Re: [ontology-summit] Hackathon: BACnet Ontology

[Simon Spero <sesuncedu@xxxxxxxxx>]

On Tue, Mar 12, 2013 at 6:00 PM, Barkmeyer, Edward J <edward.barkmeyer@xxxxxxxx> wrote:

 

That said, the sad fact is that a pump curve really is just someone's fit to a set of actually measured values for particular inputs, and the real data is just the pairs (input, measured behavior).  The fit is indeed in the eye of the sender, and if the receiver uses a different algorithm and gets a different curve, neither of them is necessarily "right" (they are just the products of different engineering intuitions or skills), and the proof of the pudding will come in the pump test.  That is, the multidimensional_property really is just a set of observations. The pump curve is not the same kind of thing as a spline curve specification for an airfoil, where that shape is what will have the desired aerodynamic properties.

 There is an interesting point here, which ties back in in to the summit theme of measurement and evaluation.  A sampling of points, without more, can be fitted to a curve using a variety of  non parametric techniques, such as splines, or LOESS (locally weighted regression).  

There are many different ways of measuring the goodness-of-fit of these curves ( e.g.  http://www.itl.nist.gov/div898/handbook/pmd/section4/pmd44.htm ), as well as ways of comparing the goodness of fit of multiple models (AIC, BIC,etc). 

There is a big difference between points representing samples taken at random/fixed intervals, and points that are chosen to precisely describe a curve, or that specify the parameters that characterize a function exactly, or to within a known range of precision- for example, being given a  sequences of 24-bit samples  48,000Hz that computation suggests approximate a 440Hz sine wave is qualitatively different from from being told that something is an audio recording of a tuning fork, which is different from being told that something is a 440Hz sine wave.

_________________________________________________________________
Msg Archives: http://ontolog.cim3.net/forum/ontology-summit/   
Subscribe/Config: http://ontolog.cim3.net/mailman/listinfo/ontology-summit/  
Unsubscribe: mailto:ontology-summit-leave@xxxxxxxxxxxxxxxx
Community Files: http://ontolog.cim3.net/file/work/OntologySummit2013/
Community Wiki: http://ontolog.cim3.net/cgi-bin/wiki.pl?OntologySummit2013  
Community Portal: http://ontolog.cim3.net/wiki/     (01)