Skip to main content

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: "hotness" (Read 201594 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

"hotness"

Reply #100
I also thought about that:
http://www.hydrogenaudio.org/forums/index....ndpost&p=281175

I mean, does hotness takes into account the size of the library? because it's for sure that the largest your library is, the less you're likely to play tracks even your favourites one. I often tends to forget about traks that I have and I love, but with so much new tracks and the size of my library going up I get 'distracted'

"hotness"

Reply #101
Quote
Anyway this give me some weird results because it doesn't match with the result I get from the column UI global, you can check it via this screenshot:

[a href="index.php?act=findpost&pid=280765"][{POST_SNAPBACK}][/a]


hey! can you upload your hotness column string?

"hotness"

Reply #102
Quote
I also thought about that:
http://www.hydrogenaudio.org/forums/index....ndpost&p=281175

I mean, does hotness takes into account the size of the library? because it's for sure that the largest your library is, the less you're likely to play tracks even your favourites one. I often tends to forget about traks that I have and I love, but with so much new tracks and the size of my library going up I get 'distracted'
[a href="index.php?act=findpost&pid=281192"][{POST_SNAPBACK}][/a]

That's what the baseline settings are for...tweak and see how it affects things.  I wish there were a more automatic way to set those, but I think that's impossible (even if you have a huge library, you might only listen to five albums in a week...either over and over or just once each...or you might have music on all day and listen to almost everything you have...very tricky.  Read the rules of the thumb I give in the code for configuring those baselines and let me know if you think they're accurate.  They don't come from practice, but from theory.

"hotness"

Reply #103
Quote
That's what the baseline settings are for...tweak and see how it affects things.

Ok, I didn't really read all the settings, thanks for the tip.

Quote
-the more often you listen to music, the LOWER baselinedecay should be
-the more variety of music you listen to, the HIGHER baselinefrequency should be


I have quite a large library (around 225 albums) and listen to music around 2-3 hours a weekday and much more on the weekend. So I don't know if I should alter the settings, they seems quite fine for me. Maybe rise a little the decay and frequency could help. But 've never been really good with statistics

"hotness"

Reply #104
Quote
Quote
Anyway this give me some weird results because it doesn't match with the result I get from the column UI global, you can check it via this screenshot:

[a href="index.php?act=findpost&pid=280765"][{POST_SNAPBACK}][/a]


hey! can you upload your hotness column string?
[a href="index.php?act=findpost&pid=281218"][{POST_SNAPBACK}][/a]

Sure.

Code: [Select]
$blend(414141,0000FF,%_hotness%,100)%_hotness% ●

"hotness"

Reply #105
Just some thoughts. I got the impression that fresh songs (recently added) get too much hotness, I mean i.e. a track very recently added get around 85 on the first play while older track gets not so great bumps.

For me it takes sense that fresh tracks get better hotness indicators that older ones, but maybe not so higher values. In the mean time, i also got the impression that old tracks even with some playcounts and good rating takes too much to take hotness indicators.

i.e. a track with 5 rating, added the 01-04-2004, played 3 times and last played 09-03-2005 doesn't get any hotness (remains at 0) when it's played today. I would tends to think it deserves at least some tiny points... 

Well, doesn't know if this makes sense or I'm totally wrong

"hotness"

Reply #106
Quote
Quote
Quote
Anyway this give me some weird results because it doesn't match with the result I get from the column UI global, you can check it via this screenshot:
[a href="http://img184.exs.cx/my.php?loc=img184&image=clipboard4nc.png" target="_blank"]

"hotness"

Reply #107
Quote
Just some thoughts. I got the impression that fresh songs (recently added) get too much hotness, I mean i.e. a track very recently added get around 85 on the first play while older track gets not so great bumps.

For me it takes sense that fresh tracks get better hotness indicators that older ones, but maybe not so higher values. In the mean time, i also got the impression that old tracks even with some playcounts and good rating takes too much to take hotness indicators.

i.e. a track with 5 rating, added the 01-04-2004, played 3 times and last played 09-03-2005 doesn't get any hotness (remains at 0) when it's played today. I would tends to think it deserves at least some tiny points...  

