Author Topic: [LEL] Visualization of LEL results  (Read 11877 times)

red marley

[LEL] Visualization of LEL results
« on: August 05, 2013, 04:54:25 pm »
Now that Greenbank has kindly provided a database of all control times, I've created a visualization of every rider's 'position graph'. This is just the accumulation of each rider's relative position at each control. The purpose of the visualization is to show where there were greatest changes in riders' relative times, due largely to sleep stops.


(see also this high resolution version for dowload).

The slope of the bottom of the graph shows the attrition rate. Surprisingly (to me at least), that would appear to be greatest between Barnard Castle and Brampton northbound. Perhaps less surprisingly, Moffat to Edinburgh has one of the lowest attrition rates but Edinburgh to Traquair one of the highest. It's also quite high between St Ives and Great Easton, presumably as riders hit the cutoff time.

The fastest riders and to a large extent, the slowest ones tend not to change position much (as seen by parallel lines without many crossings towards the top and bottom of the chart). The vertical strips with most crossings show where sleep stops have resulted in changes in rider order. So the darkest band between Pocklington and Thirsk northbound show where different riders have chosen to sleep in those two controls. Similarly, Eskdalemuir to Brampton southbound and Pocklington to Market Rasen southbound show where there have been least changes. Market Rasen to Kirton southbound is interesting in that the first half of the field remain quite stable, but the second half involve lots of swapping. I guess this is because the faster riders were able to finish on the Wednesday evening or very early Thurs am, but many of the the slower ones went for a sleep stop.

I'm sure those who have ridden it will be able to make better interpretations, but I thought it would be interesting to share.

London Edinburgh London

Re: Visualization of LEL results
« Reply #1 on: August 05, 2013, 05:07:19 pm »
Wow! Very pretty.

Barnard Castle to Brampton is a tough section, with Yad Moss to contend with. Moffat to Edinburgh is a really easy section, despite the big climb out of Moffat. I saw many riders manage this in the dark.

St Ives to Great Easton is a fiendish section, which most rode in boiling hot weather.

Reg.T

  • "You don't have to go fast; you just have to go."
Re: Visualization of LEL results
« Reply #2 on: August 05, 2013, 05:13:47 pm »
Thanks - that is good. I tend to concur with your observations.

Looking beyond relative position, it would be interesting to see a similar plot but with something like Time In Hand at each control as the vertical axis to see where the (virtual) bulges formed and how they progressed.

And/or perhaps a real-time version with a horizontal line set with the opening time of each control for the first group, IYSWIM.

Any volunteers?
Just turn me loose let me straddle my old saddle
Underneath the western skies

simonp

  • Omnomnomnipotent.
Re: Visualization of LEL results
« Reply #3 on: August 05, 2013, 05:15:47 pm »
Moffat => Edinburgh = easy section with tail wind.
Edinburgh => Traquair = tough section with head wind.


simonp

  • Omnomnomnipotent.
Re: Visualization of LEL results
« Reply #4 on: August 05, 2013, 05:17:56 pm »
I think that the histogram one someone in the US did for PBP was excellent.

Danial told me there were two bulges, and at Traquair I was between them. It would be interesting to see my position relative to those in some way.


frankly frankie

  • I kid you not
    • Fuchsiaphile
Re: Visualization of LEL results
« Reply #5 on: August 05, 2013, 05:32:00 pm »
I wonder how it would have looked with a headwind on the 1st day, instead of a tailwind.
How many people I wonder decide their first sleep stop before the ride, and how many wing it depending on conditions.
It's not dark yet but it's getting there.

Re: Visualization of LEL results
« Reply #6 on: August 05, 2013, 05:38:57 pm »
I think that the histogram one someone in the US did for PBP was excellent.

I'll have a look at the code to see how easy it is to adapt it for use with the LEL data.
"Yes please" said Squirrel "biscuits are our favourite things."

