clj-baseball

This is a Clojure library inspired by Python's pybaseball. It scrapes HTML and CSV data on baseball websites and makes it friendly to work with at the REPL. Baseball data is hugely plentiful and granular; I think it makes for an excellent match with a data-driven language like Clojure. My intention in building this library (besides having fun) is to offer a way for users to learn Clojure with abundant real world data. Even if the sport of baseball does not interest you, its ready-to-use data provides an excellent opportunity to explore the language ecosystem without hacking on your own web app or seeking out inferior APIs, second-rate "dummy data", or any other inadequate and frustrating option.

Code organization

The library is divided into a few primary namespaces. At press time, the most complete functionality is available in com/slothrop/statcast, where the user will find the functions that query MLB's Statcast dataset for real-time, pitch- and swing-level data. Query validation is provided by Clojure spec. Data is returned in CSV format and converted into a Clojure map.

In com/slothrop/sabermetrics, the user can find the existing functionality for creating their own maps of baseball stats (sabermetrics is the official term for baseball stats). These are a work in progress.

All other namespaces are stubs at this time.

Python interop!

This is an experimental feature, but for now basic DataFrames are available thanks to the excellent libpython-clj library. Two idiomatic Clojure wrappers around some DataFrame class methods (DataFrame.from_dict and DataFrame.from_records) are provided. To enable Python support, you must of course have Python installed on your machine, and use one of the jdk-?? aliases specified in deps.edn, based on your version of Java. This includes modules required only by libpython-clj when the JVM starts.

Portal

Pandas is a very powerful tool for tabular analysis, but its API can be somewhat difficult to manage. In the near future, I may investigate the panthera library for providing more idiomatic Clojure support for DFs. For now, however, I intend for users of this library to leverage Clojure's functional and data-driven approach to inspect the data provided by the Statcast API. One outstanding tool for this is Portal, which offers many different viewing options, including tables. I highly recommend it.