christophmolnar.bsky.social
Author of Interpretable Machine Learning and other books
Newsletter: https://mindfulmodeler.substack.com/
Website: https://christophmolnar.com/
131 posts
5,688 followers
966 following
Regular Contributor
Active Commenter
comment in response to
post
5/ It was stressful, but I don’t regret it. I learned a lot and definitely feel validated in my skills again.
Full story & solution details: https://buff.ly/4gHZYHD
comment in response to
post
4/ Writing Supervised ML for Science at the same time was a huge plus—competition & book writing fed into each other (e.g., uncertainty quantification).
comment in response to
post
3/ One key insight: SHAP’s reference data matters! I used historical forecasts for interpretability. Also combined SHAP with ceteris paribus profiles for sensitivity analysis.
comment in response to
post
2/ My approach:
✅ XGBoost ensemble, quantile loss
✅ SHAP for explainability + custom waterfall plots + ceteris paribus plots
✅ Conformal prediction to fix interval coverage
✅ Auto-generated reports with Quarto
comment in response to
post
1/ Years ago, I went full-time into writing & cut ML practice. At some point, I felt like an impostor writing about ML but no longer practicing. This competition about water supply forecasting on DrivenData (500k prize pool) was a way back in.
comment in response to
post
deprecated was maybe the wrong word. It's no longer the default in the shap package. There are faster alternatives
comment in response to
post
The connection between SHAP and LIME is only when we represent features differently for LIME and use a different weight function.
My take is that, while interesting, it can be misleading as SHAP and original LIME are very different, as you also say.
comment in response to
post
Not planned so far
comment in response to
post
Okay, that makes a lot of sense. Thanks!
comment in response to
post
I love that visualization. At the same time confused about why souping works.
comment in response to
post
While the results on ARC are impressive, the ARC problems are abstract games in little grid worlds. And it remains to be seen whether this translates into "general" intelligence.
comment in response to
post
Great way to frame it
comment in response to
post
Yeah, I'm mixing two very different scenarios here.
The socioeconomic ones are even tougher, so predicting the first place is extremely speculative.
comment in response to
post
As a consequence, there is disappointment.
comment in response to
post
Because I have multiple books already in Quarto, and it makes it easier to maintain them when they share the same software. Also I find Quarto easier to handle
comment in response to
post
Wait till he checks the google trend for XAI.
comment in response to
post
Alternatively, you only read the "not all you need papers"
comment in response to
post
True for almost eradicated diseases. For other diseases, the upside is also personal: For example, in my area ticks can give you Tick-borne encephalitis, but since I have the vaccine, I feel much saver. But maybe for some this upside is a bit abstract, since you might never get the disease anyways.
comment in response to
post
Actually sitting here with a bit of knee pain while writing this 😂
comment in response to
post
Why an entire weekend? A couple of TikTok videos should do 🤡
comment in response to
post
pandas.DataFrame.explosion.dodged_like_a_hero
comment in response to
post
pd.concat
comment in response to
post
pandas.DataFrame.set_axis
comment in response to
post
pandas.DataFrame.transform
comment in response to
post
pandas.DataFrame.round
comment in response to
post
pandas.DataFrame.drop
comment in response to
post
pandas.DataFrame.stack
comment in response to
post
pandas.DataFrame.max
comment in response to
post
pandas.DataFrame.min
comment in response to
post
pandas.DataFrame.ffill
comment in response to
post
pandas.DataFrame.shift
comment in response to
post
pandas.DataFrame.groupby
comment in response to
post
I love this. As well as the profiles.
comment in response to
post
The iPad is 13 inches which is not a good form factor for me for reading and the battery is already weak (it's from 2017)