8000 GitHub - amirdhami/F1-Racing-Prediction: Predicting the outcome of F1 races by classifying drivers into a given final placement based on their stats going into a race
[go: up one dir, main page]
More Web Proxy on the site http://driver.im/
Skip to content
/ F1-Racing-Prediction Public

Predicting the outcome of F1 races by classifying drivers into a given final placement based on their stats going into a race

License

MIT license
5 stars 0 forks Branches Tags Activity
Star
Notifications You must be signed in to change notification settings

amirdhami/F1-Racing-Prediction

BranchesTags

Repository files navigation

F1 Racing Prediction

CURRENT CONTENTS OF THE REPO:

raw_data

  • contains the 3 raw data files
  • these were all directly pulled from the kaggle site

F1_data.csv

  • contains the consolidated data
  • starting point for my transforms

F1_qualtime.csv

  • contains the consolidated data
  • "Qual Time" column now contains avg time of qual phases that the driver competed in if the race was new qual format (3-round quals started in 2006)
  • "Qual Time" may contain "DNS" or "DNF" if a driver failed to complete phase 1 of the qual for any reason

F1_racetime.csv

  • most recent and processed version of data--please use this as a baseline for further modification
  • contains prev data from F1_qualtime.csv
  • additional columns are "Time in Seconds" and "Adjusted Time"
  • "Time in Seconds" is an intermediate for calculations
  • "Adjusted Time" is an adjusted form of "Time/Retired" in seconds for all drivers
  • "Adjusted Time" contains either a time in seconds (for drivers who did not complete all laps in time, the avg time for the 1st place driver to complete a lap is added per lap they were behind) or "DNF" if the driver did not finish
  • "Adjusted Time" has some holes due to disqualifications--these are not yet accounted for

qualtime.py

  • transformed "F1_data.csv" into "F1_qualtime.csv"
  • note that manually filled in holes where qual was unfinished (DNF, DNS, etc.) will be erased if you rerun this script

finaltime.py

  • transformed "F1_qualtime.csv" into "F1_racetime.csv"

construct_raw.py

  • Data Loading: Reads multiple F1 datasets (results, constructors, races, circuits, drivers, status, qualifying, driver standings, pit stops) into DataFrames.
  • Data Merging: Combines these datasets to include detailed race, driver, constructor, and circuit information.
  • Column Renaming: Clarifies data by renaming columns (e.g., date to raceDate, position to finishPos).
  • Feature Engineering: Calculates historical performance metrics (e.g., points and wins before races, recent placements, seasonal averages, circuit-specific averages, driver age at race, position changes).
  • Pit Stop Data: Integrates pit stop data, calculating average number and time of pit stops.
  • Final Adjustments: Filters data for races from 2011 onwards and saves the final DataFrame to raw_data_with_pit_stops.csv.

fnn.py

  • Feed forward Neural Network based on the raw data file
  • Uses pytorch for most of the heavy lifting
  • MSE loss function

fnn_cross_entropy.py

  • Cross entopy loss function

decision_tree.py

  • contains both the random forest and id3 classifier

About

Predicting the outcome of F1 races by classifying drivers into a given final placement based on their stats going into a race

Resources

Readme

License

MIT license
Activity

Stars

5 stars

Watchers

1 watching

Forks

0 forks
Report repository

Releases

No releases published

Packages

No packages published

Contributors 2

  •  
  •  

Languages
0