Zurdo זורדו

🚪🚪🐐 Lessons in Decision Making from the Monty Hall Problem | Towards Data Science A journey into three intuitions: Common, Bayesian and Causal

Eyal Kazin dives into why the Monty Hall Problem "doesn't matter" is the wrong answer and how to leverage analytical thinking in the face of uncertainty.

This was a fun pieced to write.
Thanks for sharing @towardsdatascience.com !

🚪🚪🐐 Lessons in Decision Making from the Monty Hall Problem | Towards Data Science A journey into three intuitions: Common, Bayesian and Causal

The Monty Hall Problem... a classic brain teaser that reveals surprising flaws in our intuition about probability and decision making. Eyal Kazin dives into why "it doesn't matter" is the wrong answer and how to leverage analytical thinking in the face of uncertainty.

September 12th - Data Liberation Day! 🇪🇺

Sounds like a lot of pros. What are the cons?

Start a New Year of Learning on the Right Foot A special edition of must-read articles and resources to help you kick off a productive 2025

🎉 New Year, New Learning Goals! 🚀 Kick off 2025 with a fresh batch of accessible resources for aspiring data scientists and machine learning professionals.

🖊️ by @zur-do.bsky.social, Barr Moses, Leonie Monigatti, Jason Tamara Widjaja, Sarah Lea, Vyacheslav Efimov, Sara Nóbrega + more

#slack bug!

They converted "under a minute ago" to "55 years ago" … just after the moon landing.

"[From the Simpson’s Paradox we learn that] certain decisions cannot be made based on the basis of data alone, but instead depend on the story behind the data. … Graph Theory enables these stories to be conveyed" - Judea Pearl

Internalising this paradox is a great gateway to causality.

#Causalsky

"The equals sign does not convey any causal information" - Judea Pearl

Structural Equations make causality explicit.
Textbook physics doesn't!
E.g., dv/dt := F/m reads F causes velocity change.

Einstein's E=mc²? Depends:
E := mc² → Manhattan Project
m := E/c² → Particle accelerators

#Causalsky

"Mistaking a mediator for a confounder is one of the most deadliest sins of causal inference..." - Judea Pearl

Chains and Forks share dependencies but demand distinct treatments. Forks require adjustment to prevent bias; chains forbid it to preserve causality. DAGs clearly visualise. #CausalSky

"[The Monty Hall Problem] shows that our beliefs should depend not merely on the facts observed but also on the process that led to those facts." - Judea Pearl (Causal Primer, p.15)

Daily quote 4 #MontyHall #CausalSky

Switch or stay? Judea points that the host’s choice reveals more than it seems.

This conference slammed the door on baseless claims that causality cannot be used in industrial practice. A brand new episode from cAI London 2024. In the new episode of Causal Bandits Extra we interview participants and speakers at... 🧵 1/ #CausalSky

This conference slammed the door on baseless claims that causality cannot be used in industrial practice.

A brand new episode from cAI London 2024.

In the new episode of Causal Bandits Extra we interview participants and speakers at...

🧵 1/

#CausalSky

Wordle 1,266 4/6*

🟩⬛⬛⬛🟩
🟩⬛⬛⬛🟩
🟩⬛🟩⬛🟩
🟩🟩🟩🟩🟩

Today's #wordle

Green person with very long arms drowning.

"Our unique gift is seeing what might have been." — Judea Pearl

Quote no. 4

The Causal Ladder defines 3 rungs:

1️⃣ Seeing (what happened?)
2️⃣ Doing (what happens if we act?)
3️⃣ Imagining (what would happen?).

From observations to interventions to counterfactuals—causality evolves with us.
#Causalsky

"Our unique gift is seeing what might have been." — Judea Pearl

Daily quote no. 4

The Causal Ladder defines three rungs:

1️⃣ Seeing (what happened?)
2️⃣ Doing (what happens if we act?)
3️⃣ Imagining (what would happen?).

From observations to interventions to counterfactuals—causality evolves with us.

I just dug into my notes and I jotted down:
"(e.g, Monty Hall Let’s Make a Deal vs. Let’s Fake a Deal page 192 Book of Why, Surveys)"

I'll check when I get home! (Regarding page 192, I have the soft cover book)

➡️ Start Asking Your Data “Why?” - A Gentle Intro To Causality Begin your causal journey with visual demonstrations and resources.

I like to paraphrase "The story behind the data is as important as the data itself"

To learn more see bit.ly/start-ask-wh...

"The collection of information is as important as the information itself" - Judea Pearl

Daily quote no. 3.

Pearl reminds us that context shapes meaning: biases & assumptions lurk in how info is gathered. It's a subtle critique of Deming's 'Trust data' mantra.

#CausalSky

🧠🧹 Causality - Mental Hygiene for Data Science Harness The Power of Why with Causal Tools.

For more details see my article about causality as mental hygiene

bit.ly/causal-menta...

"Oddly, statisticians both over- and underrate the importance of confounders." – Judea Pearl (The Book of Why, p. 139)

Day 2 of of daily quotes. Pearl critiques stats for lacking tools like DAGs to discern what to control for—essential to avoid spurious correlations like colliders!

#CausalSky

➡️ Start Asking Your Data “Why?” - A Gentle Intro To Causality Begin your causal journey with visual demonstrations and resources.

Interested in asking your data Why?

This beginner’s guide to thinking beyond correlations was written as the article I needed when starting to learn causality a few years ago.

bit.ly/start-ask-wh...

#CausalSky

🧠🧹 Causality - Mental Hygiene for Data Science The Power of Why: Enhancing Data Science with Causal Tools.

Causality is great for Data Science mental hygiene. 🧠🧹

I'm looking forward to presenting this week in PyData Global.

Can't make it?

In parallel to slide preparation I posted a @towardsdatascience.com article, which dives a bit deeper.

bit.ly/causal-menta...

#CausalSky @pydata.bsky.social

Thanks for sharing my article, @alxndrmlk.bsky.social. I'm fairly new to Bluesky, feel free to contact me at m @zur-do.bsky.social.

🧠🧹 Causality - Mental Hygiene for Data Science The Power of ‘Why?’: Enhancing Data Science with Causal Tools.

In a thoughtful and comprehensive post, Eyal Kazin explores how the very act of assessing the applicability of causality "is valuable in its own right since it can improve the scientific rigour of your projects."

(9) Embrace ignorance and be humble 🤷‍♂

(8) There are many ways to skin a … problem 🐈

(7) A well designed visual goes a long way 🎨

(6) Simulations are powerful but not always necessary 🤖

(5) When confused — look for a useful analogy … but tread with caution ⚠️

(4) Be one with subjectivity 🧘

(3) With new information we should update our beliefs 🔁