|
calculated stats based on FINAL results
|
| TankNL |
Posted on 05-08-2014 00:31
|
Domestique
Posts: 440
Joined: 19-03-2007
PCM$: 200.00
|
dacolik wrote:
because of this year's results Contador should get the best MO... This is really bad database...
ignored.
Edited by TankNL on 05-08-2014 00:31
|
| |
|
|
| dacolik |
Posted on 05-08-2014 00:32
|
Junior Rider
Posts: 43
Joined: 21-06-2013
PCM$: 200.00
|
It just looks like you don't watch cycling enough  ) |
| |
|
|
| Ad Bot |
Posted on 22-11-2024 00:58
|
Bot Agent
Posts: Countless
Joined: 23.11.09
|
|
| IP: None |
|
|
| TankNL |
Posted on 05-08-2014 00:32
|
Domestique
Posts: 440
Joined: 19-03-2007
PCM$: 200.00
|
TankNL wrote:
I created a mapping with the PackITA to see how the stats would be altered if I used the created stats from actual results. Every stat column is followed by my stat (in orange) and the followed by the difference (in green).
https://dl.dropbo...20map.xlsx
I haven't created a .cdb, but that can easily done.
I did also 1 other thing (something I also did previously for the PCMDaily database); the popularity of a cyclist. I calculated the popularity based on the CQ points collected over the last years. This can easily be altered in the database so that the sponsor reacts correctly to the riders in the team.
I will do the A.S.O. DB 2014 of OlegTinkov next.
Tank
Crap, I found out that in the copying and pasting, some of the mapping has been lost. Will have to re-do it tomorrow. the link can not be used right now. |
| |
|
|
| TankNL |
Posted on 05-08-2014 00:36
|
Domestique
Posts: 440
Joined: 19-03-2007
PCM$: 200.00
|
dacolik wrote:
It just looks like you don't watch cycling enough )
have you even read my posts? |
| |
|
|
| Paul23 |
Posted on 05-08-2014 20:25
|
Grand Tour Specialist
Posts: 4411
Joined: 10-08-2011
PCM$: 400.00
|
TankNL wrote:
first off, it is not a database.
second... it only the number of points scored in races of a certain kind by a rider.
And yes the races younger the 365 are counted times 4, the races older the 720 are counted once and between get 2 points.
Then when you look at the results in this year... these are the figures... not saying that this should also be the stats, but... those are the results.
as for SAGAN;
| category | date | race_name | race_type | result | COBBLE | | GT1s | 9-7-2014 | Tour de France, Stage 5 : Ieper (Bel) - Arenberg (156 km) | Cobbles | 4 | 173,6 | | 1.WT1 | 13-4-2014 | Paris - Roubaix (257 km) | Cobbles | 6 | 280 | | 1.WT1 | 6-4-2014 | Ronde van Vlaanderen (259 km) | CobbleHills | 16 | 70,4 | | 2.HCs | 2-4-2014 | Driedaagse De Panne-Koksijde, Stage 2 : Zottegem - Koksijde (206 km) | CobbleSprint | DNF | 0 | | 2.HCs | 1-4-2014 | Driedaagse De Panne-Koksijde, Stage 1 : De Panne - Zottegem (201 km) | CobbleSprint | 1 | 39,2 | | 1.WT2 | 30-3-2014 | Gent - Wevelgem (233 km) | CobbleSprint | 3 | 156,8 | | 1.WT2 | 28-3-2014 | E3 Prijs Vlaanderen - Harelbeke (211 km) | CobbleSprint | 1 | 308 | | 1.WT1 | 31-3-2013 | Ronde van Vlaanderen (256,2 km) | CobbleHills | 2 | 187 | | 1.WT2 | 24-3-2013 | Gent - Wevelgem (185 km) | CobbleSprint | 1 | 154 | | 1.WT2 | 22-3-2013 | E3 Prijs Vlaanderen - Harelbeke (211 km) | CobbleHills | 2 | 145,2 | | 2.HCs | 26-3-2013 | Driedaagse De Panne-Koksijde, Stage 1 : Middelkerke - Zottegem (199,8 km) | CobbleHills | 1 | 30,8 | | 1.1 | 2-3-2013 | Strade Bianche (188 km) | CobbleHills | 2 | 52,8 |
ok...now it makes more sense...thank you for taking time to answer 
|
| |
|
|
| TankNL |
Posted on 05-08-2014 20:33
|
Domestique
Posts: 440
Joined: 19-03-2007
PCM$: 200.00
|
TankNL wrote:
TankNL wrote:
I created a mapping with the PackITA to see how the stats would be altered if I used the created stats from actual results. Every stat column is followed by my stat (in orange) and the followed by the difference (in green).
https://dl.dropbo...20map.xlsx
I haven't created a .cdb, but that can easily done.
