Discussion in 'Linkin Park Chat' started by Minus, Jun 20, 2012.
Also, what's your favorite LP album? If it's LT, list LT and your second favorite album.
Roads Untraveled, I'll Be Gone
Lost In The Echo, Powerless, Tinfoil, Castle Of Glass
In My Remains, Lies Greed Misery, Victimized, Skin To Bone
Until It Breaks, Burn It Down
At least for now..
Needs favorite album.
Well, LITE is in top 4 so i guess that's ok. we just need lots more voters because LITE was getting lost of love in the favorite song thread. But CoG being #1! Yes!
After seeing the spreadsheet, I would like to officially petition to have a different color for my name box, for being a special snowflake and stuff.
This might be getting a little statistics nerdy, but perhaps instead of using an average we should be using median rather than the mean to limit the effects of outliers? When I'm looking at how the data is being represented it seems to me that the extreme negative answers are disproportionately affecting the sample. When I use your data as of the last available spreadsheet and I calculate for the median of LITE, this is the result I get the median as 5.
This seems to be far more representative of the community's response to the song (I think we can agree that there is an awful lot of love for this song) in general and would remove any deviations from dragging down the average or causing significant damage. It seems to me that calculating the median of the set would produce a more accurate single number, and of course we can run some simple math to create a standard deviation. So we'll be able to say that the track is rated a 5 with a std. dev. of 1.55. In this way not only do we get a fairly accurate representation of the community's love for the song but we have an accurate representation of how contentious that love is. For extra geekiness but easy readability we could do some more very basic math and convert the std. deviation into a percentage since that's easier to understand for most people.
What do you think?
Using the example of Roads Untraveled: Our friend here would probably not actually affect the outcome at all because his outlying response would be effectively ignored by a median measurement but he WOULD affect the standard deviation slightly. This method would benefit from a larger data set for sure.
It's ok If you pull a me, sometimes its funny to see people overreact to it. But for this type of thing it would make since to stagger it
Also, I have the second highest album rating besides Dragonfly who gave 5's to everything. I just find that interesting.
Whenever you have a small sample size, any extreme outliers are of course going to have disproportionate effect on the overall rating. That's where the idea I proposed for the X-axis of the graph comes in. Using the same example I gave in my earlier post about it - consider a song that has received 9 votes for B Tier (2 points) and 3 votes for God Tier (5 points). The mean would be 2.75, rounded up to 3 in the Tier List (A Tier), but the majority of the votes are lower than that.
With those votes, on the graphic Minus makes the song would still remain in A Tier on the Y-axis, but would trend toward the negative on the X-axis, since that's how most people voted it. In other words, it's on the extreme low end of A Tier.
Right, but I feel a rating that says "Most people voted this song as a 5, but some suggested it was a 4 or a 3" is more accurate than trying to produce a single number which takes all these factors into account. As a consequence we can get results from a relatively small sample size which accurate reflect the community's general response. While I like the idea of an X axis it doesn't change the way that the disproportionately negative votes are dragging down the album as a whole. Even the people who vote one or two songs as a negative generally have most of their songs sitting in the SS or S tiers but because the negatives trend towards extreme they are disproportionately affecting the general ratings for everything. I don't see how an X axis can solve the problem in the aggregate BUT a median/stddev combo WOULD accurately reflect that data.
Added I'll Be Gone, sorry!
Fav. album is ATS
Roads Untraveled, Castle of Glass
In My Remains, Powerless, Tinfoil, Lost In The Echo, Victimized, I'll Be Gone
Skin To Bone, Lies Greed Misery, Until It Breaks, Burn It Down
I agree with maybe changing it somehow like _depression or adam.posey are saying.
If you compare this chart with the answer to this poll: http://www.lpassociation.com/forums/showthread.php?t=36822&page=3
It seems really off because LITE has the most votes. Those 1 votes really bring it down. I mean doing this helps us see things but If anything LITE seems to be along with CoG as the #1 LT song, which this chart doesn't represent
EDIT: you guys are working on a fix in the shoutbox.
The fix has been implemented. I'll add the standard deviation later.
Coolio. I was getting a little confused about Lost in the Echo being in two places there. Cheers for the fix
For some reason, it doesn't say a PM has been sent, so I'll write it here.
My favourite album is Hybrid Theory. Full list: Hybrid Theory > Living Things > A Thousand Suns > Meteora > Minutes To Midnight. But they all rule
LOST IN THE ECHO, IN MY REMAINS, I'LL BE GONE, CASTLE OF GLASS, ROADS UNTRAVELED, SKIN TO BONE, TINFOIL, POWERLESS
BURN IT DOWN, LIES GREED MISERY
- - -
- - -
VICTIMIZED, UNTIL IT BREAKS
- - -
Album Tier - GOD TIER
Fav. Album - LT/ATS
You don't choose your album tier. That's calculated in the first post as a function of EVERYONE'S score. Just score the individual songs, please.
Hmm. I did pm you my favorite album but I didn't my second. So I guess it's ATS for my favorite and a tie for reanimation and LT as my second.
In My Remains
Castle of Glass, Powerless, Skin To Bone, Victimized
Tinfoil, Lost In The Echo, Roads Untraveled, I'll Be Gone
Until It breaks, Burn It Down, Lies Greed Misery
Favorite Album: Hybrid Theory
Separate names with a comma.