Phil W

Re: Visualization of LEL results
« Reply #7 on: August 05, 2013, 06:41:29 pm »
Brampton offers an easy train exit to Carlisle  or Newcastle and Scotland many associate with remoteness and harsher weather. I had serious stomach issues by Barnard Castle / Brampton, and on arriving at each control thoughts of abandoning entered my head. The trigger for my stomach deterioration was the extremely strong and sicky sweet black currant at Kirton, it just didn't agree with me, and although i switched back to water at MR the deterioration was already in motion. At Brampton,  decided to use up nearly all my buffer time to sleep and try and let my stomach recover, and it paid off. I'd already taken 2 hours rest at Barnard castle for the same issue. By Edinburgh I was eating properly again, and mentally I was in the place that I was going to complete it. I wouldn't underestimate the physiological boost that turning south gave riders. On the hills to Traquair I was loving it, and didn't really notice the headwinds others mention.

Thursday was brutal due to the heat, some GPS thermometers read 38c. So many riders had to stop at village shops to have extra drinks. I'd estimate the heat added at least 4 hours to my time, after St Ives, due to the need to stop, and the need to cycle slower.   I was stopping at every shop I saw along the way.

Re: Visualization of LEL results
« Reply #8 on: August 05, 2013, 06:48:13 pm »
Nice work, Jo.  Tufte would like it.

Re: Visualization of LEL results
« Reply #9 on: August 05, 2013, 07:00:23 pm »
A few tweaks to the data and the LEL logo could appear in the middle.

mattc

  • n.b. have grown beard since photo taken
    • Didcot Audaxes
Re: Visualization of LEL results
« Reply #10 on: August 05, 2013, 07:06:41 pm »
Great stuff Jo.

(Like Reg, I reckon Time-in-Hand would make an interesting y-axis - it's how I analysed my own staggerings on previous events.)

Danial told me there were two bulges, and at Traquair I was between them. It would be interesting to see my position relative to those in some way.
That's good to read. Our "predictor general" was tracking two predicted bulges coming down from Scotland, but shortly before they reached us it became one predicted bulge!

I didn't bother keeping his graphs as I expected the post-event analysis to be more revealing.

(My only guess-clamation was that the field got split around the 2nd night, one lot pressing on 1 control further to sleep. Maybe. )
Has never ridden RAAM
---------
No.11  Because of the great host of those who dislike the least appearance of "swank " when they travel the roads and lanes. - From Kuklos' 39 Articles

London Edinburgh London

Re: Visualization of LEL results
« Reply #11 on: August 06, 2013, 08:11:44 am »
Here's a link to anonymised data from the LEL rider tracker. Enjoy.

https://www.dropbox.com/s/bpxqq6wuqkkeqzb/anontracker.csv

Re: Visualization of LEL results
« Reply #12 on: August 06, 2013, 09:54:52 am »
Excellent, if anyone wants to do a time in hand graph stuff then the distances and closing times (according to the brevet card) for each control (in km and seconds since the start) were:-

Distances: 0 99 180 246 336 401 468 550 624 705 745 791 849 933 1000 1065 1150 1218 1299 1373 1419

Times: 0 29460 53400 73680 98760 118020 137820 162000 184080 208020 220620 234180 251400 276120 295800 315240 340320 360600 384540 406500 420000
"Yes please" said Squirrel "biscuits are our favourite things."

mattc

  • n.b. have grown beard since photo taken
    • Didcot Audaxes
Re: Visualization of LEL results
« Reply #13 on: August 06, 2013, 10:32:28 am »
Who can tell this idiot what cols AB and AC are?
Has never ridden RAAM
---------
No.11  Because of the great host of those who dislike the least appearance of "swank " when they travel the roads and lanes. - From Kuklos' 39 Articles

red marley

Re: Visualization of LEL results
« Reply #14 on: August 06, 2013, 10:32:56 am »
Here's the time-in-hand chart for all riders:



and a high resolution version for download.

What is interesting to me is that if you look at the speed differences between the faster and slower riders in the first sections up to Market Rasen, they don't count for that much of a time difference. If you were to project them forward to the end of the ride, they would account for maybe a third of the time variation. The bigger differences come with time spent at controls and speed in the second half of the ride.

marcusjb

  • Full of bon courage.
    • Occasional wittering
Re: Visualization of LEL results
« Reply #15 on: August 06, 2013, 10:43:48 am »
Interesting stuff - impressed with how quickly you made these things (but it being your area of expertise I believe, I suspect that it's a fairly simple task for you!).