Well, doesn't know if this makes sense or I'm totally wrong
[a href="index.php?act=findpost&pid=281277"][{POST_SNAPBACK}][/a]

No, you're probably totally right.  I haven't been keeping track of playcount and added for very long at all, which is why this kind of feedback is good.  I don't know off the top of my head how you could solve that problem by altering the algorithm itself, but I would refer to the rules of thumb again (or which there are two for each setting):

baselinefrequency:
- decrease if songs stay hot too long
- increase if you listen to a large variety

baselinedecay:
- decrease if too many songs are hot
- decrease if you listen to music very often

Again, those rules don't come from practice, they come from theory.  I myself am no good at statistics, but I wrapped my head around this problem for about a week and reached those rules of thumb pretty confidently (even though I know one or two of them sound counter-intuitive).

Another issue is that you must have "start decay today" turned on, which is why those rarely-played songs are starting at 0.  Of course, changing that would just make them 100% today, and 0% tomorrow (and something really bothers me about having a playlist full of 100% hot songs...boring data.)

I would increase baselinefrequency (because songs aren't staying hot long enough) and see how that feels.  Also, this might be a good argument for implementing hourly rather than daily decay (something that might be very, very beautiful).  I don't know if this would occupy more system resources, but I can't imagine it would be noticable since it would just be doing the same math with numbers in the thousands instead of in the hundreds.  If anyone knows how this would affect things, let me know, otherwise I might just go ahead and putting together an hourly decay version...man would that be sweet, watching all those blinking dots...

"hotness"

Reply #108
@topdownjimmy: thanks for the explanations, I'll play with them

BTW, an hourly based decay would look nice.

"hotness"

Reply #109
Quote
@topdownjimmy: thanks for the explanations, I'll play with them

BTW, an hourly based decay would look nice.
[a href="index.php?act=findpost&pid=281373"][{POST_SNAPBACK}][/a]

Success.  I'm about to post an "hourly decay" version in the first post.  Seems to work over here at first glance, lemme know if there are problems. 

Softer gradients, more dynamic data.

"hotness"

Reply #110
Quote
I don't know if this would occupy more system resources, but I can't imagine it would be noticable since it would just be doing the same math with numbers in the thousands instead of in the hundreds. If anyone knows how this would affect things, let me know, otherwise I might just go ahead and putting together an hourly decay version
I don't think it will affect the system resources but you could try it, btw hourly based hotness would be perfect.

Another useful thing would be hotness to look in your entire music collection rather than only on active list, that way it would calculate even better the hotness cuz it would take in count all all the files....

EDIT: @topdownjimmy: I haven't noticed that you where going to post an hourly based version! Perfect!
Random number generation is too important to be left to chance

"hotness"

Reply #111
Quote
Another useful thing would be hotness to look in your entire music collection rather than only on active list, that way it would calculate even better the hotness cuz it would take in count all all the files....
[a href="index.php?act=findpost&pid=281387"][{POST_SNAPBACK}][/a]

I'm afraid foobar isn't capable of doing anything like that.  :/

"hotness"

Reply #112
Quote
Quote
Another useful thing would be hotness to look in your entire music collection rather than only on active list, that way it would calculate even better the hotness cuz it would take in count all all the files....
[a href="index.php?act=findpost&pid=281387"][{POST_SNAPBACK}][/a]

I'm afraid foobar isn't capable of doing anything like that.  :/
[a href="index.php?act=findpost&pid=281389"][{POST_SNAPBACK}][/a]


  Oh too bad

Thx anyway...
Random number generation is too important to be left to chance

"hotness"

Reply #113
Quote
Success.  I'm about to post an "hourly decay" version in the first post.  Seems to work over here at first glance, lemme know if there are problems. 
[a href="index.php?act=findpost&pid=281382"][{POST_SNAPBACK}][/a]

Works fine here. I've noticed that hotness values have changed a little (i.e. for the most played track on my library, that had a 55 hotness indicator, with the new version it has a 71 indicator), I guess it has to do with the new hourly decay code.

Thanks

"hotness"

Reply #114
how can i get 100% hotness?

"hotness"

Reply #115
I just tried this also, and I really have to say: good work!
One "problem" I have is that every song I play gets immediately 100 as hotness, but that is only because I dont use the %added% tag (or so I've figured). Anyway, it isn't that big problem with the hourly decay version, which seems to work quite well
I'll also need to check the baseline values for any tweaking they might need, but I'm not there yet, as I installed kode's foo_playcount yesterday (after deleting all %last_played% and %play_date% tags I had). I just don't like the idea of having them as actual metadata, the external database is much better for it (same goes with ratings).

