preliminary statistical analysis of the international
play

Preliminary statistical analysis of the international eventing - PowerPoint PPT Presentation

Preliminary statistical analysis of the international eventing results 2014 Madrid 23/1/15 Overview of the talk Statistical analysis The analysis and the data Analysis of the fall data (related to a fence) Analysis of results


  1. Preliminary statistical analysis of the international eventing results 2014 Madrid 23/1/15

  2. Overview of the talk Statistical analysis • The analysis and the data • Analysis of the fall data (related to a fence) • Analysis of results data

  3. 1. The questions addressed in this analysis What are the factors that on their own and as a combination are correlated with good results, Refusals, EL, Horse Fall, Unseated Rider ? Age of horse and rider, barely qualified, previous results, competing at international level since when, average number of successful / unsuccessful competitions per year at that level, number of competition as a combination, … Are there parameters that significantly alter the probability of a fall? Homework: questions / interesting points raised during the seminar Possibility to distinguish XC profiles??

  4. The major steps in a statistical analysis • Define the interesting questions and factors, that might affect risk • Query the FEI database and compose file with data • Encode data for statistical analysis • Perform the analysis • Interpretation of the results

  5. Giuseppe’s slide

  6. 2. The data file Excel file with data on each international starter  161 000 rows and 650 columns Identifications, Date, Officials, … Profile of rider: Profile of competition age CCI or CIC # different horses Star-level # starts 1*, 2*, 3*, 4* SJ before XC # MER 1*, 2*, 3*, 4* Speed Profile of fall Distance 38 columns Profile horse: 977 rows age Result # different horses Dressage: pen # starts 1*, 2*, 3*, 4* XC: pen, EL (R, Fall) RET,WD # MER 1*, 2*, 3*, 4* SJ: pen, EL, RET,WD Low dimensional Multidimensional uncorrelated highly correlated PCA, FA dataset dataset File: 2004-2014 but focus on 2014

  7. Fall analysis

  8. Grouping of competitions CCIY1* CCIYH1* CICYH1* CCI1* CIC1* CCIJ1* CICJ1* CCIP1* CICY2* CCIP2* CIC2* CCI2* CICYH2* CCIU252* CCIY2* CIC3* CICO3*-NC CCIO3*-NC CCI3* CCIU253* CIC4* CCI4* Championship 1* Championship 2* WEG Asian Games-C1* CH-EU-P-C2* WEG-C C.Am+Caraib Games-CCI1* CH-EU-Y-CCI2* CH-Balkan-CCI1* CH-M-YH-CCI2* CH-EU-J-CCI1* CH-N.Am-Y-CCI2* CH-M-YH-CCI1* CH-S.Am-C2* CH-N.Am-J-CCI1*

  9. Competition Format and Horse Falls # starters # competitions # XC-FH % XC-FH # XC-FR % XC-FR CCI1* 3832 126 30 0.8 141 3.7 CCI2* 1667 73 25 1.5 69 4.1 CCI3* 642 22 22 3.4 48 7.5 CCI4* 308 6 20 6.5 30 9.7 6449 227 97 1.5 288 4.5 CIC1* 5421 146 40 0.7 180 3.3 CIC2* 4576 249 47 1.0 138 3.0 ? CIC3* 2314 80 38 1.6 90 3.9 12311 475 125 1.0 408 3.3 CH1* 221 6 9 4.1 17 7.7 CH2* 196 5 10 5.1 10 5.1 WEG 90 1 3 3.3 7 7.8 507 12 22 4.3 34 6.7 Normal competitions: - numbers of horse falls and unseated riders increase with star level; - approximately 3 times more unseated riders than horse falls Championships: - alarming number of falls in particular horse falls

  10. Falls and type of fences In total 912 falls related to fences Unseated Riders: 685 (75%) Horse Falls: 227 (25%) C. square spread E. brush F. round No normalization by the number of times the type of fence is jumped!

