Emily Riederer

I feel like #lazyweb isn't what it used to be, but anyone in #econsky happen to know of a good package for implementing fuzzy regression discontinuity designs in python?

My hand holding a shortbread cookie in the shape of an airplane. There are red sprinkles in the pattern of the survivorship bias plane.

A plate of the same cookies.

Does anyone want a survivorship bias shortbread

Five years from now I’m starting a consulting firm to teach people how to use wildly novel tools called pens and an idea notebook to think through a problem

Wondering what we lose as tech writers write for AI training vs humans. The lovely `polars` docs often use nice reprexs to demonstrate syntax but then note if the technique shouldn't be used in such a trivial case. Would someone write this now or worry AI picks example w/out context?

Great taxonomy for managers, as well! So important to be clear when feedback is a must-have, nice-to-have, or curiosity question you don’t reasonably expect to get answered

Mostly a review of pretty standard methods, but some fun examples of enabling expression expansion (the magic behind column selectors!), more complex objects in a dataframe (models, vectors), and breaking the paradigm to go back to partitioned dataframes + list comprehensions

(2/2)

Welcome! | Emily Riederer

Continuing my #python Rgonomics posts for the #rstats crowd, wrote a short post about the different ways to use (non-polars) user-defined functions in a pipe

Not the most groundbreaking thing, but one of those "the thing I needed to read a bit ago" posts

(1/)

www.emilyriederer.com/post/py-rgo-...

Maybe this is an “when I was a kid” thing, but book stores (ok definitely a “when I was a kid” thing) always used to have those “page a day” tear off calendars

Every year around this time I’m surprised/disappointed to find there aren’t good places online to print custom 365 day calendars

Do you people like.. not maintain lists of 400+ pithy thoughts on the same topic?

(Seriously, does anyone know where this can be done?)

1. This is reprehensible and should not exist. What are we doing?
2. Mildly amusing picturing AI me, learned from my digitaltraces; family and friends reaching out in their moment of need, and I'm just spouting off about data quality checks and column naming conventions

Counting down to day 2, coming at me starting 3:30A Chicago time! 🤓☕

(Also, massive respect to the folks in the chat who mentioned joining from Seattle time 😴)

7/7

Non-overlap Average Treatment Effect Bounds The average treatment effect (ATE), the mean difference in potential outcomes under treatment and control, is a canonical causal effect. Overlap, which says that all subjects have non-zero probability...

Not enough overlap to estimate ATE with potential outcomes? Elegant approaches for finding bounds when you can't point estimate arxiv.org/abs/2509.20206

Reminds me of some talks from last year aiming to rank when they couldn't estimate. Lots of analytically useful outcomes beyond estimates

(6/n)

Survey of models in APSR papers, using DAGs to analyze main model described and identify incorrect controlling for colliders, mediators, etc. propping up "significant" results mikedenly.com/research/dags

(5/n)

Exploring the heterogeneous impacts of Indonesia's conditional cash transfer scheme (PKH) on maternal health care utilisation using instrumental causal forests This paper uses instrumental causal forests, a novel machine learning method, to explore the treatment effect heterogeneity of Indonesia's conditional cash transfer scheme on maternal health care util...

Speaking of robust simulations, great talk on quantifying the impact of different estimators and policy learning approaches through simulation with applications in public health (team behind this paper but can't find the actual paper arxiv.org/abs/2501.12803)

(4/n)

Calibration Strategies for Robust Causal Estimation: Theoretical and Empirical Insights on Propensity Score-Based Estimators The partitioning of data for estimation and calibration critically impacts the performance of propensity score based estimators like inverse probability weighting (IPW) and double/debiased machine lea...

No more thoughtless logistic regressions! Very intuitively, correctly calibrating propensity scores can improve bias *and* variance of casual estimates arxiv.org/abs/2503.17290

(3/n)

Position: Causal Machine Learning Requires Rigorous Synthetic Experiments for Broader Adoption Causal machine learning has the potential to revolutionize decision-making by combining the predictive power of machine learning algorithms with the theory of causal inference. However, these methods ...

Survey of causal method evaluation and areas of opportunity -- focused on more credible benchmarks and simulations www.arxiv.org/abs/2508.08883

(2/n)

Favorite talks from day 1 of @causalscience.org 's #CDSM2025 🧵

The way you could ask a random question and the most qualified person in the world would just stop by to answer

Or how someone would ask the one thing you asked yourself last week and had the perfect link

Truly one of the best data science events that I look forward to all year long! No sales, no hype, just exquisite thoughtful yet practical methods. Taking two days off work to attend. Don't miss out!

A twinge of jealousy for the ecologist-types who are getting one heck of a natural experiment out of the FAA flight reductions

Terrible if you’re trying to live in a functional society, but pretty lovely if you do causal analysis on pollution or emissions

PEP 798 – Unpacking in Comprehensions | peps.python.org This PEP proposes extending list, set, and dictionary comprehensions, as well as generator expressions, to allow unpacking notation (* and **) at the start of the expression, providing a concise way o...

Python SC accepted PEP 798

PEP: peps.python.org/pep-0798/

Acceptance: discuss.python.org/t/pep-798-un...

So this:
[*row for row in list_of_lists]

Will do the same thing as this:
[x for row in list_of_lists for x in row]

⏰ Only 1 week left to register for #CDSM2025!

Join us Nov 12–13 for two days of talks & debates at the intersection of causality, data science & AI.

💻 Online | 🎟️ Free
👉 causalscience.org

My theory is that for practitioners, regression models should be like pocket money: you get a fixed number per week to do whatever you like with until we're sure you won't blow the whole lot on silly stuff, get caught up in a get-rich-quick scheme, or accidentally leave them in a drawer somewhere.

The way python and R foster inclusion directly contributes to their success: joyful places to exist, a steady flow of new maintainers, and a delightful collection of niche tools empowered by wildly different expertise coming together

Watch the new python documentary for more on PSF’s work here

The Majority Of Your Users The majority of your users don’t read your changelog. The majority of your users only upgrade to new versions when forced to.

jacobtomlinson.dev/posts/2025/t...

Highly relatable for anyone that has ever written a line of code used by other people

Lovely little post from @jacobtomlinson.dev

Chartle - A daily chart game Guess the country in red by analysing today's chart

Do you enjoy playing chartle.cc?
We’re opening up to the community!

Are you a scientist, journalist, researcher or just a data-enthusiast and would like to submit your own data (it should be a country-based time series)? Get in touch with us - we’d love to feature it!

I have no idea whether or not this is a good paper, but where the abstract cut off is the ultimate cliff hanger

It occurs to me that the CFPB epitomizes what Abundance calls for: Legitimacy through results, not process. They got a lot done, and that is precisely why conservatives put a target on their back. Yet, Klein and the other abundance bros never talk about them.

That said, not sure what the counterfactual 2025 rstats Twitter is. IMHO, part of nostalgia is for 2019 - prepandemic, ZIRP job market, rapid innovation in less mature data tools, pre AI, etc

A philosopher once said “You can’t step in the same feed twice. Neither you nor the feed are the same” 🙃

Discord/Slack also hard bcs it buckets content into different streams. Starting out, I didn’t know what I was interested in, but you’d swim in a pleasantly incoherent mix of specialties. Peak serendipitous intellectual diversity that it’s hard to intentionally seek (don’t know what you don’t know)