Sign in

Eric Leung

@erictleung.bsky.social
890 followers 488 following 280 posts

marketing data scientist, generalist, math and library enthusiast, data scientist of the third kind, loves good stationary and pens, low tech enthusiast, open source tinkerer, opinions = mine #rstats

PostsRepliesMedia
Eric Leung @erictleung.bsky.social · 13/08/2026
i've created a datasette web app on Vercel so anyone can immediately go explore my Pixar films dataset using SQL pixarfilms-datasette.vercel.app
pixarfilms-datasette.vercel.app
Datasette: pixarfilms
130
Eric Leung @erictleung.bsky.social · 24/06/2026
TIL about marino pair, cool tool to collaborate with agents in interactive notebook-like sessions marimo.io/blog/marimo-...
marimo.io
Introducing marimo pair
A canvas where agents and developers collaborate on research and data work; structured working memory and a reactive Python runtime for agents.
110
Eric Leung @erictleung.bsky.social · 24/06/2026
just learned about ediff in emacs. M-x ediff-files n — Navigate forward to the next difference p — Navigate backward to the previous difference a — Overwrite the highlighted difference in File B with the text from File A b — Overwrite the highlighted difference in File A with the text from File B
100
Eric Leung @erictleung.bsky.social · 17/06/2026
TIL about Google's AI edge gallery, a smartphone app (Android and iOS) to try out LLMs locally. could be useful on planes and spotty cell service www.xda-developers.com/ran-a-full-l...
xda-developers.com
I ran a full LLM on my phone with no internet, and it's more useful than I expected
I owe the local LLM community an apology.
110
Eric Leung @erictleung.bsky.social · 17/06/2026
been ssh-ing into servers a lot, TIL about the -vv flag to get more details about the handshake for troubleshooting if errors arise ssh -vv user@hostname
000
Eric Leung @erictleung.bsky.social · 16/06/2026
SQL = thinking in shapes Python = thinking n steps
100
Eric Leung @erictleung.bsky.social · 11/06/2026
I sometimes wonder about how vague world of advancing your own skills and careers. ran into this more developer-centric rubric, but might be a helpful roadmap for growth and learning github.com/Semalab/deve...
github.com
100
Eric Leung @erictleung.bsky.social · 10/06/2026
reminds me of task views that are listed on CRAN, giving a curated list of R packages to help with different tasks or types of analysis cran.r-project.org/web/views/
cran.r-project.org
CRAN Task Views
CRAN task views aim to provide guidance which packages on CRAN are relevant for tasks related to a certain topic. They give a brief overview of the included packages which can also be automatically in...
010
Eric Leung @erictleung.bsky.social · 10/06/2026
TIL that the @librarycongress.bsky.social has these cool research guides guides.loc.gov
guides.loc.gov
Research Guides
Index of Library of Congress Research Guides Research guides to the Library's collections, as well as subject guides prepared by Library of Congress staff, are listed below. More online guides coverin...
100
Eric Leung @erictleung.bsky.social · 03/06/2026
more reminders, I had forgotten about some of the details around multicollinearity, besides it not being a good thing in a model statisticsbyjim.com/regression/m...
statisticsbyjim.com
Multicollinearity in Regression Analysis: Problems, Detection, and Solutions
Multicollinearity is when independent variables in a regression model are correlated. I explore its problems, testing your model for it, and solutions.
100
Eric Leung @erictleung.bsky.social · 03/06/2026
also TIL about the phi coefficient (aka Matthews correlation coefficient) as an evaluation metric for machine learning and statistical models medium.com/@cdefaux/phi...
medium.com
Phi Coefficient A.K.A Matthews Correlation Coefficient (Binary Classification)
In machine learning/data science, we often run into problems where we’re trying to classify binary (two-class) data. In this case, you…
100
Eric Leung @erictleung.bsky.social · 03/06/2026
random reminder of how to do SQL rolling average queries SELECT product_id, sale_date, daily_sales, AVG(daily_sales) OVER ( PARTITION BY product_id ORDER BY sale_date ROWS BETWEEN 2 PRECEDING AND CURRENT ROW ) AS rolling_avg_3_days FROM sales_data;
110
Eric Leung @erictleung.bsky.social · 02/06/2026
I've been trying to learn more about machine learning engineering to teach about it and found this nice overview about it. makes sense to be some blend of software engineering, data science, and mathematics, plus some knowledge of your application domain. www.maxmynter.com/pages/blog/b...
maxmynter.com
Max Mynter - Full Stack Machine Learning Engineer
Max Mynter - Full Stack Machine Learning Engineer. Explore my blog and portfolio. Multiple years of experience in data science, software engineering, and machine learning.
110
Eric Leung @erictleung.bsky.social · 21/05/2026
I've been using SQL for 5+ years now and only now have been paying a bit more attention. I got some code from a colleague and saw they did a join not using "on". at first I was confused, but learned this is an older standard. the following thread shows the difference. the more you know!
100
Eric Leung @erictleung.bsky.social · 18/05/2026
"Good docs answer what a new analyst would ask."
