# Good clues from Twitter A secret subtitle for this publication is _"Can you learn econometrics from Twitter and Stack Overflow alone without distracting yourself to data science tutorials"_. Links collcted in no particular order, some will show in other sections of the Navigator. ## No links between leading macrotextbooks
## OLS interactively exposedThat Fabio Canova's 2007 textbook didn't even cite Jorda (2005) shows the weakness of citations as a measure of quality of work.
— Lance Bachmεiεr (@LanceBachmeier) 1 октября 2018 г.
## R language guide is an econometrics guideVery cool teaching tool here to help introduce students to OLS regression: https://t.co/eQV4rAQt7O
— Nicholas Hillman (@n_hillman) 10 сентября 2018 г.
## OLS, MML, Bayes and MCMC(!) for linear regression https://peterroelants.github.io/posts/linear-regression-four-ways/ ## Very true on R2You might take a look at our Guide to R, which we use, along with tutorial scripts, to teach intro econometrics. https://t.co/G5zPoL9Iyj 1/2
— Michael Kevane (@mkevane) 14 сентября 2018 г.
## Instrumental variablesIf a psychologist sees an R^2 of .62 they go back to find the error they made https://t.co/Nf33adiHKR
— Alexander Etz (@AlxEtz) 3 ноября 2018 г.
## Interpreting coefficients 1I see way to many people confused about this: You can't test the validity of your instruments. Don't even try! https://t.co/AjPvy6dOkL #CausalInference #DataScience #Econometrics #Stats #MachineLearning
— Paul Hünermund (@PHuenermund) 30 октября 2018 г.
## Interpreting coefficients 2Struggling to recall how to interpret logistic regression outputs?https://t.co/jfJxgwh4gW pic.twitter.com/L0CodvWGHG
— Displayr (@Displayrr) 10 ноября 2018 г.
## Doing PCA approachI highly recommend this article to anyone who uses regression. Very helpful explanation of how to interpret partial regression coefficients, why the "answer" changes when you add more variables to your model, etc. https://t.co/xdxl0x3eGL
— Maggie R Wagner (@maggieRwagner) 22 июля 2018 г.
## OLS explained for social scientistsOK twitter, what's your opinions on the PCA-axis-as-new-variable approach? ie
— Andr(é|ew) MacDonald (@polesasunder) 13 марта 2018 г.
1. Collect many correlated variables
2. do PCA
3. interpret the first few axes as being about something specific
4. use those axes in a multiple regression#rstats #ecology
## Suggested stat excercisesReverse Engineering a Regression Table: A simple (?) introduction for social scientists https://t.co/mPIdcuNUuf
— AK (@AchKem) 17 января 2018 г.
## Coding a treeProve the Lindeberg-Levy Central limit theorem.
— mac strelioff (@macstrelioff) 12 ноября 2018 г.
Prove the Gauss-Markov theorem.
Show the parameters in a regression model are independent of the residuals.
Find the Score and Fisher Information for every common distribution. @SachaRoseUritis did I miss anything?
## Consumer demand modellingNice explanation, and breakdown of the algorithm. I agree - coding it is the best way to determine if you REALLY understand the math! https://t.co/GEcbxFhLwV
— Dave Giles (@DEAGiles) 10 ноября 2018 г.
## Scott Cameron learning methodKey developments in modelling the microeconomics of consumer demand, from @jdube6 https://t.co/HOoTO8nXKq #Econometrics #Microeconomics #ConsumerDemand pic.twitter.com/EnvXr3UdRD
— NBER (@nberpubs) 7 ноября 2018 г.
## Model evaluation compendium (on classifier)Most I ever learned was from “pick a paper and recreate its results”
— Scott Cameron (@twitscotty) 14 сентября 2018 г.
