Skip to content

Latest commit

 

History

History
40 lines (31 loc) · 2.35 KB

DATAGUIDE.md

File metadata and controls

40 lines (31 loc) · 2.35 KB
Raw

Data Guide

Explanation of Collected Data

All data falls into five categories, and represents game information and stats from 1994 to 2020.

The five categories of data collected are as follows:

  1. Games: week, winner, loser, score, location, day of week, time
  2. Season Offenive Stats for Each Team: team, avg points, avg yards, net yards gained per pass attempt, and rushing yards gained per attempt
  3. Season Offensive Conversion Stats for Each Team: team, 3rd down conversion %, 4th down conversion %
  4. Season Defensive Stats for Each Team: team, avg points allowed, avg yards allowed, net yards allowed per pass attempt, and rushing yards aallowed per attempt
  5. Season Defensive Conversion Stats for Each Team: team, 3rd down conversion % allowed, 4th down conversion % allowed

From there the data was compiled into two types of files:

  1. Stats: includes every collected stat for each team in a given season (created using this script)
  2. Matchups: includes game information and stats for both teams for each game in a given season (created using this script)

The model will be trained with data from 1995 to 2018 for ther following reasons:

  1. From 1995 on there were at least 30 teams in the NFL for any given season.
  2. From 1995 on the average number of points scored per team per game never fell below 20 for any given season.

Collection of Data

All data was collected with a set of scrapers utilizing the selenium and pandas packages. The scrapers and any scripts used to manipulate the CSV files can be found here.

Possible Changes to Data

Including the following stats may create a more accurate model:

  • FG%: compares total field goals made vs total field goal attempts
  • INT%: percentage of times intercepted when attempting to pass
  • SK%: percentage of times sacked when attempting to pass
  • Pen: penalties committed by team and accepted
  • Record and strength of record

In addition, the following information, even if not easily quantifiable, may be useful in future attempts:

  • Injured or benched players for a game
  • Weather for a game
  • Whether a game is a division game or not
  • How signifigant a game is for a team's playoff chances

Source

All data has been scraped from Pro Football Reference .