Auto Rating of Songs & Radio Announcer v5.02 [Script]

rovingcowboy · Post by **rovingcowboy** » Thu Apr 27, 2006 9:07 am

DiddeLeeDoo wrote:Cool stuff!! You have to report how they work out in real-life!
And yes, you definition of the 'catch-up' script is better, thank you.

Actually, when I think about it, I mentioned in the script that a weekly option is on the To-Do list.. and those considerations should be in there.

Could turn out really cool... I'll test the weekly stuff with the data I have available,and put some priority on a weekly option!

the codes i put there are working in real life i tested it with a lot of short songs and ran the song up to 33 playcount and they changed the stars each time that the count matched the playcount codes.

and the adding the half star for never before played songs works. since it was your code i just made it use 10 instead of 50.

DiddeLeeDoo · Post by **DiddeLeeDoo** » Thu Apr 27, 2006 9:15 am

Here's a new toy for you. Weekly or Monthly Estimator

Note, this is a new script, and needs another scripts.ini section.

This one takes a bit longer to execute, but is interesting.
I would upload results, but the image server is not responding at the moment.

\Scripts\AutoRateCalibratorWeekly.vbs

Code: Select all

New Code coming soon

rovingcowboy · Post by **rovingcowboy** » Thu Apr 27, 2006 9:42 am

i think i know what you mean by being a little strange with the results..

the code for the play counts i posted are to use the database table of song's playcounter. which is able to be changed or left alone.

so the counts of 2 to 30 keep the rating correct for about a month with out having to reset the playcount to zero.

but in the catch up ratings script the playcounts used for a week would make for the songs to be rated lower then needed.

if you use the playcounts from 2 to 30 in the ratings catch up script then it should be set for a month to do the check and not a week.

so the count would need 31 plays in a month's time to be 5 stars

is that what your finding out?

rovingcowboy · Post by **rovingcowboy** » Thu Apr 27, 2006 9:58 am

no longer needed as the script was removed. so i removed my comment

DiddeLeeDoo · Post by **DiddeLeeDoo** » Thu Apr 27, 2006 10:08 am

Try this new Monthly version, I think that is actually better than weekly, you can set 'ww' for weekly, and 'm' for monthly in the one just posted above.

I had to quickly edit the one above, as I found I left a 'testing' error in it, should be alright now.

rovingcowboy · Post by **rovingcowboy** » Thu Apr 27, 2006 10:21 am

this is the ratings the new one you just updated above will give me for a month.

and i too have a 3 year old database.

0 songs for no less 0.0
0 songs for no less 0.5
2671 songs for no less 1.0
1776 songs for no less 1.5
1152 songs for no less 2.0
726 songs for no less 2.5
686 songs for no less 3.0
483 songs for no less 3.5
201 songs for no less 4.0
108 songs for no less 4.5
83 songs for no less 5.0

that look better to you over a month that looks good to me.

DiddeLeeDoo · Post by **DiddeLeeDoo** » Thu Apr 27, 2006 10:30 am

Great to see it in 'real-life' !!
Thank you for posting your results!

Just have to 'fancy' it up a bit, so it becomes a bit more user friendly, and get an 'Execute' button too....

Will start fresh after some zzzzz

rovingcowboy · Post by **rovingcowboy** » Thu Apr 27, 2006 10:41 am

i think you sould make this run once a month and fix the songs automaticly.

so instead of having give the estimate it would write the ratings for the songs. each month.?

maybe it would be easier to use auto ratings script and have it use the history play count field instead of the songs table's count field?

that way it would be correct when adjusting any song. this whole catch up script would not be needed then and would have been a great learning tool..

what do you think about that??

agentsmart · Post by **agentsmart** » Thu Apr 27, 2006 9:52 pm

yes i also think the switch in which the way the auto ratings gets the play count now to get it from the history count would be better then getting 3 or 4 scripts to do one thing?

DiddeLeeDoo · Post by **DiddeLeeDoo** » Thu Apr 27, 2006 10:01 pm

I agree with that! I took out this rating catch-up script for now, and will post a more advanced one soonish.

DiddeLeeDoo · Post by **DiddeLeeDoo** » Fri Apr 28, 2006 6:06 am

rovingcowboy wrote: so instead of having give the estimate it would write the ratings for the songs. each month.?

maybe it would be easier to use auto ratings script and have it use the history play count field instead of the songs table's count field?

that way it would be correct when adjusting any song. this whole catch up script would not be needed then and would have been a great learning tool..

what do you think about that??

Don't know really. Currently I have a catch-up script that works fine, and actually do the ratings in the songs-database properly too..

Plus quite a few options has been added. Currently trying to crack a nut about singular plays in the period set. I'll try to spread out these singular plays in some facion, so it all turnes out closer to perfect.
Testing various ways to do this, and will see what turnes out the best.

All up a nice little puzzle to finally get to feel on top of the ratings in the database....

rovingcowboy · Post by **rovingcowboy** » Fri Apr 28, 2006 7:28 am

i just was thinking that since the auto rating script adjusts the ratings by the playcount field of the songs table in the database.

that it would be messing up the songs if the catchup script had set one to something else? thus making it a messed up rating.

but since the catch up script uses the history playcount it would always adjust the ratings to the real play count because in that count list it would just be adding the counts to it.

but the songs table's count list can be changed at the users will.

now i like that being able to change it at will but after doing all this and thinking i do not like it for rating the songs.

for what i am trying to do with the weekly top of the charts list, i will use the songs table play count and keep resetting that back to zero. so the playlist would show only the songs mostly played that week.

in doing that the ratings from the songs i thought would be getting messed up all the time.? not sure on that since i been resetting the play count and the ratings every time i tested the script, so it might not mess it up but i am not sure?

rovingcowboy · Post by **rovingcowboy** » Fri Apr 28, 2006 11:04 pm

for those of you with a large database like mine

i just tried using the ratings over a 1825 day time scale

i ended up with one song that had one count and 4 stars.

so that month long count i posted on the other page does wonders

on longer time scales also.

DiddeLeeDoo · Post by **DiddeLeeDoo** » Sun Apr 30, 2006 4:03 am

Trying ways to get it all calibrated based on that 3 years old database, with 25,000 plays.... Just stuff in progress basically...

The levels tops out at 3.0 Stars here, and go down toward the 2.5 Star mark, and drops to nothing for lower levels.. Leaves some room for skipping songs down some steps too..

rovingcowboy · Post by **rovingcowboy** » Sun Apr 30, 2006 9:09 am

i think you are on the right track with the 3 year limit

i mean since it is almost the same on the 5 years i did it does not change much that song only had one play count because i keep reseting that count field.

but the history count set it to 4 stars, i also think most do not have a 5 year 24 / 7 playcount database they might not even have a 3 year old database.

so just hard coding the catch up script to check for 3 years and no older then the ratings would be a fairly correct rate. since most will listen to a song a lot in a year or two and slow down on the number of times they listen to a song or stop with it for a year or so after that, they might start again in the 4 or 5 year.

but since the count was done for a 3 years then the rating would still be the same.

of course setting the catch up script to go for only 3 years would also mean you have to set it to keep it running if the database Does not go that far back in time. other wise it would toss the zero'ed out error and not run.