Do you have something to say about Prog...
...and would you like to submit it as material for this blog, apply for membership at the ProgRockin blog planning Google group.
Recent Comments

Tuesday, March 5, 2013

Interesting trends in Progulus Radio ratings data

I've been taking a close look at various aspects of the user ratings data on Progulus recently. Here's the first of a few posts about my findings. What I was interested in finding out was what the best years for prog were. Was there a "golden age" of the 2000's modern prog like there was with the golden era of prog in the 70's?

First, a digression into the method that I use to rate albums on the station. What I did first was to take the raw average album scores based on user ratings for each song and obtain the bayesian average. Why baysian averaging? Well using this method allows you to weight ratings based on the number of accumulated votes. For example, an album with only two "5" votes is really not a "5" rated album when compared to another album that has accumulated 100 "5" votes. Only after an album gets sufficient votes will the album's true rating become apparent. The bayesian formula takes this into account and adjusts the rating of albums with fewer votes closer to the mean.

After I adjusted the user ratings for each album, what I did next was to rank them all from highest to lowest in sequence from 1 to 4,500+, up to the total number of rated albums. In case you're wondering, the #1 ranked album on the station is Dream Theater - Images and Words. No big surprise there. Next, I took the albums from each year and looked at the top 50 albums of that year individually. One could tally an average rating for each year from those 50 CDs and then plot them on a graph, but what I did was to simply sum up the overall rankings of those 50 albums for each year. I inverted the result so that a higher score is better, as it's what we are used to seeing on graphs. Here's the result:

20130304040904

A couple of things to note on the result. First, as you go back in years to the early 90's, there are fewer and fewer representative albums. Before 1994 there were fewer than 50 albums for each year so they couldn't be included in the poll. Compare this to the late 2000's where there were upwards of 200 - 250 albums each year. I believe there is a direct correlation between the number of albums in the pool and the number of highly rated albums. Granted, Progulus does not have every possible album for a given year available, but I'd like to think that we have all of the most important albums. So by this reasoning, there are simply more albums available due to the popularity of the genre, and as a result we have a certain percentage of highly rated albums out of that pool to draw from.

Second, the graph drops sharply as we get to the present time. Fortunately for us I don't believe that this is wholly the result of a tail off in music quality over the last 3 years. Part of the reason goes back to the bayesian averaging. Newer albums have fewer votes, so it may just take a few years to accumulate enough votes to adjust the rating up away from the mean. I don't think this is the whole picture though. I have a hypothesis that people tend to rate music lower that they are less familiar with. To test this theory I would have to look at the raw user ratings going back a few years. That may be the subject of a future article.

The real finding here is that we started to hit our "peak" or golden age somewhere around 2004. This result wasn't surprising to me since I always personally considered 2004 as a banner year for prog, and a year that really opened the flood gates and inspired many new bands who would follow. Where the surprise was for me was that this peak continued well into 2008 - 2009, with 2007 marking the high point in the annual ranking. Lets take a look at the top 10 rated albums for 2007 as an example:

7, Symphony X- Paradise Lost, US
20, Porcupine Tree- Fear Of A Blank Planet, UK
26, Redemption- The Origins Of Ruin, US
30, Circus Maximus- Isolate, NO
44, Riverside- Rapid Eye Movement, PL
55, Myrath- Hope, TN
56, Spheric Universe Experience- Anima, FR
79, Dream Theater- Systematic Chaos, US
84, Dead Soul Tribe- A Lullaby For The Devil, US
87, Sieges Even- Paramount, DE

The number listed before each album is it's overall rank on the station. The #10 album that year is still in the top 100 of all albums. The only other year that I looked at that came close to this was 2005. In 2008 the #10 album ranked #191 overall, and in 2006 the #10 album ranked #228 overall. We can take a look at this in more depth with the following graph that shows the total number of top 10 albums each year that presently rank in the current overall top 100:

20130305091143

To be in the overall top 100 rank an album has to be in the highest tier, or better than about 98% of the other 4500+ albums on the radio station. Looking at the result, 2005 and 2007 both had the highest number of "hit-producing" years with all 10 of the top 10 CDs falling within the overall top 100 ranking. While 2000, 2002, and 2004 did produce a quite a few albums that rank in the top 100, the indication from the first graph is that a lot of other CDs that came out those years ranked quite a bit lower than the top 50 did in 2004 - 2009. It's also interesting to note here that there are a lot of peaks that seem to happen every other year.

We may need to wait a few more years to find out if 2009 - 2012 is really marking a downward swing in the quality of music coming out, or if it's simply an error in the way the data was analyzed as I described above. But it's clear from the data that the modern "golden era" of prog began to peak around 2004 and continues at least into 2008-2009, if not further.

0 comments :