I recently offered AAVSO members lots of targets based on data mining I did but never published (Bad boy!)
http://www.martin-nicholson.info/freegratis/freegratis.htm

Its a very interesting work you do that gives new insights.

Looking just at the total number of observations is probably misleading
as some variables have been observed for a very long time (like e.g. Mira
that was discovered long ago) and others not (e.g. variables that were
discovered recently, variables that gained attention or were added to the
observing programs (or got a sufficient comparison star chart) just lately).
So you might want to use the observations per year for some of the sta-
tistical evaluations to get more meaningful results.

An other approach would be to use ‘sufficient coverage’ as measure
for some evaluations. In general a lightcurve is considered to be suf-
ficiently covered if one has collected 30-100 observations per period
for variables without very steep bightness changes (e.g. unlike EA
or EP type eclipsing, erruping and cataclysmic variables). Then one
could bin the observations in time ranges of e.g. period/50 and just
count the number of occupied bins instead of the total number of ob-
servations.
Then one could also get a measure for the ‘lightcurve coverage’ (oc-
cupied bins divided by all bins since the first observation) which
gives a good measure for regularly vs. unregularly/underobserved
stars. And an other measure for the ‘observation strength’ (avarage
observations per occupied bin) which is a good measure for over-
observed stars. As not all stars are and can be observed for the
same time during the year, one could apply a declination depend-
ent correction.

For the total number of references its similar. Here you could use the
references per year since the first year of the first observation/refer-
ence. As the referencing is not complete in ADS, you might want to
consider just the year range where the referencing in ADS is pretty
complete.
Here one might also find a change in interest of special variables or
variable types over the time.

Clear skies
Wolfgang

Martin Nicholson
AAVSO code is NMR

I just read your Msg on the AAVSO-DIS list Titled “Mining GCVS and AAVSO data”.

I realize this is unprofessional but WOW. I Bow before you.
I was blown away with your efforts in Datamining. A man after my own heart.
I just wanted to let you know that Your effort IS appreciated and I will be following any more of your msgs with close interest.

Thank you Again. WOW..
Phil

Clear Skies