Tim Radtke

I posted the slides from my presentation at yesterday's #satRdayBerlin conference: minimizeregret.com/post/2019/06/1…

@psobot @normconf Thanks for the talk and the preparation that surely went into it, captivating storyline!

Up next at @normconf: my talk! On how to deploy machine learning products within budget by thinking like an engineer. Watch live here: youtube.com/watch?v=pR3QUe…

@BBayesians @twiecki @mexiamorelli Among my favorite reasons: Statistics and decisions become exercises in counting. And ABIYLFOYP, kids!

Great @BBayesians meetup tonight: Large turnout for a talk by @twiecki. Thanks to @mexiamorelli for the organization! Presentation showed how a model can be iteratively tuned to fit the business problem, recapped why Bayesian modeling can be a great fit for business use cases.

@seanjtaylor This presentation by Fernanda Viégas and Martin Wattenberg from 2011 features data visualizations with a focus on repetitions and motifs (music, text, Wikipedia usage). Just watched it, thought you might like it: youtu.be/kNEmil66bfo

Very excited the Berlin Bayesians meetup is happening again *in-person*! And the topic of Buy 'Til You Die models is one of my all-time favorites, looking forward to @juanitorduz's take on them👌

Kick-off with @ogrisel presenting a notebook introducing different strategies to leverage time-related features for a bike sharing demand regression task. bit.ly/3pPnqdN #timeseries #euroscipy @EuroSciPy 22, Basel 🇨🇭

“Calling if() or while() with a condition of length greater than one gives an error rather than a warning.” Can’t say that I’ve been reading the R changelog much in the past, but if it’s always that good I sure will going forward. This is great news! cran.r-project.org/doc/manuals/r-…

Ah yes, the wonders of modern machine learning explanations.

ML systems debugging rewards a willingness to dig across the entire stack, to chase a slightly suspicious signal back to its source, and to derive chains of failures from surprising end results. High cognitive burden but also some of the most exhilarating work upon success.

Books are the best.

@canyon289 @mihagazvoda @PyData That's the way to go for presentations! Lesson for me: It's only live once, next time then. But: This way I can keep pretending that you only said kind things about Minimize Regret. ;)

@mihagazvoda @TimRadtke @PyData To @mihagazvoda's credit I requested it not be recorded so I could say all sorts of extra fun stuff 😀

Join us on the virtual @PyData Ljubljana Meetup on Wednesday, 2022-04-13, at 6pm CEST!📅 @canyon289 will talk about how to pick the right data science problems for your organization and influence decisions (with examples from his work). meetup.com/PyData-Sloveni…

@statwonk @zapier Well done and congrats!

@hspter @NSSDeviations Sounds like a great fit, congrats!

Having built out an extensive digital database of my entire wardrobe (as listeners of @NSSDeviations have heard :laugh:), I was already convinced that this was the *only* way to make the dream of digital styling possible. It is simply too much work for the average woman to do!

I am SOOOOO excited to share that I’ve joined the founding team of a startup that is very near & dear to my heart: myindyx.com

@seanjtaylor I've been watching a couple of your talks--thanks for never presenting the same talk twice, learning something new every time! You mentioned your interest in model evaluation and continued quality control. Have you been reading something worth sharing?

@ChelseaParlett Spotify's annoy (github.com/spotify/annoy) in case recommendation makes for a valid example. "After running matrix factorization algorithms, every user/item can be represented as a vector in f-dimensional space. This library helps us search for similar users/items."

I'm lecturing on KNN tomorrow, and unlike many of the other algorithms in my class, I've never used KNN in my own work (just with simulations and "toy" datasets, not for an official analysis). 📢Do you have any cool applications of KNN you've done or seen?📢

Another week, another #F1 qualifying result: Pérez less than a car length ahead of Leclerc. Big gap between Red Bull/Ferrari and the rest of the field.

What better way to pass the time until race start than with some #rstats? The Ergast API is fantastic (ergast.com/mrd). #F1 #BahrainGP

@forde_nathaniel Yes, Europe, maybe just a copy with a short shipping distance. It also said 29th.

@TimRadtke @aloctavodia @canyon289 @junpenglao Was this in Europe? Mine says it won't ship til the 29th.

Christmas came early this year--thanks @aloctavodia @canyon289 @junpenglao !