  11. Similar to 2013 In 2013 total 817 falls related to fences 2014 2013

  12. Unseated Riders: 685 (75%) Horse Falls: 227 (25%) More (as compared with FR) horse fall on type C less on type E Significant difference between the distribution of Unseated riders and horse falls (chi-square p=0.014)

  13. 2013 Distribution Unseated riders and horse falls over the fence types Relative more Horse Falls than Unseated Riders on type C Relative more Unseated Riders than Horse Falls on type G (corner) and J(step) Not in 2014

  14. Where does the fall occur: fence number ? Horse falls At the end?! Unseated athlete NB: 44 of the 974 falls are not related to fences

  15. Falls and ground conditions

  16. Other fall characteristics: frequencies of individual parameters FALSE TRUE valid Combination 370 542 912 Fence Type Option 901 11 912 Frangible 841 71 912 Portable 477 435 912 Related to water 728 246 974 Terrain 443 531 974 Ground Line 196 716 912 Off a Bend 629 345 974 Air Jacket worn 303 545 848 Air Jacket activated 452 504 956 Frangible break 5 Horse fell or tread on Rider 923 33 956 Accident Type 661 295 956 Horse slipped? 883 73 956 Wind 839 117 956 Poor Visibility 949 7 956 Horse refused? 674 238 912 Horse hit Fence? 411 563 974 Horse tipped portable? 4 Horse broke fence? 880 32 912 Horse somersaulted? 880 32 912 Inexperience 624 350 974 Rider fatigue 953 3 956 Horse out of control 937 19 956 Horse too fast 851 105 956 Horse too slow 887 69 956 Light situation 905 39 944 Horse impaired by health 953 3 956 Horse fatigued 936 20 956 Horse distracted 945 11 956

  17. Four (to 5) possible classes of falls related to a fence Component 1 2 3 4 5 Fall Fence Combination 0.513 Fall Fence Portable -0.572 Fall Fence Related to water 0.457 Fall Fence Ground line -0.341 Fall Fence Off a bend 0.468 -0.463 Did horse fall on or tread on athlete Did the horse slip Injury level 0.421 -0.316 Did horse refuse 0.364 -0.548 Did horse hit the fence 0.684 Did horse break the fence 0.448 Did horse somersault 0.345 Inexperience 0.435 Athlete impaired by fatigue Horse out of control -0.527 Horse too fast -0.524 0.444 Horse too slow 0.384 0.565 Horse fatigued Horse distracted 0.346 NB: falls not related to fence are not included in this analysis

  18. Falls in relation to other results

  19. Results in function of competition format 70.0% 60.0% 50.0% 40.0% Downward trend compensated by Falls, WD and RET 30.0% 20.0% 10.0% 0.0% CIC1* CIC2* CIC3* CCI1* CCI2* CCI3* CCI4* clear 66.3% 63.5% 59.2% 63.0% 58.8% 50.9% 40.5% 20 10.3% 10.9% 13.3% 11.3% 11.1% 13.2% 11.3% 40 1.7% 2.0% 1.8% 2.1% 2.4% 1.6% 1.9% 60 .9% 1.2% .5% 1.0% 1.0% .5% .3% XC-R 4.4% 3.6% 2.3% 6.1% 4.7% 2.8% 2.3% XC-FR 3.2% 2.9% 3.7% 3.7% 4.2% 6.9% 9.7% XC-FH .7% 1.0% 1.6% .8% 1.7% 2.7% 6.5% XC-OT 1.0% 1.3% 1.0% 1.0% .9% .6% 1.3%

  20. 16.0% RET 14.0% 12.0% 10.0% FR 8.0% FH WD 6.0% WD RET 4.0% 2.0% 0.0% CIC1* CIC2* CIC3* CCI1* CCI2* CCI3* CCI4* 40 60 XC-R XC-FR XC-FH XC-OT XC-EL XC-RET XC-WD

  21. Results and TD TD’s with more than 200 starters # starters 80.0 70.0 60.0 50.0 40.0 30.0 20.0 10.0 0.0 clear 1 20 2 40 3 60 4 XC-R 5 XC-FR 6 XC-FH 7 XC-OT 8 XC-EL 9 XC-RET 10 XC-WD 11 XC-AH-DR 12 SJ-EL 19 SJ-RET 20 SJ-WD 21 SJ-WD 100

