Scribd
Upload
31 views

Read Me

The document includes a zip package with 31 CSV tables containing play-by-play data from NFL games since 2000, as well as SQL and documentation files. The CSV files can be loaded into Excel or a MySQL database for analysis. Weekly updates with new game data will be available each season for subscribers. The data is estimated to be over 99.8% accurate, with about 1,000 plays potentially needing correction, but should meet the needs of most users. The data can be used freely without restrictions.

Uploaded by

clanshe
Copyright
© Attribution Non-Commercial (BY-NC)
Available Formats
Download as PDF, TXT or read online on Scribd
31 views

Read Me

The document includes a zip package with 31 CSV tables containing play-by-play data from NFL games since 2000, as well as SQL and documentation files. The CSV files can be loaded into Excel or a MySQL database for analysis. Weekly updates with new game data will be available each season for subscribers. The data is estimated to be over 99.8% accurate, with about 1,000 plays potentially needing correction, but should meet the needs of most users. The data can be used freely without restrictions.

Uploaded by

clanshe
Copyright
© Attribution Non-Commercial (BY-NC)
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

May 5th, 2012

Whats included in this zip package 1) 2) 3) 4) 31 CSV Tables Setup.sql Data Reference.pdf These notes.

Quick Start Excel: Load the CSV file named Flat-File into Excel. To speed operation times, it is best to re-save it as a binary (.xlsb) file. Details on all 518,035 plays since the 2000 season are contained in this file. Player codes (i.e., XX-000) are referenced in Roster.csv and different totals (both by team and player) can be found in the Games, Team, Offense, Defense and Kickers.csv files. All of the remaining CSV files will be of little use if a relational database is not being used (i.e., MySQL). PHP / MySQL (using phpMyAdmin): At the end of the day, the flat-file is a highly inefficient way of storing NFL play-data. Any serious application development or research will require the use of a MySQL relational database. To that end, a Setup.sql file is included which will create 30 different tables for you. All you need to do is create a new database in phpMyAdmin (the name of it doesnt matter) and then copy and paste the Setup.sql file into the MySQL code editor. Click Go and a few moments later the database will be ready. In order to fill each table with data, you will need to select them individually in phpMyAdmin and then click on the Import tab. Choose the appropriate CSV file to import and make sure to enter 1 for the Number of Rows to Skip (the first one contains the column names). Also, ensure that the Allow Interruption of an Import option is selected so that some of the larger tables can be uploaded. Other options exist for importing the data in a more automated fashion; however, you will typically need to be running a dedicated server and will also have to write a few custom PHP/MySQL scripts to handle the process. Weekly Updates (Subscription Only) Each week of the season, shortly after the completion of the Monday night game, a new .zip package containing CSV files for the entire 2012 season will be available for download (for subscribers). These can then be imported into your existing database.

Data Accuracy The Core data table currently contains 518,034 plays and without a doubt, some of these have not been parsed correctly particularly the more complex plays that feature more than 1 turnover or a lateral or two thrown in for good measure. These kind of plays are rare, luckily, and I currently estimate that the dataset has around a 99.8% accuracy rate covering all variables. This means that there are still about 1,000 plays that could use some sort of correction whether it be with the players that were marked as being involved or the yardage that was gained/lost, etc. If we were to look at this from the perspective of a single week of the season, this adds up to roughly 5 plays out of 2,500 where an error or difference might exist between the final official league record and my database. So, if youre goal is to provide a source of NFL data that matches NFL.com yard-for-yard with exacting historical accuracy, then, this data may not be for you. For the vast majority of interested parties the aforementioned level of accuracy should easily meet your needs, especially with regards to pro handicappers , researchers and businesses engaged in player analysis from a Fantasy perspective. Restrictions on Usage This one is easy: there are none! Go forth and write some awesome research papers or build the next kick-ass Fantasy Football app. I dont own this data. Thats why its free in the off-season. The in-season updates do carry a cost ($35 per-week) to reflect the fact that I basically have no life for 20+ weeks every year its not a bad deal, really! If there is one thing that I do ask, its that your site/application/research paper contains a reference to ArmchairAnalysis.com as the source of the data. Enjoy!

You might also like