0
$\begingroup$

When I was reading a project paper, I came across this phrase:

particularly if you have a lot of data and a model without many degrees of freedom.

What is meant by a model without many degrees of freedom?

I have gone through this thread, but the thread just gives a general definition for df.

$\endgroup$
  • $\begingroup$ That refers to the number of free parameters that can be adjusted while fitting the model. $\endgroup$ – Marc Claesen Mar 4 '15 at 12:46
  • $\begingroup$ @DLDahly I have updated a question. $\endgroup$ – Elizabeth Susan Joseph Mar 4 '15 at 13:02
  • $\begingroup$ @MarcClaesen - Could elaborate the answer. $\endgroup$ – Elizabeth Susan Joseph Mar 4 '15 at 13:02
  • 1
    $\begingroup$ Examples: a multiple regression model has as many degrees of freedom as there are parameters to be estimated, which is equal to the number of regressors (the intercept and the error variance may or may not count, but that is just a minor detail). ARMA model has $p+q$ degrees of freedom where $p$ is the autoregressive lag order and $q$ is the moving-average lag order (again, intercept and error variance may be included in the count). A "model without many degrees of freedom" is similar to a "parsimonious model". $\endgroup$ – Richard Hardy Mar 4 '15 at 13:30
  • $\begingroup$ @RichardHardy - So degrees of freedom are number of paramters. am I right? $\endgroup$ – Elizabeth Susan Joseph Mar 4 '15 at 13:51