openWAR
openWAR copied to clipboard
beanumber
explore use of RGameDay backend
All of the MLBAM scraping code doesn't really belong in openWAR. It belongs in RGameDay.
The pitchRx-like database functionality is also a must have. I think this can be achieved with @cpsievert's help and the etl package.
I don't have much time to work on this till late June, but this seems to me a must-do now. I ran into some "R CMD CHECK" issues and decided to remove openWAR from one of my toolsets. Today I decided to put it back in and now I'm getting install failures in devtools::install_github("cboettig/Sxslt")!
I'm going to fork Sxslt and see if I can figure out what's happening, but at the moment everything I'm doing is based on cpsievert/pitchRx, which is working and which gives me a nice SQLite database. Note that the CRAN pitchRx has a couple of minor bugs that I fixed; the fixes are in cpsievert/pitchRx.
Meanwhile, my code is in znmeb/masteringdfsanalytics
Fork done - removed a few lines of code from R/nodeSet.S and my fork of Sxslt now installs ... see https://github.com/znmeb/Sxslt/issues/1 and https://github.com/znmeb/Sxslt/commit/94af3909317facd7b862c0938b6e117ae67d708e
I'm going to file a pull request since the cboettig/Sxslt repo doesn't have issues enabled.
Did you try omegahat/Sxslt? @duncantl updated this a few days ago and then it compiled for me.
I wasn't aware of its existence. But I am now and it's working. Thanks!
But switching to pitchRx / SQLite still sounds like a good idea, if only to get things working on Windows. My code's trivial: https://github.com/znmeb/masteringdfsanalytics/blob/master/R/mlb_database.R.
I've been trying to use openWAR on windows lately and after giving up on the sxslt dependencies, I'm really interested in the suggestions here of using pitchRx to just put it all in a database. Are there any examples of what the data frames that openWAR needs should look like so I have a target to shoot for? I've started combing through the code trying to see what it's doing, but seeing the end would be faster.
@lqsullivan @beanumber I've got some free time now to look at this. Meanwhile, for Windows users with a decent amount of RAM there's always Docker - there are Docker images with RStudio, and I've got one with Jupyter (Notebooks, currently Python 3, R and Julia.