Matt Harris

RandomForest is a cool algo but be careful of the default tree depth hyper-parameter; it want's to overfit your data Tree depth is how "deep" the decision tree goes, e.g. how many decisions are made before coming to an answer. If you use RF, but don't know about max tree depth...

@SEPTA_SOCIAL Thank Enjoil! The Septa websites said that the Sports Express trains were no longer and to expect delays. Even if there are no system issues, is the BSL running a regular schedule? Thanks!

@Md_Harris Greetings. At this time, there are no reports of the BSL experiencing any issue. ^ER

Goood Saturday morning! Andy and Enjoli here by your side for any SEPTA-related questions. We'll be here until 5 PM.🌇🚉🍵 ^AS

@zachlanebryan Tickets for the Philly show were super expensive. The. I saw a video of you in a Phillies hat and I bought those tickets immediately! Go Phils!!!

yes, glad I’m not the only one waiting for dinger bohms

@CecilyTynan what is this cloud formation?

@CecilyTynan Yes!!!! Flourtown. Was very evident.

Anyone just feel like we just had an #earthquake ? 10:24 AM

Look here von Neumann, the particle does not give a rats ass about your consciousness.

@vboykis Sorry, the DistributedSquid deployment of ParaaS is only available in the Azure-Denmark-South region. However, the UniSquid Parameter Pond is available bare metal.

“Oh wow! I hear you can deploy the parameter service on DistributedSquid now”

Being an MLE is just talking to other MLEs and being like, “did you hear about ModelReservoir? I hear they added compatibility with LambdaFeatureLake in their 2.3.46 release. Would love to know what you think about their parameter as a service offering” until you die

Executing queries on spatial data can be difficult due to the complex polygonal shape of geographical boundaries. That's why geospatial systems resort to a strategy called “Filter-Refine” to evaluate geospatial predicates efficiently. Details in🧵 #geospatial #DataAnalytics #SQL

@bnummerdor Genius! Lets give that a whirl…

@Md_Harris Ask ChatGPT to settle marital arguments, that should end well

2023-02-24 6:50pm: Our first marital argument over a ChatGPT prompt. The future is bleak.

Every Philadelphian deserves access to safe & reliable public transportation. RT if you agree.

@ArynnPost 🎉 🙌

@MoSarwat The Anime (top right) versions are the most fun for me. But top left is a pretty authentic and interesting version. I think it’s a fun and engaging use of deep learning 🤷

A little late to the game, but I kinda have mixed feelings about AI-generated portraits / Avatars. Which one you think is best?

@machsci So true. Love that quote.

"The most exciting phrase to hear in science, the one that heralds new discoveries, is not 'Eureka!' but 'That's funny...'" Isaac Asimov

@auren I’m living on the edge at 2:45 in the afternoon.

What is your most irrational fear? Mine: that my phone will suddenly run out of power (so am always charging)

@beeonaposy Just started it!

About 1/4 of the way through and really digging this book so far

@machsci twitter.com/md_harris/stat…

Just because a model can classify something with 99% accuracy doesn't mean it's useful.

@stewartbryson @SnowflakeDB Congrats! ❄️

I’m honored to included in the #DataSuperheros program. Thank you @SnowflakeDB.

"With hard work and perseverance, anything is possible despite your initial circumstances. After 100 rejection letters, I was recruited by Snowflake in 2018 and have been enjoying every minute since!” — Marion B. Consulting Manager at Snowflake #BlackHistoryMonth

Finally learned some basic raster analysis, and took the USGS topographic elevation + National Land Cover datasets to create a least-cost-distance path from Harrisburg to Pittsburgh. (Corridors in background). Results are interesting for high speed rail implications!