Missing values

[John W. Eaton]

On 28-Feb-2002, Marcelo Bruno <address@hidden> wrote:

| i have a matrix with missing values (e.g. 9999 or -9999).
| How does octave work with missing values ???

Octave doesn't currently have a good way of working with missing data.

It is possible to use NaN (or some other special value, if you know
that it is outside the range of your data) but you will have to do
some work to get the results you want.  If you want to use this
method, the functions written by Alois Schloegl would probably be a
good place to start.  However, I don't think that this method should
be the default in Octave, because NaN can be generated by invalid
operations and as it is now, there would be no way to tell whether any
given NaN value was intended to represent missing data or was
generated by an invalid operation.

In R (www.r-project.org), missing data is represented by a particular
NaN value (the range of possible NaN values includes more than one bit
pattern) which presumably cannot be generated by a calculation.

Another possibility is to use separate data objects to represent
matrices that contain missing values.  This might be more work to
implement than using a special NaN value, but it might be more
efficient (we wouldn't have to have checks for missing data in all
functions that handle matrices).

jwe



-------------------------------------------------------------
Octave is freely available under the terms of the GNU GPL.

Octave's home on the web:  http://www.octave.org
How to fund new projects:  http://www.octave.org/funding.html
Subscription information:  http://www.octave.org/archive.html
-------------------------------------------------------------