And every intro course should make students do one OLS regression in matrix form in a spreadsheet. Invest six hours to live the rest of your life with a fundamental understanding of what your software is doing https://t.co/WGsUO0MCQu
## Causality by Judea PearlBtw. I forgot to mention that I had also made a single PDF of all 4 model evaluation articles, which is maybe easier to read (and/or if you prefer print versions) https://t.co/BDuQ1RBjD0 https://t.co/7GQ5cIMwJ5
— Sebastian Raschka (@rasbt) 15 ноября 2018 г.
## A thesis turned tutorial on probabilistic programming and MC inference by Tom RainforthReaders asked for a conceptual summary of my book Causality (2009), free of the mathematics that decorates most pages. I'VE FOUND ONE: https://t.co/BhxvqcgTqx -- a lecture given to AI audience in 1999, and its still fun to read. #Bookofwhy #Causalinference #ecobookclub
— Judea Pearl (@yudapearl) 8 сентября 2018 г.
## Value of logitI have written a tutorial (masquerading as my thesis) on probabilistic programming and Monte Carlo inference. It starts with the absolute basics and takes you all the way through to some cutting edge developments. Check in out here https://t.co/ISNoyvN3Zh pic.twitter.com/CjkmUGgSSI
— Tom Rainforth (@tom_rainforth) 13 апреля 2018 г.
## Traditional statistics vs MLI'll use deep learning even if I can solve the problem using a single logistic regression just to use my GPU
— Abhik Sarkar (@abhiksark) 18 ноября 2018 г.
## Program evaluation by John HolbeinTraditional statistical methods often out-perform machine learning methods for time-series forecastshttps://t.co/4ENmeJ4FnW #AI #statistics #bigdata
— Charles-Abner DADI (@DadiCharles) 14 июля 2018 г.
## Which estimator are you?New @AnnualReviews (economics): "Econometric Methods for Program Evaluation"
— John B. Holbein (@JohnHolbein1) 7 августа 2018 г.
This piece provides an overview of common methods for program evaluation and discusses recent developments with these methods.
Very useful for teaching.#SocSciResearchhttps://t.co/FfMZYP0Vyo pic.twitter.com/pcATE81Qzu
## 218 pages on probabilistic programmingWhich estimator are you? I like being BLUE hence OLS!#econtwitter #Econometrics #Statistics #DataScience #estimators #BLUE #OLS #AcademicTwitter #Academia #ResearchHighlight #research pic.twitter.com/ENGWDcchzD
— Alessia Paccagnini (@Alessia_metrics) 17 сентября 2018 г.
## Reviewing Lucas critiqueBookmark: This 218-page Intro to Probabilistic Programming. 🤓https://t.co/QVybd8us4A
— Kaggle (@kaggle) 18 октября 2018 г.
Many thanks to 🙏 @hyang144, Jan-Willem van de Meent, Brooks Paige, Frank Wood #ebook #programming pic.twitter.com/82gZ1jGFio
## MCMC recommendation by Neil Shepard1/ My new working paper @BristolEcFin : What is the link between microfoundations and the Lucas Critique(LC) ? A historical appraisal. Thread.
— Francesco Sergi (@cescoeco) 23 октября 2018 г.
## A4 econometric artI am very excited by this line of work by Pierre Jacob and his coworkers. Their coupling of MCMC chains to produce unbiasedness is simple, beautiful and important. https://t.co/p01wimrDwA
— Neil Shephard (@shephard_neil) 15 сентября 2018 г.
## CLT mattersLegalicenla: permito a mis alumnos venir a mi examen con una hoja escrita de ambos lado, con cualquier cosa. Cada año doy el premio "Arte Econometrico" a la mayor cantidad de info por cm2. Aqui uno de los ganadores. pic.twitter.com/i1cTpSpxLp
— Walter Sosa Escudero (@wsosaescudero) 28 ноября 2018 г.
## A 1910 must-readThe Central Limit Theorem matters in real life. https://t.co/GxFPx2PsMV
— Alex Combessie (@alex_combessie) 3 декабря 2018 г.