Ignoring the one bump at Brampton, the fast boys clearly have the smoothest lines due to little stopping time and decent pacing etc. - quite amazing. 

There's also 1 rider who went off like a rocket and had more time in hand than everyone up until Thirsk before sinking back into the pack.

And as for the rider who started way out of time at St Ives and never packed (the line goes all the way to Loughton at about -6 hours) - I assume they set off very late or something (as they were -20ish hours at St. Ives!)? 

But it does show where many riders chose to sleep and has a lot of value for future rides.
Right! What's next?

Ooooh. That sounds like a daft idea.  I am in!

Re: Visualization of LEL results
« Reply #16 on: August 06, 2013, 10:45:14 am »
Who can tell this idiot what cols AB and AC are?

AB and AC are times that they dropped out. The last row matches the times of rider C62. Notice s/he is marked down as dropping out twice.
"Yes please" said Squirrel "biscuits are our favourite things."

Re: Visualization of LEL results
« Reply #17 on: August 06, 2013, 10:50:30 am »
And as for the rider who started way out of time at St Ives and never packed (the line goes all the way to Loughton at about -6 hours) - I assume they set off very late or something (as they were -20ish hours at St. Ives!)? 

Also note that one rider (rider_id=1136 or D31) had their start time put in 24 hours early which is making that graph go down quite a bit more than it needs to. In reality they finished with 18 hours in hand.
"Yes please" said Squirrel "biscuits are our favourite things."

red marley

Re: Visualization of LEL results
« Reply #18 on: August 06, 2013, 11:00:37 am »
Interesting stuff - impressed with how quickly you made these things (but it being your area of expertise I believe, I suspect that it's a fairly simple task for you!).

I produced the graphics with Processing - very good for producing beautiful and interactive graphics. It requires a bit of programming ability, but not that much, being designed originally for digital artists who had previously no programming experience.

simonp

  • Omnomnomnipotent.
Re: Visualization of LEL results
« Reply #19 on: August 06, 2013, 11:23:26 am »
Is it runnable via command line? The output is much prettier than I’ve achieved using gnuplot.


red marley

Re: Visualization of LEL results
« Reply #20 on: August 06, 2013, 11:52:30 am »

Re: Visualization of LEL results
« Reply #21 on: August 06, 2013, 12:06:57 pm »
It would be amazing to be able to see your own individual result in a different colour over the top of the 'cloud'.

Anyone magical enough to make that work?

:)

Re: Visualization of LEL results
« Reply #22 on: August 06, 2013, 12:53:00 pm »
It would be amazing to be able to see your own individual result in a different colour over the top of the 'cloud'.

Anyone magical enough to make that work?

If jo wants to make 1000 odd graphs, one for each rider with their line highlighted in red, then I'll gladly host the resulting (non-hi-res) images (~250KB each isn't a problem). Same for the 'relative position' graphs.

Like simonp I only know how to drive gnuplot to make far less pretty graphs. I suppose I should take this opportunity to learn Processing...
"Yes please" said Squirrel "biscuits are our favourite things."

mattc

  • n.b. have grown beard since photo taken
    • Didcot Audaxes
Re: Visualization of LEL results
« Reply #23 on: August 06, 2013, 12:56:04 pm »
If it's just for Simon_'s wall (framed with his medal and photo) I would recommend a felt-tip as the quickest solution.
Has never ridden RAAM
---------
No.11  Because of the great host of those who dislike the least appearance of "swank " when they travel the roads and lanes. - From Kuklos' 39 Articles

red marley

Re: Visualization of LEL results
« Reply #24 on: August 06, 2013, 01:25:20 pm »
I've updated my web-page version to allow you to overlay your own profile on the blue 'cloud' of other riders. This works both for the position chart and the time-in-hand chart. To use, you need to customise the parameters to the web page:

id=XX to show rider XX's profile.
&all=true to show all riders
&co=true to show the time-in-hand chart



So for example, to show rider G28's time in hand in comparison to all other riders:

http://gicentre.org/lel2013/?id=g28&all=true&co=true

To show rider V34's position relative to all others:

http://gicentre.org/lel2013/?id=v34&all=true