Exploring the 2005-2017 MotoGP Racing Dataset

As a lifelong Motorcycle Grand Prix (MotoGP) enthusiast, stumbling across Ankur Vishwakarma’s Predicting MotoGP Race Finish Times using Linear Regression was a fantastic read, and I immediately realized that MotoGP data is now much more available than it once was.

As part of an Undergraduate Research Project in 2008, I had briefly tried to use similar datasets to explore the effects of rider attrition and optimal expected performance timeframes, but unlike baseball the data was very difficult to obtain. Fast forward now to official repositories and web scraping resources I could have only dreamt about in 2008.

Vishwakarma’s .csv dataset was scraped from the MotoGP Official Website. I have since forked his research into my Github account. The following is his Data Dictionary:

Summary of Data

This file contains the following data points for racing sessions only between 2005 and 2017 in all 3 racing classes.

• Year: int value of racing season.
• TRK: three character str abbreviation of track.
• Track: full name of the race in str format.
• Category: class of racing, whether it’s MotoGP, Moto2, 125cc, etc. as a str.
• Session: str if the race was a RAC (race) or RAC2 (race2) session of the day.
• Date: pandas readable datetime (date only) of the race.
• Track_Condition: str format racetrack conditions, whether dry, wet, or dry-wet.
• Track_Temp: float value of track ground temperature in celsius (C).
• Air_Temp: float value of air temperature in celsius (C).
• Humidity: float value of local humidity between 0 (0%) and 1 (100%).
• Position: finish order of the particular rider. Denotes ‘crash’ if rider crashed before finishing.
• Points: float value of points earned by rider during this race session.
• Rider_Number: rider’s racing number.
• Nationality: rider’s nationality as a str.
• Team_Name: team that the rider belongs to during this race session.
• Bike: manufacturer of the motorcycle ridden by rider during this race session.
• Avg_Speed: average speed of the rider during this race session in km/h.
• Time: str value of the finishing time for the winning rider. Riders finishing 2nd or worse have the added time value. Riders who crash have the number of full completed laps indicated.
• Finish_Time: pandas timedelta value obtained from the Time column. This is each rider’s total time to finish the race. Crashed riders are ignored.
• GP: concatenated track abbreviation and name used to pull in track data.
• track_length_km: total track length in kilometers (km).
• l_corners: number of left corners.
• r_corners: number of right corners.
• width_m: average width of racetrack in meters (m).
• straight_m: length of track’s longest straight in meters (m).
• GP_avg_speed: average speed of MotoGP motorcycles around the track in kilometers per hour (km/h) if available.
• gp_dist: distance raced by MotoGP class bikes in kilometers (km).
• m2_dist: distance raced by Moto2 class bikes in kilometers (km).
• m3_dist: distance raced by Moto3 class bikes in kilometers (km).

Here is the .csv data table:

library(pander)

URL <- "https://raw.githubusercontent.com/nbugliar/motogp_regression/master/MotoGP_2005_2017.csv"
dat <- read.csv( URL, stringsAsFactors=F )

m <- dat[1:5, 2:22]
pandoc.table(m, style= "rmarkdown")

## 
## 
## | Year | TRK |                       Track                        | Category |
## |:----:|:---:|:--------------------------------------------------:|:--------:|
## | 2017 | QAT | Grand Prix of Qatar - Losail International Circuit |  MotoGP  |
## | 2017 | QAT | Grand Prix of Qatar - Losail International Circuit |  MotoGP  |
## | 2017 | QAT | Grand Prix of Qatar - Losail International Circuit |  MotoGP  |
## | 2017 | QAT | Grand Prix of Qatar - Losail International Circuit |  MotoGP  |
## | 2017 | QAT | Grand Prix of Qatar - Losail International Circuit |  MotoGP  |
## 
## Table: Table continues below
## 
##  
## 
## | Session |    Date    | Track_Condition | Track_Temp | Air_Temp |
## |:-------:|:----------:|:---------------:|:----------:|:--------:|
## |   RAC   | 2017-03-26 |       Dry       |     22     |    21    |
## |   RAC   | 2017-03-26 |       Dry       |     22     |    21    |
## |   RAC   | 2017-03-26 |       Dry       |     22     |    21    |
## |   RAC   | 2017-03-26 |       Dry       |     22     |    21    |
## |   RAC   | 2017-03-26 |       Dry       |     22     |    21    |
## 
## Table: Table continues below
## 
##  
## 
## | Humidity | Position | Points | Rider_Number |    Rider_Name     |
## |:--------:|:--------:|:------:|:------------:|:-----------------:|
## |   0.96   |    1     |   25   |      25      | Maverick VIÑALES |
## |   0.96   |    2     |   20   |      4       | Andrea DOVIZIOSO  |
## |   0.96   |    3     |   16   |      46      |  Valentino ROSSI  |
## |   0.96   |    4     |   13   |      93      |   Marc MARQUEZ    |
## |   0.96   |    5     |   11   |      26      |   Dani PEDROSA    |
## 
## Table: Table continues below
## 
##  
## 
## | Nationality |       Team_Name        |  Bike  | Avg_Speed |   Time    |
## |:-----------:|:----------------------:|:------:|:---------:|:---------:|
## |     SPA     | Movistar Yamaha MotoGP | Yamaha |   165.5   | 38'59.999 |
## |     ITA     |      Ducati Team       | Ducati |   165.5   |  +0.461   |
## |     ITA     | Movistar Yamaha MotoGP | Yamaha |   165.4   |  +1.928   |
## |     SPA     |   Repsol Honda Team    | Honda  |    165    |  +6.745   |
## |     SPA     |   Repsol Honda Team    | Honda  |    165    |  +7.128   |
## 
## Table: Table continues below
## 
##  
## 
## |        Finish_Time        |                 GP                 |
## |:-------------------------:|:----------------------------------:|
## | 0 days 00:38:59.999000000 | QAT - Losail International Circuit |
## | 0 days 00:39:00.460000000 | QAT - Losail International Circuit |
## | 0 days 00:39:01.927000000 | QAT - Losail International Circuit |
## | 0 days 00:39:06.744000000 | QAT - Losail International Circuit |
## | 0 days 00:39:07.127000000 | QAT - Losail International Circuit |

There are many different avenues one could pursue with this information. I hope you find it as valuable as I have, and welcome any and all feedback and recommendations as to best use this dataset in the future.

Upon an initial inspection, I think that the following should take place:

• Cleaning A lot of capitalization and the potential for inconsistencies make this worth taking a closer look at.
• Visualization Overlaying the victories, losses, championships, and durations of competitiveness would be cool.
• Dashboard Lastly, an interactive dashboard that the casual racefan and die-hard data scientist alike could explore.

Thoughts? my email