Corrie Bartelheimer

Data Scientist. Bayesianist. Data Visualist. Mathematician. She/Her. @[email protected]

[Corrie Bartelheimer @corrieaar]

Slides for my talk about Bayesian Workflow with PyMC3 and ArviZ https://t.co/HFYJBJeqgn with accompanying code and notebooks on Github https://t.co/lejg2v1aSy

[Corrie Bartelheimer @corrieaar]

@Natascha_Strobl Ich hätte auch noch gerne einen 🙈

[Corrie Bartelheimer @corrieaar]

@RVAwonk Over on mastodon, @CT_Bergstrom did a long Thread how this paper is badly done and doesn’t support the conclusions they’re making

https://t.co/EimCfhYHJt

[Alyssa Kann @AlyssaKann]

After months of declining follower counts, some state-affiliated media outlets on Twitter started to gain more followers, all around the same date. Here's the sudden shift of an account from Russia's RT, beginning a period of follower growth March 28 and onwards:

[Corrie Bartelheimer @corrieaar]

First time I’m getting a notification to subscribe to someone 🙄
Looks like Twitter became someone’s personal ego trip

[Corrie Bartelheimer @corrieaar]

And it's merged 🤩🤓🐼

[Corrie Bartelheimer @corrieaar]

Just submitted my first PR to @pandas_dev 🙌🏻
Thanks to @PyConDE @jorisvdbossche and the rest for organising 🐍#PyData #PyConDe

[Corrie Bartelheimer @corrieaar]

@wagnerpeer Oh thank you 😊 that is too kind 😌

[Corrie Bartelheimer @corrieaar]

Very nice talk about giving and receiving feedback for better PRs by @JDKAndersson #pydata #pyconde @PyConDE

[Corrie Bartelheimer @corrieaar]

Just submitted my first PR to @pandas_dev 🙌🏻
Thanks to @PyConDE @jorisvdbossche and the rest for organising 🐍#PyData #PyConDe

[Corrie Bartelheimer @corrieaar]

@neorutje @jorisvdbossche @TomAugspurger I come from R dyplrs so I’m a big fan of method chaining, makes sense that that doesn’t work with inplace

[Corrie Bartelheimer @corrieaar]

All in favour of this proposal: the pandas 🐼 inplace stuff always seemed dirty to me 😅 @jorisvdbossche #pyconde23

[Corrie Bartelheimer @corrieaar]

And some hard truths for data scientists: we need to deliver value. We need to be mindful of business value, efficiency and communication. Too often DS teams are just very costly without much ROI

[Corrie Bartelheimer @corrieaar]

The “zoo of titles”. Are you actually a data scientist or rather a data translator? Or ML engineer? Or an analyst? Who knows….
Great point by Noa Tamir that this also has adverse effects on job satisfaction & career development 🤔
#PyData #PyConDe @PyConDE 🐍

[Corrie Bartelheimer @corrieaar]

@wagnerpeer Oh and I was hoping to have a chat with you over a beer or coffee 🙃 but happy to hear you’re around online

[Corrie Bartelheimer @corrieaar]

It’s starting! 🤩 my first PyData con since 2019! 🐍 I’m soo pumped and looking forward to talk and learn all things Python and data the next three days 🤓#pydata #pyconde23

[Corrie Bartelheimer @corrieaar]

@TMrtSmith @vboykis Our team started using airflow some months ago and I still secretly hope we switch to metaflow 🫠 at the same time, migrating is really the last thing I wanna do… too much pain. I just wanna complain about a different *flow

[vicki @vboykis]

Tell me your title, I'll tell you your job:
DS - Feeding daily metrics into Airflow
MLE - Feeding batch model updates into Airflow
MLOPs - Airflow
AI Researcher - Figuring out how to get to Airflow

[PyConDE & PyData @PyConDE]

Does your production code look like it’s been copied from Untitled12.ipynb? Are your eng… @corrieaar at #PyConDE #PyDataBerlin
https://t.co/1XGpF82Vyp

[Corrie Bartelheimer @corrieaar]

Already happening so soon! 🤩

[PyConDE & PyData @PyConDE]

Does your production code look like it’s been copied from Untitled12.ipynb? Are your eng… @corrieaar at #PyConDE #PyDataBerlin
https://t.co/1XGpF82Vyp

[Corrie Bartelheimer @corrieaar]

@HamelHusain I remember a few occasions where our problem was the other way round: some SW engineer had helped out with our DS/ML code but didn’t know pandas and reimplemented a bunch of inbuilt pandas functions with inefficient for loops 🫠

[Corrie Bartelheimer @corrieaar]

@Glennwithtwons @robmarkcole @HamelHusain Second this. Even when most of the time I’m just doing SWE, every once in a while I’m faced with a problem that requires a good intuition & deeper knowledge of the data and model at hand