[patch #6650] one pass covariance algorithm

[Jason H Stover]

Follow-up Comment #11, patch #6650 (project pspp):

"1. I'm confused that covariance_matrix.h now has a _init and a _create
function. Aren't they the same thing? Similarly there is both a _destroy and a
_free function. "

They are supposed to be the same, but I created new functions so I wouldn't
clobber any other functions that might be using the old ones. If the patch
looks OK, I will keep only the newest ones and rename them.

"2. The non-static functions in covariance_matrix.c could all do with
comments. "

I'll add them.

"6. Again in run_glm, the first reference to n_data is n_data++, but n_data
has never been initialised. "

I plan to remove n_data by placing the number of valid cases for each
variable inside the covariance matrix struct. Then there will be no more
n_data.

"7. What should I pass to argument 2 of covariance_matrix_compute when
computing a simple covariance matrix with no categorical variables involved?
Is zero appropriate?"

After I drop n_data, covariance_matrix_compute will take only one argument.

I will post an updated patch soon. In it, the covariance matrix struct will
have the sample size stored, which will allow handling of missing values and a
simplification of the interface.



    _______________________________________________________

Reply to this item at:

  <http://savannah.gnu.org/patch/?6650>

_______________________________________________
  Message sent via/by Savannah
  http://savannah.gnu.org/