## Amazing statistics animationHow a book written in 1910 could teach you calculus better than several books of today [Calculus Made Easy, 1910] https://t.co/W4gspJoGQc pic.twitter.com/rm45Ot0xSm
— Massimo (@Rainmaker1973) 2 декабря 2018 г.
## Undergrad econometrics condensed to 3 pagesAs requested, slower graphs! Also added a graph on collider bias, the webpage explanation helps there.
— Nick HK (@nickchk) 29 ноября 2018 г.
These graphs are intended to show what standard causal inference methods actually *do* to data, and how they work.
This is what controlling for a binary variable looks like: pic.twitter.com/dTZxqY5JxA
## Microeconometrics refresher course wantedAlso, in case it's useful, here is an entire overview of undergrad econometrics (no matrix algebra) in 3 pages: https://t.co/kQYrn92v7w pic.twitter.com/UBe7OZJTJh
— Tyler Ransom (@tyleransom) January 15, 2019
## ML tricksChristmas wish: an econ online course that teaches new developments of estimation in empirical micro, in an applied way, for people who have already completed their PhD but want to keep up with latest advances. E.g. one lecture a month, with exercises. #EconTwitter
— Dina D. Pomeranz (@DinaPomeranz) December 26, 2019
## ML methods in perspectiveBecause there is nothing more useful than a mathematical trick, and thanks to your feedback, find the updated ML trick slide -:) Please RT, add even more to the list !!! #MachineLearningTrick pic.twitter.com/S0EL3Jh7le
— Frank Nielsen (@FrnkNlsn) January 9, 2020
## Trouble with R2 (again)Why not look at econometrics/ML methods in historic perspective? Source: https://t.co/hcvCFzWWNE
— Evgeny Pogrebnyak (@PogrebnyakE) July 30, 2019
Attn: @EconometricsN pic.twitter.com/20gp1xjFYg
## Writing 101The trouble with R² by Cosma Shalizi (famous stat prof)
— Ted Petrou (@TedPetrou) December 16, 2019
1. Doesnt indicate model is good
2. Can be low for correct model
3. Can be high for incorrect model
4. Doesn't indicate prediction error (MSE does that)
Read several more critiques here:https://t.co/RTvMVY9spu
## Writing 102Person: how does writing work?
— Meredith Ireland (@MeredithIreland) January 9, 2020
Writer: well you type & you delete. You rethink. Then you do 187 min of research & correct it. You reread & wonder if you have a grasp of English. Then you revise
Person: then you’re done with the book?
Writer: then you move to the next sentence
## A what of economistsWriting advice: don’t write like other people. Dont follow stylistic convention. Don’t read the writing guide. https://t.co/VRoy4BWx5p
— Harry Crane (@HarryDCrane) January 11, 2020
## Shady sitesFILL IN THE BLANK: A pod of whales. A murder of crows. A __________ of economists.
— Ben Casselman (@bencasselman) January 6, 2020
## THE master plana gentle reminder that people who don't have access to a university library should definitely not resort to using shady sites like sci-hub where you can get free but probably illegal access to many journal articles
— Jasmine Artemis (@JazzArtemis) December 27, 2019
### Brownian motionsI’ve found the master plan of every academic!#AcademicChatter #phdchat pic.twitter.com/n4sHXqZzsH
— Stephen Aguilar (@stephenaguilar) December 22, 2019
Also here: https://github.com/IlievskiV/Amusive-Blogging-N-Coding/blob/master/Random%20Processes/geometric_brownian_motion.ipynbMy blog post on Geometric Brownian Motion, which is the foundation in modeling some stock indices. It includes full simulation in #Python and cool animations with #Matplotlib. Check it out! #pythonprogramming #maths #blogging #visualization #animations https://t.co/gj9KALPIFT
— Vladimir Ilievski (@VladOsaurus) May 23, 2020