MLB Ranking Documentation¶
Welcome to the MLB Ranking Documentation!
Having played baseball for such a long time this project is very interesting to me on many levels. Much of the baseball specific analysis will be based upon is The Hidden Game by John Thorn and Pete Palmer. However, a significant portion of the project will be based on principles of data science not specific to baseball including data mining (with scheduling), association rules, recommender systems utilizing Jaccard Similarity.
A application to predict which MLB teams will be contenders in the range of 3-5 years. Also suggest what a particular team can do to make their team a contender in 3-5 years. Focus which stage a team is in (buying, selling, rebuilding, etc.) and how aggressive they are in that mode. Determine value of players focused on WAR and years of control as primary factors.
This will also be an exploration of ES2015 / ES6, Map and Reduce, and draw many ideas and algorithms from data science.
Links to some of the important sections.
- Algorithms - Algorithms for similarity comparisons to be used
- Player Model - Shows which stats will be focused on and why
- Simulator Application - High level synopsis of how the simulator will work
Assumptions and Constants¶
Work in progress
Decide on source for WAR - Baseball Reference, FanGraphs, ESPN, etc
- ES6 - NoSQL database system which stores data similar to JSON documents
- Mongo DB - standardized single modern database management system
- Nginx - high performance HTTP server
- Digital Ocean - simple cloud infrastructure for hosting
- Front End Framework - TBD
- Statistical Analysis Framework - TBD
- Read the Docs - Incrementation hosting with Sphinx generator