Jensen Holm

I am beyond excited to announce that I will soon be joining the St. Louis Cardinals as a Machine Learning Engineer! After graduating from GVSU College of Computing next spring I'll be moving to St. Louis. This has been a dream of mine for a long time, and I owe this to the countless number of people that helped me get to this point. I can't wait to get started!

@choice_fielder You should use plotnine

I rarely use R anymore and man do I miss ggolot2. This update is sick.

This is awesome. A former intern just accepted a new job in the Analytics Department with the St. Louis Cardinals 👏 Jensen said that this is his dream job and his experience in Kalamazoo have him the needed experience with Trackman and other software. It made his resume stand out in front of 700+ other applicants from around the nation. Experience plus lifelong friendships and great memories led him to a job in Major League Baseball. Help us congratulate our 2022 Statistics Intern on landing his dream job 🙌 Interested applicants can find our internships at bit.ly/GrowlersIntern…

@TJStats @SpinDoctor_BB It’s going to be changed to daily updates during the season, and weekly refreshes for updates like this

@_holmj_ @SpinDoctor_BB So weekly, it re-downloads all games from Baseball Savant, not just new games?

Just updated the statcast_pitches package to include new arm angle data and all other new columns! Download as much or as little baseball savant pitch data from 2015-present in 3-45 seconds. github.com/Jensen-holm/st…

@TJStats @SpinDoctor_BB Yeah so I manually re scraped it all and looked for updates to existing data during the end of this season, but I’m currently testing a new workflow that automates this.

@TJStats @SpinDoctor_BB It’s all automated with github actions The hugging face dataset is automatically updated one a week and looks for any new data or changes to the data. You’ll never have to reinstall the package or make any code changes since the package just points to the hugging face dataset.

@_holmj_ @SpinDoctor_BB Do you need to re-scrape and re-upload when MLB makes changes to the data (i.e. adds a pitch type to a pitcher)

@gabe_appelbaum @saberpowers I might make an R package for it as well sometime in the near future ...

@_holmj_ @saberpowers Jensen, I'm so intrigued that I may just have to dust off vscode

just scraped an entire MLB season in under two minutes thanks for the package and the code @saberpowers

@SpinDoctor_BB Thanks, that is exactly why I made this

@_holmj_ this is really really useful (and hopefully will save me the 20 minutes required to download data from pybaseball)

2024 top pitch combinations by Sequence+ 🔥⚾️ 1. Sonny Gray sweeper / four seam: 106 2. George Kirby splitter / four seam: 105 Here is a clip of the best of each of these sequences. Sonny clip Sequence+: 115 Kirby clip Sequence+: 117 #baseball #pitching #analytics @PitchingNinja @SonnyGray2 @gkirb98

@SpinDoctor_BB That’s pretty cool, yeah I am also curious about prior pitches and not just back to back ones. Curious as to what you find out

@_holmj_ see if pitches prior to previous pitch have some impact, albeit a deafened impact. anyways very good work

if you have any tunnelling/pitch synergy articles/tweets/anything that you really like, let me know so I can reference it in my tunnelling article

Sequence+ progress update: Changes in feature selection have led to better separation between Sequence+, Location+ and Pitching+ while maintaining moderate/strong relationships with K/BB, FIP & WHIP. This is a good sign because it means that to an extent, I am isolating the effect of pitch tunneling. github.com/Jensen-holm/Se…

@gabe_appelbaum @TJStats It doesn’t have arm angle data yet, but working on updating it github.com/Jensen-holm/st…

@TJStats Unreal! I'm both incredibly excited and also not looking forward to re-scraping all my data

The Statcast Team has done it again! You can now get pitch-by-pitch arm angle from the Savant CSVs PyBaseball and/or BaseballR will likely need an update to download this data

I've been working on modeling cumulative expected run value for two pitch sequences using a more advanced tunnel score, and other sequence related metrics. Still a WIP but pretty pleased so far with how it relates to other metrics. github.com/Jensen-holm/Se…

@zo6logan @scikit_learn yeah, MacBook Pro M3

@_holmj_ @scikit_learn what’s this running on? macbook pro i assume?

njobs=-1 go brrrrrr @scikit_learn

I added the ability to access the statcast-era-pitches Hugging Face dataset through a simple python library I created called statcast_pitches. Below is an example on how you can get started either using the default Polars method (recommended) or query the data into a polars LazyFrame from a DuckDB SQL query. Each code snippet has the same result. Recently the popular open source baseball computer vision project, BaseballCV (created by @DrummeyDylan), adopted the statcast_pitches library and saw a significant speed up in their data download times. BaseballCV: lnkd.in/gtew6Dm6 statcast_pitches Source code: lnkd.in/g8Ad_Xsi HuggingFace Dataset: lnkd.in/gJTgWGaD