110
Eric Leung @erictleung.bsky.social · 08/05/2026
I've been digging deeper into better SQL queries, and looks like Google has a good documentation page on it. it's split out in a few sections: - reduce processed data - optimize query operations - reduce query output - avoid anti-SQL patterns docs.cloud.google.com/bigquery/doc...
docs.cloud.google.com
Optimize query computation  |  BigQuery  |  Google Cloud Documentation
Best practices to optimize query performance.
110
Eric Leung @erictleung.bsky.social · 07/05/2026
🧵 some other cool built-in functions in R I found while I was already in there
100
Eric Leung @erictleung.bsky.social · 07/05/2026
i'm sure there are more optimized or vectorized versions, but it is cool to know there's a built-in edit distance function in R, utils::adist(): adist("kitten", "sitting") # [,1] # [1,] 3 stat.ethz.ch/R-manual/R-d...
stat.ethz.ch
R: Approximate String Distances
011
Eric Leung @erictleung.bsky.social · 06/05/2026
ah I was thinking of doing a word cloud and got reminded of a "better word cloud". couldn't remember but then eventually found them again, they're called "chatterplots", bringing a little order to an aimless word cloud medium.com/data-science...
a modified scatter plot where the points are the words themselves when analyzing text, the color representing sentiment, and size is frequency. moreover, the x-axis is sentiment and y-axis is word frequency
110
Eric Leung @erictleung.bsky.social · 08/04/2026
i've had to start using google bigquery, so this post was useful to get up to speed about it towardsdatascience.com/burn-data-ra...
towardsdatascience.com
A Definitive Guide to Using BigQuery Efficiently | Towards Data Science
Make the most out of your BigQuery usage, burn data rather than money to create real value with some practical techniques.
110
Eric Leung @erictleung.bsky.social · 05/04/2026
for R users wanting to get a list of the working titles of all R versions, you can run this SPARQL query from Wikidata query.wikidata.org#SELECT%20DIS...
query.wikidata.org
120
Eric Leung @erictleung.bsky.social · 02/04/2026
a great explanation and illustration on why bayesian statistics these days uses Hamiltonian Monte Carlo methods instead of Metropolis-Hastings algorithm that's behind Gibbs Sampling elevanth.org/blog/2017/11...
elevanth.org
Markov Chains: Why Walk When You Can Flow?
In 1989, Depeche Mode was popular, the first version of Microsoft Office was released, large demonstrations brought down the wall separating East and West Germany, and a group of statisticians in the ...
110
Eric Leung @erictleung.bsky.social · 01/04/2026
I hear a lot about "propensity score matching" en.wikipedia.org/wiki/Propens...
en.wikipedia.org
Propensity score matching - Wikipedia
100
Eric Leung @erictleung.bsky.social · 27/03/2026
i'm slowly moving my dotfiles from using stow to chezmoi, but also reuniting with my old friend tmux. i almost forgot how to scroll within a window. you can either do Ctrl-b [ (or whatever your prefix is), or set set -g mouse on # >=v2.1 set -g mode-mouse on # <v2.1 superuser.com/q/209437/941...
superuser.com
How do I scroll in tmux?
How do I scroll with either the keyboard or mouse? The tmux man page indicates one must enter copy-mode to scroll. Is there a way to quickly scroll without manually entering copy-mode?
100
Eric Leung @erictleung.bsky.social · 27/03/2026
also trying to learn more about causal inference methods. i'm familiar with the statistical rethinking materials on the subject, just need the practice to make it intuitive for myself
100
Eric Leung @erictleung.bsky.social · 27/03/2026
i'm trying to upskill in more experimentation techniques, and one method is to do geo lift analyses. this article here shows the GeoLift R package from Meta towardsdatascience.com/how-to-analy...
towardsdatascience.com
Analyzing Geo-based Campaigns with Synthetic Control | Towards Data Science
Using Meta's GeoLift Library
100
Eric Leung @erictleung.bsky.social · 27/03/2026
ah on Linux systems when upgrading R and you need to migrate all your old R packages over to the new version, you can run update.packages(checkBuilt = TRUE, ask = FALSE)
110
Eric Leung @erictleung.bsky.social · 10/03/2026
learning some API development with FastAPI has been pretty slick fastapi.tiangolo.com
fastapi.tiangolo.com
FastAPI
FastAPI framework, high performance, easy to learn, fast to code, ready for production
100
Eric Leung @erictleung.bsky.social · 05/03/2026
need a quick Python REPL, but you're away from your main computer? but you also need libraries like numpy and scikit-learn? the python.org homepage has got you covered
the official Python website with an interactive Python shell built into the front page, showing libraries like numpy and sci kit learn are installed
152
Eric Leung @erictleung.bsky.social · 05/03/2026
i may be behind the times regarding Python REPLs, but just found ptpython. it works super well! i'm impressed. the last innovation in Python REPLs i saw was years ago with ipython github.com/prompt-toolk...
github.com
GitHub - prompt-toolkit/ptpython: A better Python REPL
A better Python REPL. Contribute to prompt-toolkit/ptpython development by creating an account on GitHub.
