It's been a busy few days at the Quick service. Here's a query just in,
and I don't know whom to ask. Any comments or suggestions?
From the limited information given, it sounds to me like a
cluster-analysis problem, with the emphasis on percolation (well he
wants to write in Java (-:). It's been a long time since I read about
Voronoi diagrams in "Algorithms" so I may be completely misunderstanding
the problem.
Re. FITS in Java, STIL or Tom McGlynn's package comes to mind. If we
pointed him to the former, we might get some classes useful for TOPCAT.
I am a computing student in my final year at university. It is only
recently that I have been introduced to the Starlink project via my data
mining tutor who is also a astrophysicist.
For my dissertation I have chosen to do a project based on finding
outliers in data. To do this I intend to calculate a voronoi diagram
for data and calculate the outliers byt the area of the polygon. I.E.
The points having the greater area are widely dispersed from the main
group and may be outliers.
The problem I am having is when the outlier becomes a cluster of
outliers. My method of detection will not detect them based soley of
the polygon area.
AlthoughI have done research into this area,it has not yielded any
signifcant results as yet and I was hoping that you may have some
suggestions that could point me in the right direction,.
I will be getting the data from FITS tables and will be doing all the
coding in java.
Malcolm
|