I did also 1 other thing (something I also did previously for the PCMDaily database); the popularity of a cyclist. I calculated the popularity based on the CQ points collected over the last years. This can easily be altered in the database so that the sponsor reacts correctly to the riders in the team.
I will do the A.S.O. DB 2014 of OlegTinkov next.
Tank
Crap, I found out that in the copying and pasting, some of the mapping has been lost. Will have to re-do it tomorrow. the link can not be used right now.
Ok,
redid my the mapping...
https://dl.dropbo...%20v2.xlsx
so now the popularity should be good... (for instance Vanmarke goes from 28 to 77 and more top riders get proper popularity stats).
Will start to work on the mapping with the A.S.O. DB of OlegTinkov |
| |
|
|
| TankNL |
Posted on 07-08-2014 23:22
|
Domestique
Posts: 440
Joined: 19-03-2007
PCM$: 200.00
|
Also did a mapping for the ASO database.
https://dl.dropbo...%20v1.xlsx
however, found out that I missed half of 2012 in copying and pasting again... so added that. This now is fine, but that brings the file up to 40Mb... oh well.
Same here goes for the popularity of the riders. The stats are not something that can be copied 1-on-1... As earlier discussions mentioned, domestics are not good rated (especially the mountain and hill), because they don't score points, but they are still very capable and that has to reflect in the stats...
So the stats could be used to reflect on the capebilities of a rider, as all his results are there, but it doesn't tell the whole story.
Edited by TankNL on 07-08-2014 23:24
|
| |
|
|
| sammyt93 |
Posted on 07-08-2014 23:29
|
Classics Specialist
Posts: 3634
Joined: 03-07-2012
PCM$: 300.00
|
TankNL wrote:
Also did a mapping for the ASO database.
https://dl.dropbo...%20v1.xlsx
however, found out that I missed half of 2012 in copying and pasting again... so added that. This now is fine, but that brings the file up to 40Mb... oh well.
Same here goes for the popularity of the riders. The stats are not something that can be copied 1-on-1... As earlier discussions mentioned, domestics are not good rated (especially the mountain and hill), because they don't score points, but they are still very capable and that has to reflect in the stats...
So the stats could be used to reflect on the capebilities of a rider, as all his results are there, but it doesn't tell the whole story.
A couple of ideas would be to give domestiques X% of the points scored by their leader in a race so that their input is modelled in some way.
And to possibly include some sort of points coefficient based on the strength of the field to multiply the points by. (though this would probably be really hard to get right)
|
| |
|
|
| TankNL |
Posted on 08-08-2014 00:12
|
Domestique
Posts: 440
Joined: 19-03-2007
PCM$: 200.00
|
sammyt93 wrote:
A couple of ideas would be to give domestiques X% of the points scored by their leader in a race so that their input is modelled in some way.
And to possibly include some sort of points coefficient based on the strength of the field to multiply the points by. (though this would probably be really hard to get right)
Thanks for the input.
I like the idea of giving a domestique a partial of the points collected by a race leader, or by the best man of his team in that stage... could be do-able i would say. It has some challenges though... just watched the Tour of Utah and it could apply to stage 5 of the Tour de France.
What to do with the likes of Anacona or Fuglesang, he is a helper, though scoring points on his own. Well, I will just have to say if the riders doesn't score points, that he gets 25% or something like that of the highest scoring team member... or even maybe the total scored points from the team. Will have to build a piece of code to calculate that... will have to look into that.
And the second part I have thought of also, but is near impossible to implement. For instance... Every time i load a race, I could calculate the total rider value of the participants, but that doesn't tell me if the battle is hard. Sprinters with a lot of points in a mountain stage... and I can only look back... so Wilco Kelderman beating Nibali in the Critérium du Dauphiné this year... didn't count for much, as Nibali scored most of his points in the Tour... will be very difficult... |
| |
|
|
| sammyt93 |
Posted on 08-08-2014 00:26
|
Classics Specialist
Posts: 3634
Joined: 03-07-2012
PCM$: 300.00
|
TankNL wrote:
Thanks for the input.
I like the idea of giving a domestique a partial of the points collected by a race leader, or by the best man of his team in that stage... could be do-able i would say. It has some challenges though... just watched the Tour of Utah and it could apply to stage 5 of the Tour de France.
