Joshua Feldman

So, Threads is a thing now. See you there? @feldesque" target="_blank" rel="nofollow noopener">threads.net/@feldesque

@DontWantNoTweet @thetimes Haha, it was purely tongue-in-cheek – I would go for a crossword over Wordle any day.

Crosswords – Wordle for boomers? Here's a crossword of the 20 most common answers in @thetimes crossword. The clues are samples from previous editions. Code 👨‍💻: github.com/joshua-feldman… Full-size graphic 📈 : github.com/joshua-feldman… #rstats #tidytuesday #ggplot2

@mcmullarkey The University of Bristol have a great online course on multilevel modelling (including practical lessons with R): bristol.ac.uk/cmm/learning/o…

@foodforthot_ Nooo, what. That's insane. I genuinely thought that was a picture of Putin.

@the_numerist Aw, thanks Chloe!

@JoshuaFeIdman This is fantastic! Beautiful informative viz, amazing to see the whole process in R. Great project really shows off your skills! 👏🏽

I ❤️ 90s sitcoms (esp. Friends, Frasier and Seinfeld), so I looked into which was most popular at the time. It blew my mind that Seinfeld had higher ratings than Friends in the US (at least during its later seasons). A true master of its domain. FYI @JerrySeinfeld #rstats

@IgorBrigadir You're right! Once you take that into account, the #1 word is actually SOARE; LARES comes in a very respectable second place.

@JoshuaFeIdman I think the actual game has 2 arrays, 1 smaller one for possible answers and 1 larger one for possible guesses, so while it's 12972 valid words total, the real list of candidate answers (var Aa in js) is only 2315!

So I was playing Wordle by @powerlanguish and I wondered – what's the optimal first guess, i.e. the word that maximises your chances of winning? As I'm self-isolating and apparently have nothing better to do, I figured out that word is LARES. Rationale and #rstats code below 👇

Update! I learned via @MeijaJuris that Wordle only uses a subset of words as possible solutions. If we take this into account, LARES is only #2 – SOARE is #1, which is the same conclusion that @bertrandom came to.

@MeijaJuris @powerlanguish @MeijaJuris Just reran my code with the filter – I come to the same answer! #1 is SOARE and #2 is LARES. Will update my tweet, thanks for the heads up ☺️

@MeijaJuris @powerlanguish That said, I didn't know about the distinction between 'solutions' and 'herrings' in the underlying JS (I used the total Wordle dictionary as a database), so the result may be different if I filter for solutions only.

@MeijaJuris @powerlanguish I hadn't seen that actually. The approach is very similar, but it looks like the author only uses a heuristic method for the final ranking (when he says that 'right spots' are twice as important as 'wrong spots').

Here's the link to the (currently very messy) #rstats code for anyone who wants to explore further: github.com/joshua-feldman…. I deleted my earlier tweet because I noticed a bug, but the end result was the same after fixing it.

The word LARES is that which – on average – reduces the solution space the most, specifically from 10k to 300. In general, good first guesses are those that use common letters but do not repeat them.

New datasets: The 100 Best Footballers in the World 2021 (scraped from the @guardian using #rstats) ⚽️📊 👨 Male version: github.com/joshua-feldman… 👩 Female version: github.com/joshua-feldman… In male football, Man City have the most players featured, followed by Chelsea and PSG.