This is an archive from Klassic Note Web Reports

Sunday, 20 August 2017 Hi, everyone. From here on I'll (probably) do up a series of posts to discuss about #KN10YEARS (this will not be in a new label btw) and everything I would like to discuss about, from statistics to opinions to... everything under the sun. Music, I mean. This week, we shall talk about... album rankings?


THE BACKGROUND & PROBLEM
A little background if anyone actually gives a damn. Every year, I post Song Awards results on the full results site (which can be found here) and in that is the album rankings for Top 10 Albums, currently ranked by how many songs I've given out of the release, and sorted by track count in descending order if their percentages meet.

This is alright and all, but it has certain caveats which are mainly:

• The ranking takes into account songs that have been given from previous releases, specifically albums with single songs (revised albums, I call it) and more commonly, best albums with only a few new songs or even none
• The ranking does not fully represent the amount of songs I have given with respect to the rest of the new songs in every album release, thus including (for example, best albums) all the releases I potentially have reviewed before

THE SOLUTION(S)
To this, I came up with two main alternatives to the usual (simple) method I've been using over the years, which loosely named in this post, shall be referred to as "songs in list" and "initiation percentage (%)" approaches.

The "in list" method differs from the original approach by only considering how many new songs from the album have been given, regardless of total track count. So for example if I give 5 songs in an album and 3 songs in another album of the same track count, the former will rank higher. The problem here is that it differs with album track size, thus ignoring how many songs I have not given and also brings back point 2 of THE PROBLEM listed before because larger albums will have a larger pool of new songs to give and thus, will rank high compared to mini-albums and revised albums.

The "initiation %" method is for now my preferred approach for now but it requires more calculation than before. To put it simply, it includes the previous method mentioned above PLUS I now have to compare it to how many songs I have not given in the album itself. So for example if there are 5 songs that I've not given in this 13-track album and I only give 4 new ones, then the size fof the album effectively has been reduced to 5, instead of the full 13 songs. This example will apply to current rank no.1 album Aki-chan's 3rd album "all time lovin'", just in case you wonder.

The advantage here is that it will compensate to an extent the second problem in the original approach (large albums will rank lower because a larger pool of songs to give will drop it's ratio down) and to a lesser extent the first problem (including revised albums), in contrast to the first alternative I proposed. However, the differences may not be accountable when relating to albums with smaller track count, which I worry can still skewer the ranking. This is because a revised album (with at least half the songs already given) is equal in track count to say, a new mini-album. This is fine, but it will not work if the mini-album is revised, ie. it'll have track count equivalent to singles, which will definitively put the release high in the ranking.

My current solution to this then is to further separate rankings to mini-albums and full albums, but that may not be enough to reflect on certain mini-albums, so... maybe manual sorting? I've done that for singles. Maybe this needs a little tuning after all.

All theory but no examples, huh. Well, let's get to one!

THE RESULTS
We'll be taking last year's ranking as example, just so you know. Before this, note that the images below follow the column order of - after album titles and artist name - total track count (a full album has ~12 tracks, a mini-album has about half), rating as of original approach, number of new songs given in the year (in list method), number of songs already in Klassic Note before year of review, the subtraction of the previous two values, and its percentage with respect to number of available new songs (initiation % method). Also, click to zoom in the images to see everything, yes?

As I assume you have seen this in the results site if you ever went there, here are the Top Albums of Klassic Note 2016, but with more detail. This original method was sorted according to "overall rating %" column (in yellow) and then "total tracks" column in descending album size.
As mentioned above, the original method suffers from two major problems. Firstly, best albums with only one or two new songs given are given priority (red font "in list" column) instead of other releases with more new songs given. Second, large albums which have many new/old songs not given are all included, thus is always higher if those tracks are counted even though they do not contribute much to the year's song list (red font "initiation %" column).

The "in list" method was sorted according to the "in list" column (in yellow) and then "total tracks" column.
As you can see, new albums have surfaced (red font on "initiation" column, shows 0 old songs) and revised and best albums have mostly left the ranking. This is unhealthy as it ignores revised albums that have been given most of their tracks (Aki-chan's album has dropped to 7th because of this) and albums with smaller track count, which removes ALL mini-albums. ("total tracks" less than 6 in the original list)

The point with the former problem here is that revised albums are not credited on the number of new songs available in the track list, and dependent on the number of songs from singles that are available and how many I have given before. Therefore giving new songs will have a smaller limit compared to new full albums. Now you see the problem.

The "initiation %" method was sorted according to "initiation (%)" column and then "total tracks" column.
In this case, Aki-chan's album returns back to first place because I've only not given one of the new songs, thus its Initiation Percentage is very high. This also helps mini-albums come back in the ranks (see "total tracks" column) and help maintain new, non-revised albums ("initiation" column = 0) to stay in the ranking. This is because revised albums are calculated the same way as a new mini-album would, and thus there is bias for lower track count, making them stand out more. So for Aki-chan's album and Miyawaki Shion's mini-album, the latter has ranked higher than many because of its size compared to the best albums (2 given out of 3 new songs). Problem now is the full albums mentioned in the earlier method are now not present, and the ratio of old to new songs is amplified (3 new songs compared to 12 new songs) greatly resulting in polar differences between full albums and revised mini-albums.


THE CONCLUSION
You may be as confused as I do, so let me sum it all up. The original approach of just comparing number of songs given out of the album's total tracks is not adequate to represent which are the better albums of the year because some of them have already been given in earlier years. The "in list" method takes care of promoting new full albums but ignores revised albums and mini-albums in general. The "initiation %" method then is able to recover those releases back, but still ignores for the most, newly reviewed full albums. Both however, have effectively removed best albums from the competition (due to their high track count) and failed to acknowledge some exceptions that can make a large difference due to the main factor, album track count/album size. So yeah, it's rather complicated, you see.

I established a statement that in order to know which albums are better, I talk about the number of new songs I have NOT YET given to be accounted for, so this has bias to the first method mentioned. But I'm inclined to the second one as I know this is a solution to compensate for the repeated tracks we see in Japanese revised albums, but would be unfair to mini-albums.

I will not make a decision right here right now, but I'll like you to think of what makes of giving a good album. Is liking the whole album important to you or giving many songs in the album you've never anything from inside before already the trigger for you to give the whole release? Klassic Note has taken such an apprehensive approach over the past 10 years in order to ensure maximum satisfaction because of how expensive the releases are, but in recent years best albums have unnecessarily taken up the charts. So, what do you think? I will not reveal my answer here for now, for I will do so when #KN10YEARS draws closer. Until then, I suppose. But I think I gave it away somewhere in this post, you can get a rough idea what I mean xD


P.S. I'll be doing this more often and posting on Saturdays instead, so stay tuned. Cya guys tonight for Entertainment News.

---

Copyright (c) 2014-2024 Klassic Note Web Reports