the true difference will lie in it. ...
    (sci.stat.math)
  • Re: How to change robustfit function arguments?
    ... I want t-stats values andnot confidence intervals. ... it is easy to do with regress function because alpha ... combination of robustfit and nlparci. ... I am using both the regress and robustfit functions. ...
    (comp.soft-sys.matlab)
  • Re: Old bug in corrcoef not yet fixed
    ... % confidence intervals. ... the 'alpha' value is simply ignored (as far back as in R13) ... Sorry for the repetition but I forgot one very important thin in the subject - the BUG word ... To show any differences between the default 'alpha' value .05 and the .5 value you entered, you would have to display these confidence interval bounds. ...
    (comp.soft-sys.matlab)
  • Old bug in corrcoef not yet fixed
    ... % confidence intervals. ... the 'alpha' value is simply ignored (as far back as in R13) ... Sorry for the repetition but I forgot one very important thin in the subject - the BUG word ...
    (comp.soft-sys.matlab)
  • Re: Old bug in corrcoef not yet fixed
    ... % confidence intervals. ... the 'alpha' value is simply ignored (as far back as in R13) ... Sorry for the repetition but I forgot one very important thin in the subject - the BUG word ...
    (comp.soft-sys.matlab)