What to do with the likes of Anacona or Fuglesang, he is a helper, though scoring points on his own. Well, I will just have to say if the riders doesn't score points, that he gets 25% or something like that of the highest scoring team member... or even maybe the total scored points from the team. Will have to build a piece of code to calculate that... will have to look into that.
hmm, maybe if a teammate that doesn't score gets 10% then one that does score should get his position finish plus 10% of the difference between his points and his leaders points. (only used 10% as it's an easy number to work with, no idea how much it should be to give teammates realistic points/stats)
e.g. If it was 10% then I think Bardet would get 118.5 points instead of 110 from Peraud also taking second for the Tour GC.
TankNL wrote:
And the second part I have thought of also, but is near impossible to implement. For instance... Every time i load a race, I could calculate the total rider value of the participants, but that doesn't tell me if the battle is hard. Sprinters with a lot of points in a mountain stage... and I can only look back... so Wilco Kelderman beating Nibali in the Critérium du Dauphiné this year... didn't count for much, as Nibali scored most of his points in the Tour... will be very difficult...
As far as I under stand it you can break down the points scored by race type, if so then it should only take into account points scored on the same terrain as the race is on. So Nibali wouldn't get a bonus for winning a tour mountain stage featuring Kittel, Kristoff, Greipel etc as it would only take into account the points they scored on mountain stages.
But they wouldn't get a massive bonus from Nibali's value on a sprint stage as it would only count Nibali's points scored on sprint stages instead of all his points.
I'd imagine after collating the points for each type of terrain for the races you could then use some sort of distribution curve for the percentage added once you've found your mean.
But that's about as far as my knowledge would get me.
If it worked like that would it make it any easier to implement or not?
Edited by sammyt93 on 08-08-2014 00:31
|
| |
|
|
| TankNL |
Posted on 08-08-2014 08:38
|
Domestique
Posts: 440
Joined: 19-03-2007
PCM$: 200.00
|
sammyt93 wrote:
TankNL wrote:
Thanks for the input.
I like the idea of giving a domestique a partial of the points collected by a race leader, or by the best man of his team in that stage... could be do-able i would say. It has some challenges though... just watched the Tour of Utah and it could apply to stage 5 of the Tour de France.
What to do with the likes of Anacona or Fuglesang, he is a helper, though scoring points on his own. Well, I will just have to say if the riders doesn't score points, that he gets 25% or something like that of the highest scoring team member... or even maybe the total scored points from the team. Will have to build a piece of code to calculate that... will have to look into that.
hmm, maybe if a teammate that doesn't score gets 10% then one that does score should get his position finish plus 10% of the difference between his points and his leaders points. (only used 10% as it's an easy number to work with, no idea how much it should be to give teammates realistic points/stats)
e.g. If it was 10% then I think Bardet would get 118.5 points instead of 110 from Peraud also taking second for the Tour GC.
TankNL wrote:
And the second part I have thought of also, but is near impossible to implement. For instance... Every time i load a race, I could calculate the total rider value of the participants, but that doesn't tell me if the battle is hard. Sprinters with a lot of points in a mountain stage... and I can only look back... so Wilco Kelderman beating Nibali in the Critérium du Dauphiné this year... didn't count for much, as Nibali scored most of his points in the Tour... will be very difficult...
As far as I under stand it you can break down the points scored by race type, if so then it should only take into account points scored on the same terrain as the race is on. So Nibali wouldn't get a bonus for winning a tour mountain stage featuring Kittel, Kristoff, Greipel etc as it would only take into account the points they scored on mountain stages.
But they wouldn't get a massive bonus from Nibali's value on a sprint stage as it would only count Nibali's points scored on sprint stages instead of all his points.
I'd imagine after collating the points for each type of terrain for the races you could then use some sort of distribution curve for the percentage added once you've found your mean.
But that's about as far as my knowledge would get me.
If it worked like that would it make it any easier to implement or not?
I think you are on the right track, but the algorithm to calculate the stats is now done by adding al the points and then comparing them to the other cyclists to get a stats value PCM style. If I were to implement such a routine, them it would be a much heavier to process all the results. Bear in mind that all the results from 2012 until now already is 140.000 rows....
I think I will first go for the domestics solution to play around with that. Doing that would also mean that I need the team of the rider in the result (which I now ignored and therefor not stored). So that also would mean reprocessing all the results... Will do some test and than will come back here.
Was thinking, could do something like domestiques on stages get a % of all he collected points accept acceleration, sprint, stamina, resistance and recuperation. Also ITT stages should be excluded.