I also felt that a new column for another "rating" would be too much so I added it nicely to my rating column:
Code: [Select]
%_color_background%'('%_hotness%') '
%_color_pale%$repeat('•',$sub(5,%rating%))
%_color_info%$if(%rating%,$repeat('•',%rating%))

1) color_background is naturally the background color of the foobar window so the actual hotness value shows only in tooltips (the rating column is just wide enough for the 5 stars to show).
2) color_pale is just a light grey color ($rgb(234,234,234)), so if a song has a rating of 3, it shows 2 dots with this color and 3 with the:
3) color_info, which is defined like this in global variables
Code: [Select]
color_info=$blend($rgb(179,179,179),$rgb(77,107,97),$get(hotness),100)

This goes from a dark blue to the default color of my ratings, so the higher hotness, the darker blue. Works great!
Oh and it's called color_info just because I used to have more info than just the rating with that color

"hotness"

Reply #116
Quote
Quote
Success.  I'm about to post an "hourly decay" version in the first post.  Seems to work over here at first glance, lemme know if there are problems. 
[a href="index.php?act=findpost&pid=281382"][{POST_SNAPBACK}][/a]

Works fine here. I've noticed that hotness values have changed a little (i.e. for the most played track on my library, that had a 55 hotness indicator, with the new version it has a 71 indicator), I guess it has to do with the new hourly decay code.

Thanks
[a href="index.php?act=findpost&pid=281474"][{POST_SNAPBACK}][/a]

Yep, that's how it should work.  Instead of showing 24 hours of decay, that track is showing 17 hours of decay (or whatever the values).  How are the baseline experiments going?  What's convenient about that is that you can change those settings in your globals config and immediately see the effects of those changes.  A good way of calibrating your settings would be to create a playlist with all your music and reverse sort by hotness.  See which track is at 50%, or which track has the lowest non-zero hotness, see if that "feels correct," then adjust. 

"hotness"

Reply #117
Quote
I just tried this also, and I really have to say: good work!
One "problem" I have is that every song I play gets immediately 100 as hotness, but that is only because I dont use the %added% tag (or so I've figured). Anyway, it isn't that big problem with the hourly decay version, which seems to work quite well
[a href="index.php?act=findpost&pid=281484"][{POST_SNAPBACK}][/a]

This has nothing to do with %added%, in fact hotness should operate fine even without %added%.  The reason that happens is because I turned off immediate decay in this version.  Immediate decay was introduced because previously, every song you played "today" stayed at 100% until tomorrow, which was boring and annoying (especially if tomorrow it went down to 30% or something).  I guess I could put that option back in, although in that case, the 24-hour "clairvoyance" of hotness (i.e., using "tomorrow's" value instead of "today's" value) would be kind of arbitrary (since our unit is now the hour, not the day).

"hotness"

Reply #118
Quote
How are the baseline experiments going?  What's convenient about that is that you can change those settings in your globals config and immediately see the effects of those changes.  A good way of calibrating your settings would be to create a playlist with all your music and reverse sort by hotness.  See which track is at 50%, or which track has the lowest non-zero hotness, see if that "feels correct," then adjust. 
[a href="index.php?act=findpost&pid=281530"][{POST_SNAPBACK}][/a]

I'll play with it tomorrow and see the results. For soem reason I can't seems to get sort by hotness working...

BTW, as mentioned here I've also noticed that every track gets now a 100 hotness when played, no matter if this has been never played and added 1 year before. What I suppose it's that this song hotness will decay quicker than other more fresh or played more often.

"hotness"

Reply #119
Quote
Quote
How are the baseline experiments going?  What's convenient about that is that you can change those settings in your globals config and immediately see the effects of those changes.  A good way of calibrating your settings would be to create a playlist with all your music and reverse sort by hotness.  See which track is at 50%, or which track has the lowest non-zero hotness, see if that "feels correct," then adjust. 
[a href="index.php?act=findpost&pid=281530"][{POST_SNAPBACK}][/a]

I'll play with it tomorrow and see the results. For soem reason I can't seems to get sort by hotness working...