  22. Falls and TD TD’s with more than 200 starters # starters 8.0 118 of 210 CCI4* 7.0 6.0 5.0 4.0 3.0 2.0 1.0 0.0 XC-FR 6 XC-FH 7

  23. Results and CD C D’s with more than 300 starters # starters 80.0 70.0 60.0 50.0 40.0 30.0 20.0 10.0 0.0 clear 20 40 60 XC-R XC-FR XC-FH XC-OT XC-EL XC-RET XC-WD XC-AH-DR SJ-EL SJ-RET SJ-WD rest

  24. Falls and CD C D’s with more than 300 starters # starters 9.0 8.0 7.0 6.0 5.0 4.0 3.0 2.0 1.0 0.0 XC-FR XC-FH

  25. Dressage/Show Jumping score Result code, Starlevel Result code N Mean Std. Deviation Dressage penalties clear 12098.0 54.3 7.5 57.5 20 2166.0 55.7 7.4 57.0 40 375.0 56.3 7.7 56.5 60 179.0 56.9 7.6 56.0 XC-R 817.0 57.1 9.0 55.5 XC-FR 699.0 55.5 9.6 55.0 XC-FH 225.0 55.4 9.6 54.5 XC-OT 204.0 57.0 8.4 54.0 XC-RET 839.0 55.5 8.4 53.5 XC-WD 53.0 751.0 56.2 10.0 Total 52.5 18355.0 54.9 7.9 Result code N Mean Std. Deviation Show Jumpiing fence penalties clear 12098 5.1 5.7 8.0 20 2166 5.7 5.6 7.0 40 375 6.7 6.3 6.0 60 179 6.4 6.3 5.0 XC-R 817 1.9 4.3 4.0 XC-FR 699 1.8 4.1 3.0 XC-FH 225 2.4 4.9 2.0 XC-OT 1.0 204 2.2 4.7 XC-RET 0.0 839 1.8 3.9 XC-WD 751 5.3 7.7 Total 18355 4.7 5.7 ANOVA: the ResultCode has an influence

  26. Rider profile  Result code 2 types of riders - 1 and 2* riders - 3 and 4* riders No clear clustering

  27. Rider profile  Horse Falls X-axis: 3 and 4* riders Y-axis: 1 and 2* riders

  28. Horse profile Result code Horse profile 2 types of horses - 3 and 4* - older 1 and 2* No clear clustering

  29. Horse profile Horse Faslls Horse profile X-axis: 3 and 4* riders Y-axis: 1 and 2* riders

  30. 2013: summary To do 2014 ResultCode PC1rider PC2rider PC3rider PC1horse PC2horse PC3horse XC-clear Mean SE XC-20pen Mean -0.133 No experienced 1 and 2* horses SE 0.025 Mean XC-40pen SE XC-60pen Mean No 2 and 3* rider , Not many horses -0.318 -0.298 No experienced 1 and 2* horses SE 0.060 0.084 Mean XC-EL-R -0.351 -0.256 No 2 and 3* rider , Not many horses SE No experienced 1 and 2* horses 0.023 0.035 XC-EL-FR Mean -0.227 -0.193 No 1, 2 and 3* rider , Not many horses SE 0.030 0.032 Mean XC-EL-FH -0.188 0.220 No 1* rider SE Horse start in 2 and 3* 0.061 0.083 XC-RET Mean SE XC-WD Mean 0.180 0.176 2and 3* rider with one horse SE Horse start in 2 and 3* 0.039 0.055 Total Mean 0.001 0.000 -0.001 -0.001 0.001 0.001 SE 0.008 0.008 0.008 0.009 0.009 0.009

Download Presentation
Download Policy: The content available on the website is offered to you 'AS IS' for your personal information and use only. It cannot be commercialized, licensed, or distributed on other websites without prior consent from the author. To download a presentation, simply click this link. If you encounter any difficulties during the download process, it's possible that the publisher has removed the file from their server.

Recommend


More recommend