Parameters

[James Malone]

Dear All,

I wish to model parameters used within processes in data
transformations in OBI. I know variable has been added to the latest
IAO file, but will parameter? If you've not thought about it, what do
you need from me so that we can begin discussions to add it?

Thanks,

James

[Alan Ruttenberg]

Melanie outlines the current "variables" we have. None of these is
what you mean I think.
The general schema for a parameter is that it is some piece of
information about something. I think it would help if we started with
a short list of some representative "parameters" and work out what
they are about so that we can get a feel for the shape of the problem.

-Alan

[James Malone]

Alan,

Here is a snippet of a correspondence with Monnie regarding parameter
(statistician that works with us on DT):


In statistics a parameter is a characteristic of a population. A
population is the entire universe of individuals about which we are
interested in gathering data. A parameter has no variance - it is a
true value, sort of like a universal constant. A statistic is a
characteristic of a sample. A sample is a subset of a population from
which we actually take measurements. So, the statistic is subject to
variability and error, and we use it to estimate the parameter of
interest.

For example, we want to know the mean cholesterol level of ALL
INDIVIDUALS who are taking some cholesterol-lowering drug. The mean
is a parameter because we want the mean for all individuals who take
the drug. However, we have only the sample mean - the mean calculated
from the individuals in a study. The sample mean (a statistic) is
meant to estimate the population mean (a parameter).


James

[Bjoern Peters]

First, this definition of 'parameter' is very different from my
intuitive one, which would be more along the lines of 'argument of a
computer function', or 'constant in a mathematical equation'. Asking
google to define:parameter includes the definition below, but only as
hit #10. Therefore I would strongly urge not to use the label
'parameter' for this, but something more specific. Maybe 'population
characteristic', which Monnie used as well?

Secondly, I am unsure where to put this. A population characteristic is
typically unknown, which means it is not an information artifact.
Rather, it is 'the thing that is approximated by a sample
characteristic' (which is an information artifact). That makes me want
to treat population characteristics as qualities of populations. That
won't be trivial either, but roughly:

'average height of German postdocs in 2008' is a population
characteristic of the population 'German postdocs in 2008'. Each member
of that population has_quality height. So I would argue that there is a
quality 'average height' inhering in a group of entities. Measuring the
height of 20 postdocs gives a sample mean.

'average sample height' is_measurement_of some (average_height and
inheres_in some sample_population)

To relate it to the population characteristic, a new subproperty of
is_about is needed, like 'is_estimate_of', so that we can state:

'average sample height' is_estimate_of some (average_height and
inheres_in some entire_population)

- Bjoern

--
Bjoern Peters
Assistant Member
La Jolla Institute for Allergy and Immunology
9420 Athena Circle
La Jolla, CA 92037, USA
Tel: 858/752-6914
Fax: 858/752-6987
http://www.liai.org/pages/faculty-peters

[James Malone]

Hi Bjoern,

Yes there are at least two uses of the word. The computer function
parameter is quite close to the mathematical use. I can find several
definitions on the net some better than others, the definitions
further down this wiki page http://en.wikipedia.org/wiki/Parameter are
ok for defining where they lie within a function but little more. The
statistical one is more complicated as you've spelled out but we need
both. So to push on, considering the mathematical/computer function
use, one thing that seems to strike me is that for a data
transformation, that a paremeter is a part of the process. If we say
that a parameter mu is part of some equation involved which defines a
data transformation, there are two things we need to say 1. mu is part
of the equation (i.e. the DT) and 2. that mu is a paremeter. Does
this help Alan?

James

[Fostel, Jennifer (NIH/NIEHS) [C]]

there is yet another use -- we have protocol parameters also, probably similar to the mathematical sense.
for instance, if we had one protocol to administer a drug to a rat, and we varied the dose administered, we would consider the dose a parameter of the protocol. canwe include this as well?

...jennifer

[Alan Ruttenberg]

On Wed, Mar 25, 2009 at 10:21 AM, Fostel, Jennifer (NIH/NIEHS) [C]
<fos...@niehs.nih.gov> wrote:
>
> there is yet another use -- we have protocol parameters also, probably similar to the mathematical sense.
> for instance, if we had one protocol to administer a drug to a rat, and we varied the dose administered, we would consider the dose a parameter of the protocol.  canwe include this as well?

This one is very *different* from the mathematical one, which claims
to be divorced from reality. In your case the dose sounds like an
independent variable. If not, could you explain the difference?

-Alan

[Alan Ruttenberg]

Averages considered qualities worry me as they don't feel like
qualities in the usual sense. For comparison consider your average
blood pressure over a week. Now the averaging is over a process. It's
almost like a defined class - not primary in the way that qualities
are.

Maybe one of our philosopher readers can offer some thoughts.

-Alan

[Alan Ruttenberg]

OTOH, a *distribution* as a quality seems more correct for a
population. The average, mean, deviation, are entities with some
relationship to that quality.

-Alan

[Bjoern Peters]

Agree with Alan. Jennifer's case is already covered in OBI by the
'controlled variable specification'. The fact that e.g. a drug dose is
varied in different administration should be modeled as part of the
study design specification. (I don't find the terms in the current file
though; will follow up with Melanie). But this should be discussed in
OBI / DENRIE, not IAO.

- Bjoern

:

[Alan Ruttenberg]

On Wed, Mar 25, 2009 at 11:48 AM, Albert Goldfain
<albertg...@gmail.com> wrote:
> Alan,
>
> I agree.  It seems to me that averages over space (e.g., average height of a
> population) and averages over time (e.g., average weekly blood pressure)
> don't seem like qualities...or, at least, they don't seem like first-order
> qualities, since they depend on other qualities for their definition...and
> since BFO doesn't make a first-order, second-order distinction, (or have
> something like a dependent quality) it seems like something that IDO should
> take up...perhaps making averages an information artifact.
>
> Against this view is the fact that a population does have an average height
> even if no one performs the measurement...and I have an average weekly blood
> pressure even if no one slaps a cuff on me.  So no one needs to perform an
> "averaging" for the average to exist.  Also, it seems crystal clear in both
> cases what IC the averages inhere in.

Well, a population also has a whole bunch of subpopulations with
different members. For those that have 3 or more members there is a
3rd one to come into existence. There is also the average luminosity
of the shoes they wear, if they wear shoes. (to put on a Barry hat)

It isn't crystal clear to me what average inheres in. It does seem
crystal clear what the average *depends on*. But inherence is only one
kind of dependence.

> I guess my only comment would be that however averages are handled...the
> spatial and the temporal should be handled in a uniform way.

My inclination is towards the same. The technicality that processes
don't have qualities gets in the way of treating average height as a
quality, btw.

-ALan