In the settings for the hotness column, you need to check "Use custom sort spec" and put this code in the "Sort" tab:
Code: [Select]
$num($sub(100,%_hotness%),3)

You also need to check "Use global variables when sorting by column" checked in the globals tab.

Quote
BTW, as mentioned here I've also noticed that every track gets now a 100 hotness when played, no matter if this has been never played and added 1 year before. What I suppose it's that this song hotness will decay quicker than other more fresh or played more often.
[a href="index.php?act=findpost&pid=281571"][{POST_SNAPBACK}][/a]

This is because I turned off immediate decay.  I could add a setting for "clairvoyance" or something that will look x hours ahead and use that hotness (the default daily decay setting was to look 24 hours ahead) so that songs aren't given a hundred percent right off the bat.  I didn't think this would be an issue since songs will start decaying in an hour, but maybe I'll at least make it optional.  Default might be 12 hours or something I guess.

"hotness"

Reply #120
Been following this for some time now and seeing its getting into an interesting phase, ppl playing about with differnent values to make it more meaningful.

I had an idea, i think hotness would be a great way to show people what you are listening to at any point in time.

I let you hash out the details of how much is to be presented etc. but the idea i had was to give a useful picture of whats going on currently, whats hot etc.

If there was a way to generate HTML out of it and post it would make it even better.

"hotness"

Reply #121
Quote
Quote
Another useful thing would be hotness to look in your entire music collection rather than only on active list, that way it would calculate even better the hotness cuz it would take in count all all the files....
[a href="index.php?act=findpost&pid=281387"][{POST_SNAPBACK}][/a]

I'm afraid foobar isn't capable of doing anything like that.  :/
[a href="index.php?act=findpost&pid=281389"][{POST_SNAPBACK}][/a]


something you could do to work this in, you could have a variable the user defines, where they just insert the number of tracks in their library, or 0 if they don't want to use this particular feature.
sure, it's not so dynamic, but it would get the job done...

"hotness"

Reply #122
Quote
Quote
Quote
Another useful thing would be hotness to look in your entire music collection rather than only on active list, that way it would calculate even better the hotness cuz it would take in count all all the files....
[a href="index.php?act=findpost&pid=281387"][{POST_SNAPBACK}][/a]

I'm afraid foobar isn't capable of doing anything like that.  :/
[a href="index.php?act=findpost&pid=281389"][{POST_SNAPBACK}][/a]


something you could do to work this in, you could have a variable the user defines, where they just insert the number of tracks in their library, or 0 if they don't want to use this particular feature.
sure, it's not so dynamic, but it would get the job done...
[a href="index.php?act=findpost&pid=281711"][{POST_SNAPBACK}][/a]


yeah, but it's an ugly one

cya!
Random number generation is too important to be left to chance

"hotness"

Reply #123
Quote
Quote
Quote
Another useful thing would be hotness to look in your entire music collection rather than only on active list, that way it would calculate even better the hotness cuz it would take in count all all the files....
[a href="index.php?act=findpost&pid=281387"][{POST_SNAPBACK}][/a]

I'm afraid foobar isn't capable of doing anything like that.  :/
[a href="index.php?act=findpost&pid=281389"][{POST_SNAPBACK}][/a]


something you could do to work this in, you could have a variable the user defines, where they just insert the number of tracks in their library, or 0 if they don't want to use this particular feature.
sure, it's not so dynamic, but it would get the job done...
[a href="index.php?act=findpost&pid=281711"][{POST_SNAPBACK}][/a]

I'm not so sure that would make things any easier for calibration...even if you have a huge library, you might only listen to a few albums at a time for a few weeks, or you might listen to a broad range of things.  In other words, it's not really about how big your library is, it's about how much of your library you listen to, and how often you listen to anything at all.

"hotness"

Reply #124
I've just changed baselinefrequency to 21 and I think everything seems correct. More old tracks are rated as little hot, and I think this fits more my listening habits

BTW, I've just requested to cwbowron to make available a %systemdate% function with system hour (hh:mm:ss) so I can try to adapt it to PT in an hourly basis. Still I can't seesm to get it working correctly (still getting those weird numbers). topdownjimmy, did you get the time to take a look at it?