Route Guide - iPhone / Android - Partners - Forum - Photos - Deals - What's New - School of Rock
Login with Facebook
 ADVANCED
MPs 50 classic climbs
View Latest Posts in This Forum or All Forums
   Page 2 of 2.  <<First   <Prev   1  2
Follow replies to this topic? Notify me at the top of web site.
1

Email me.
 
By toddgordon
From Joshua Tree, California
May 2, 2012

I only have 29 on the list of 50;.....I need to get busy. I hang my head in shame......................I shouldn't even call myself a rock climber...........


FLAG
By CJC
May 2, 2012

cool list but the algorithm needs some tweaking

so does the photo of the moment selector which requires only one 'great' vote to be included in the queue


FLAG
By Rob Dillon
May 2, 2012

Hang it low Todd, it's the only list where I I have more than you!


FLAG
By Will S
From Joshua Tree
May 2, 2012

Page view weighting leans to CO routes, since this site started as a CO-centric and those entries are older. Also probably leans toward moderates since more people have done them, seems way heavy on 5.8s.

But still, c'mon, no Rostrum?


FLAG
By Boissal
From Small Lake, UT
May 2, 2012

CJC wrote:
so does the photo of the moment selector which requires only one 'great' vote to be included in the queue

One downvote is all it takes to keep the choss and the buttshots out of the rotation FOREVER. I've been busy...


FLAG
By Conor Byrne
May 2, 2012

Rob Dillon wrote:
32/50. Obviously the algorithm skews towards easier climbs in Colorado, but hey, I live in Colorado and I don't climb hard! +1 For Lord of the Thais...overhanging broccoli stalks 300' over the beach! Interesting how Ancient Art's Stolen Chimney only gets 2 stars but is still on the list.



Stolen chimney gets 2 stars for you because i believe thats what you graded it. MP's global average grade is 3.8 stars out of 235 votes.


FLAG
By Doug Hemken
Administrator
May 2, 2012
On Everleigh Club Crack.  Photo by Burt Lindquist.

There might be a reason they chose to stratify their preference algorithm by State rather than do a global ranking ... if the global ranking is so obviously biased.

You could simply censor data before 5 years ago... but you would still have to weight for where the climbing population lives and vacations.

My question is, how have these preferences changed over time? Was Ancient Art in the top 50 list before Katie and Citi? What impact will Fred's new list have over the next 5 years?

John, glad you asked that question because Gokul & I were asking it just last night.


FLAG
By J. Thompson
From denver, co
May 2, 2012
Trundling a death block. Photo by Dan Gambino.

39!!!

Not bad!

Now how many have you done solo???

I've got 9...free and roped.

Kinda fun...

josh


FLAG
By Andy Novak
From Golden, Co
May 2, 2012
Living the High Life.

11 1/2 (Tr'ed Supercrack). What an amazing bucket list. Thanks MP!


FLAG
By JLP
From The Internet
May 2, 2012

Will S wrote:
But still, c'mon, no Rostrum?

Not sure if this feature was there before, but it seems you can now get the "50 Classics" by area. The Rostrum was in California, Yosemite and Yosemite Valley.

My stats...

CA 36
CO 39
WY 20
Yosemite 41
Yosemite Valley 39
Tuolumne 16
All 42

Pretty cool, thanks for the feature.


FLAG
By Nick Wilder
Site Landlord
From The Bubble
May 2, 2012
Personal Photo

I released all these features over a year ago... maybe we should have advertised them better!


FLAG
By Rob Dillon
May 4, 2012

Algorithms are funny things- go to the California page, pull up the Top 50, and right between Positive Vibrations and Astroman is...Coarse and Buggy?


FLAG
By berl
From Oregon
May 4, 2012

how about normalizing the votes or ticks to the routes or ticks per state?


FLAG
By Doug Hemken
Administrator
May 4, 2012
On Everleigh Club Crack.  Photo by Burt Lindquist.

Berl, not sure if you mean pick the 50 classics per State, which is where this all started, or if you mean to compare across States.

If you compare across States after normalizing for State popularity (b1*posts + b2*ticks + b3*stars + b4*views), you end up picking 50 routes that stand way above their in-State peers. So at the extreme, an area/State with one classic route and a bunch of choss (Illinois, say) would pop onto the list, and all those Yosemite routes would disappear because they don't really stand that far above their neighbors?

As a side note, once you get down to smaller areas/units, then you start getting "less intuitive" results again, just like if you try to compare across too much geography.


FLAG
By JLP
From The Internet
May 4, 2012

You have to define what you are looking for first - something better than "satisfy my intuition".

Popularity = classic = page views isn't a bad formula, IMO. Next, consider this is a basically a CO website.

The lists generated make sense enough to me.

Holy Cow - you claim there's something worth climbing in IL that should replace one of the routes on the "All" list? You Midwesterners never quit. Your "intuition" is warped.


FLAG
By Doug Hemken
Administrator
May 4, 2012
On Everleigh Club Crack.  Photo by Burt Lindquist.

JLP wrote:
Holy Cow - you claim there's something worth climbing in IL that should replace one of the routes on the "All" list?


No, I'm telling you that if you normalize poorly you'll end up with "Ant Killer" knocking "Central Pillar of Frenzy" off the list. And you are telling me your intuition doesn't like that!

You need to think a little bit longer about what kind of data generates a large z-score.

You need some give and take between intuition and data analysis, for the benefit of both, IMO.


FLAG
By JLP
From The Internet
May 4, 2012

Read your post wrong. You Midwesterners are still warped, though.

I do like that page views can be counted here. This site has the database and the popularity. I think the lists generated have some merit. I don't see a need to attempt to artificially include climbs that don't get looked at.

I wonder if some "normalization" took place to distribute routes through the different grades? Seems like it - like a bell curve was applied - hence the # of 5.8's on the lists.

My only wish would be to see them also generated strictly per grade and/or per type (boulder, sport, trad).


FLAG
By Doug Hemken
Administrator
May 4, 2012
On Everleigh Club Crack.  Photo by Burt Lindquist.

Can't really argue with "warped."

I think the lists based on "good" data tell us a lot about people (us), while the "bad" lists tell us a lot about the limits of the data.


FLAG


Follow replies to this topic? Notify me at the top of web site.
1

Email me.
Page 2 of 2.  <<First   <Prev   1  2