@davidzipper I think people know this and consider it a safety feature. I pointed it out to a friend and they said "well other people should get bigger cars too."

Michael Chavinda @mchav@mastodon.social

· 25mo ago

Replying to

@swope @davidzipper agreed but sadly ethics can't regulate people's behavior at scale. Probably need better legislation.

Michael Chavinda @mchav@mastodon.social

· 17mo ago

Replying to

Tutorials in the repo: https://github.com/mchav/dataframe

github.com

Michael Chavinda @mchav@mastodon.social

· 6mo ago

I’ve been thinking a lot about how to scale symbolic regression: this seems like a plausible direction.

https://mchav.github.io/grow-and-mow/

mchav.github.io

Grow and mow: interpretable models with boosting, symbolic regression and e-graphs

This post is the convergence of two ideas that have been floating in my head for about a year. Can we learn messy stochastic models and use algorithmic/algebraic tools to rein in model complexity to make models interpretable?

Michael Chavinda @mchav@mastodon.social

· 11mo ago

Wrote a new article where I checkpoint the work we’ve done so far enabling Kaggle style EDA-to-model workflows in Haskell.

https://mchav.github.io/iris-classification-in-haskell/

mchav.github.io

Michael Chavinda @mchav@mastodon.social

· 12mo ago

Have some pretty cool examples of feature engineering using program synthesis on Haskell data frames.

Given a function space, we run a breadth first search to find what functions (and their compositions) have the highest correlation with a target variable.

https://github.com/mchav/dataframe/blob/feature_engineering/app/Main.hs#L32

github.com

Michael Chavinda @mchav@mastodon.social

· 13mo ago

Starting a series on programming synthesis

https://mchav.github.io/an-introduction-to-program-synthesis/

mchav.github.io