Piero Ferrante

AVP & Data Science Fellow @CVSHealth | Advising, investing, teaching, and always learning #ml #python #data #infra | Retweets are not my own

[Khuyen Tran @KhuyenTran16]

Have you ever wanted to add a math description for your #Python function but found it time-consuming to do so? latexify_py allows you to add LaTex math description with only one decorator.

Link to latexify_py: https://t.co/2v8eygXgBj

[Piero Ferrante @the_real_pdf]

@willmcgugan More dynamic Cookiecutter and/or parameterized DB ERDs would be straightforward and very useful

[Piero Ferrante @the_real_pdf]

@josh_wills What about an API call or a utility function from a package so we can meet users with an aversion for *any* traversals where they typically work (read: inline)? 🚀

[Andrew Carr 🤸 @andrew_n_carr]

I had no clue that just 6 lines of code would improve my research productivity / reproducibility so dramatically

[GeoPandas @geopandas]

Long time in the making, but we are happy to officially announce:

✨ Dask-GeoPandas for scalable spatial analysis in Python! ✨

Blogpost: https://t.co/mGcqscjlxX
Docs: https://t.co/GJ3FXyhN8b

Small thread ⬇️

[Mark Tenenholtz @marktenenholtz]

The WRONG way to significantly improve your models:

• Different architecture
• Hyperparameter tuning
• Bigger ensemble

The RIGHT way:

• Thorough EDA
• Meticulous error analysis
• Robust model evaluation

Spend your time on the 100%+ gains, not the 1% gains.

[Hector Yee @eigenhector]

So excited that xmanager is open sourced https://t.co/QCWkWK0Csb

[Piero Ferrante @the_real_pdf]

Strap in for a wild ride...

[Lightning AI ⚡️ @LightningAI]

🎉 Write your own training loops for any type of research from meta-learning to recommendation systems with the new Lightning Loop API in v1.5!

⚡️ Dive into the Lightning Loop API: https://t.co/LJfcbdJabJ

[François Chollet @fchollet]

The highest-leverage way to add value is to *help others create value*, at scale. This is why tech is magical.

[Chip Huyen @chipro]

New blog post! Discussing:

1. Whether a data scientist should be fullstack
2. What caused the unreasonable expectation that DS should know Kubernetes
3. An overview of the tools that can help abstract away infra to allow DS to own a project end-to-end 🚀

https://t.co/4I3fDPaHe6

[Chip Huyen @chipro]

I was never a good systems engineer, so I always avoided the topic of compiling and optimizing ML models.

However, as I work with ML on devices, the topic keeps coming up. So I spent the last 3 months learning about ML compilers.

Here’s what I learned. https://t.co/VNXbUcfJ2h

[Robert Yi 🐳 @imrobertyi]

Imagine if you were responsible for every analysis/dashboard/model/pipeline you made for the rest of your life

[Piero Ferrante @the_real_pdf]

@mekkaokereke Sincerely, thank you for this

[Mekka 💉x7 @[email protected] @mekkaokerekebye]

I love how some folk are like "Principal engineers would never sully themselves with fixing a small bug! Their time is reserved only for meeting heads of state and other distinguished people!"

And Principals are like, "I am a janitor LOL! Can I clear your lunch tray? I gotchu!"

[Sean J. Taylor @seanjtaylor]

Kats is another promising time series package from Facebook's Infrastructure Data Science team, covering detection, forecasting, feature extraction/embedding, multivariate analysis: https://t.co/C7OOUWlqcG

[Piero Ferrante @the_real_pdf]

@databozo @SNHU Still remember your simulation workshop from PyData Seattle in '15... You're a great speaker/teacher and your prospective students are lucky!

[Simon Willison @simonw]

PostgreSQL 14 adds new syntax for accessing JSON data:

SELECT *
FROM shirts
WHERE details['attributes']['color'] = '"neon yellow"'

I like this so much more than the -> operators, which stubbornly refuse to stick in my head

[Sean J. Taylor @seanjtaylor]

Being extremely skeptical about complicated solutions is a superpower.