Re: Stats Toolbox - DFITTOOL Error?

Hi,

Please search this newsgroup for a similar discussion on 08/29/2005.

Hope this helps.
Bin


"R Gow" <r.d.gow_no_spam@xxxxxxxxxxxxxxxxxxxxxxx> wrote in message
news:ef13bff.-1@xxxxxxxxxxxxxxxxxxx
>I am using the DFITTOOL of the Stats Toolbox, and for each fit made,
> the fit summary shows a log likelihood value. Fits which appear good
> visually are in the order of -ve millions, for example -3.4243E006.
> Here, the fit with log likelihood -3E006 would appear closer than one
> with say, -5E006, implying that the closer the log likelihood is to
> zero (or unity?) the better?
>
> Nevertheless, some poor appearing fits have L.L. in the order of -ve
> thousands. Again, a fit with L.L. -1500 would appear better than one
> with -5000, again implying that the lower the value the better, but
> this cant be the case since -1500 is obviously less than -5E006, yet
> the -5E006 fit would look better?!
>
> Is there an error perhaps in the toolbox - for example, should the
> good fits have a negative exponent, i.e. -5E-006? What basically is
> the purpose in the likelihood value - how should it be used when
> comparing fits??
>
> Thanks, i very much appreciate any advice you have.


.

Relevant Pages

  • Stats Toolbox - DFITTOOL Error?
    ... I am using the DFITTOOL of the Stats Toolbox, and for each fit made, ... the fit with log likelihood -3E006 would appear closer than one ... with say, -5E006, implying that the closer the log likelihood is to ...
    (comp.soft-sys.matlab)
  • Log Likelihood - Statistics Toolbox
    ... The results produce a log likelihood ratio value and ... estimates of covariance etc. ... fit is better, is it true that the closer the log likelihood is to ...
    (comp.soft-sys.matlab)
  • Re: model selection problem
    ... I cannot see a reason why the Log Likelihood ... What I need especially is a good fit in the ...
    (sci.stat.math)
  • Re: working storage values
    ... Are you implying that tools should be optimized to fit their task? ... A train engine won't run a cargo ship, nor will it run a sports car. ... The world land-speed record for a steam driven car is in excess of 120 ...
    (comp.lang.cobol)
  • Re: Great show - it is good again!
    ... >>> If you are implying that I fit ANY of the above, you are smoking the ... >>> rope, son... ...
    (alt.fan.howard-stern)