100
Eric Leung @erictleung.bsky.social · 04/03/2026
need to reproducibly take a screenshot of a webpage using R? you can use the chromate package to do this, so cool rstudio.github.io/chromote/art...
rstudio.github.io
Taking a screenshot of a web page
000
Eric Leung @erictleung.bsky.social · 02/03/2026
preparing for some interviews, and here are some things i've looked up in prep
110
Eric Leung @erictleung.bsky.social · 01/03/2026
notes on studying recommendation systems (recsys)
000
Eric Leung @erictleung.bsky.social · 25/02/2026
i've been using the Linux terminal on my Chromebook, and frustratingly, I can't update packages to the latest. but then updated the /etc/apt/source.list with the line deb ftp.de.debian.org/debian forky main
ftp.de.debian.org
Index of /debian
110
Eric Leung @erictleung.bsky.social · 17/02/2026
some convenient calculators for A/B testing tools, like sample size calculation and even some basic survival curve analysis www.evanmiller.org/ab-testing/
evanmiller.org
Evan's Awesome A/B Tools - sample size calculator, A/B test results, and more
000
Eric Leung @erictleung.bsky.social · 13/02/2026
a fun stats read on trying to predict Taylor Swift's dress color used in the acoustic set. some cool stats, like a correspondence analysis I haven't seen in a while, along with some good visuals. for anyone needing some light hearted content academic.oup.com/jrssig/artic...
academic.oup.com
We were in screaming colour: Colours of the Taylor Swift universe: a statistical story
Abstract. Could statistics predict which dresses Taylor Swift would wear to perform the Eras Tour's surprise songs? Marina Ferrari de Aquino Klemm and Char
110
Eric Leung @erictleung.bsky.social · 01/02/2026
in case you need to change the email or name for git commits git rebase -r <some commit before all of your bad commits> \ --exec 'git commit --amend --no-edit --reset-author' stackoverflow.com/a/1320317/24...
stackoverflow.com
How do I change the author and committer name/email for multiple commits?
How do I change the author for a range of commits?
100
Eric Leung @erictleung.bsky.social · 29/01/2026
notes on studying PyTorch
100
Eric Leung @erictleung.bsky.social · 29/01/2026
notes on studying machine learning
110
Eric Leung @erictleung.bsky.social · 28/01/2026
"Every dataset has a perspective" zine by DAIR. a quick read, but a nice meditation on how we design and pick the data we record that may impact our interpretation later zines.dair-institute.org/datasets-per...
zines.dair-institute.org
The DAIR Zine Library
A collection of zines from the DAIR Institute
010
Eric Leung @erictleung.bsky.social · 28/01/2026
Bayes credible int: "given the observed data, the effect has 95% probability of falling within this range" vs Frequentist (95% CI): "there is a 95% probability that when computing a confidence interval from data of this sort, the effect falls within this range" easystats.github.io/bayestestR/a...
easystats.github.io
Credible Intervals (CI)
010
Eric Leung @erictleung.bsky.social · 23/01/2026
TIL John Tukey, whom i know of for his work on the box plot, also is credited for the fast-fourier transform, or rather for the most common fast-fourier transform en.wikipedia.org/wiki/John_Tu...
en.wikipedia.org
John Tukey - Wikipedia
110
Eric Leung @erictleung.bsky.social · 22/01/2026
love how there's a whole org, the European Spreadsheet Risk Interest Group, that's focused on spreadsheet productivity and risk management eusprig.org
eusprig.org
European Spreadsheet Risk Interest Group
110
Eric Leung @erictleung.bsky.social · 21/01/2026
documenting my data work like a lab notebook from college feels nostalgic, but also seeing the wisdom in documenting along the way
110
Eric Leung @erictleung.bsky.social · 20/01/2026
notes on using GitHub Copilot
100
Eric Leung @erictleung.bsky.social · 19/01/2026
cool package find, the {parsermd} #RStats package to programmatically interact with RMarkdown documents, maybe it could be used for Quarto documents too github.com/rundel/parse...
github.com
GitHub - rundel/parsermd
Contribute to rundel/parsermd development by creating an account on GitHub.
070
Reposted by Eric Leung
Andrew Barnas @andrewbarnas.bsky.social · 16/01/2026
well now what do I do
1235
Eric Leung @erictleung.bsky.social · 16/01/2026
build build build
010
Eric Leung @erictleung.bsky.social · 16/01/2026
notes on studying LLMs
000
Eric Leung @erictleung.bsky.social · 13/01/2026
still one of the best explanations of principal component analysis (pca), explained at different levels from layman to the more math inclined stats.stackexchange.com/a/140579/132...
stats.stackexchange.com
Making sense of principal component analysis, eigenvectors & eigenvalues
In today's pattern recognition class my professor talked about PCA, eigenvectors and eigenvalues. I understood the mathematics of it. If I'm asked to find eigenvalues etc. I'll do it correctly li...
16118