Edited by TankNL on 08-08-2014 09:53
|
| |
|
|
| TankNL |
Posted on 08-08-2014 14:50
|
Domestique
Posts: 440
Joined: 19-03-2007
PCM$: 200.00
|
tried to work on the idea today, but as it turns out, the site that I'm using only posts partial results (top20 or top50) for stages of stage races below World Tour level. So can't use that for calculating the domestic points... Then it would be only the one day races and the WT stage races. |
| |
|
|
| Selwink |
Posted on 08-08-2014 15:01
|
Grand Tour Champion
Posts: 8856
Joined: 17-05-2012
PCM$: 200.00
|
Why not use ProCyclingStats? It's very accurate and quite quick as well
|
| |
|
|
| MARSUPILAMI |
Posted on 08-08-2014 15:04
|
Team Leader
Posts: 5597
Joined: 10-08-2013
PCM$: 300.00
|
TankNL, I would like to thank you your work. I have FDb 14 for PCM 13 and I'm using your Excel for editing stats. I didn't look at your formules and your stats, but I looked at last results. It's very useful!
(If you want my update of stats of FDB 14 for PCM 1t, tell me and I'll upload it)
|
| |
|
|
| TankNL |
Posted on 08-08-2014 20:41
|
Domestique
Posts: 440
Joined: 19-03-2007
PCM$: 200.00
|
@Selwink; if only It was so simple. I've created code to grab the results from www.cqranking.com because the have a certain logic in their site navigation. www.ProCyclingSta... is indeed very good, but is harder to program to grab. Would have to read the calendars, find the links to the races, get to the stage results from there etc... I'm not saying it is impossible, but much harder to do.
@MARSUPILAMI; thank you, good to hear. That is exactly what I have created it for.
Edited by TankNL on 09-08-2014 09:01
|
| |
|
|
| MARSUPILAMI |
Posted on 09-08-2014 08:41
|
Team Leader
Posts: 5597
Joined: 10-08-2013
PCM$: 300.00
|
I look results in your Excel because in PCM's computer I don't have Internet
|
| |
|
|
| TankNL |
Posted on 09-08-2014 15:09
|
Domestique
Posts: 440
Joined: 19-03-2007
PCM$: 200.00
|
TankNL wrote:
tried to work on the idea today, but as it turns out, the site that I'm using only posts partial results (top20 or top50) for stages of stage races below World Tour level. So can't use that for calculating the domestic points... Then it would be only the one day races and the WT stage races.
Kept on going with the idea. I still have all the WT results and single day races on which this can apply. So I altered the code to also include the team name alongside the rider. Now I am able to give the helpers also points, depending on how the "leader" has scored.
I altered the points table to also tell per race type what the percentage is of the scored points and if it implies on the whole result, or only the top x of the result. Because I figured that If your in the top50 of a sprint, that your a helper deserving the points, but If you are in the top150, than your not? up for discussion. Any way, that can be altered per race type. Also some of the race types have been put at 0% (like the TT).
Will post the new stats tonight. |
| |
|
|
| MARSUPILAMI |
Posted on 09-08-2014 15:15
|
Team Leader
Posts: 5597
Joined: 10-08-2013
PCM$: 300.00
|
I will download it!
Are there new results in the Excel?
|
| |
|
|
| TankNL |
Posted on 09-08-2014 19:57
|
Domestique
Posts: 440
Joined: 19-03-2007
PCM$: 200.00
|
MARSUPILAMI wrote:
I will download it!
Are there new results in the Excel?
Nope no new results. Just that I now give points to helpers if the leader is doing good, in order to improve the domestic helpers stats.
Still everything up until San Sebastián. Tomorrow I will add tour of Utah and the rest of tour of Portugal etc... Will doubt that the stats will be influenced much... Maybe Horner and Danielson will get a boost. Will post some changes, to see how adding a couple of races influences the calculated stats. |
| |
|
|
| TankNL |
Posted on 09-08-2014 19:57
|
Domestique
Posts: 440
Joined: 19-03-2007
PCM$: 200.00
|
MARSUPILAMI wrote:
I will download it!
Are there new results in the Excel?
Nope no new results. Just that I now give points to helpers if the leader is doing good, in order to improve the domestic helpers stats.
Still everything up until San Sebastián. Tomorrow I will add tour of Utah and the rest of tour of Portugal etc... Will doubt that the stats will be influenced much... Maybe Horner and Danielson will get a boost. Will post some changes, to see how adding a couple of races influences the calculated stats. |
| |
|
