Eden @edensaesthetic@witches.town

And I think that's probably best for me. Manipulation and feature engineering in R, pull the results into python and build models.

So the python I'm playing with is just doing my pipeline in R for Kaggle, but in python instead.

And like, it's not awful. I still prefer rapid data exploration and manipulation in R, but I think model setup is actually cleaner in python. It's easier to modify evaluation and update on step progress.

Like in R I'll just be left waiting without a clue how long it has left, so this is way preferable.

Apparently an alarm was pulled on the tube, so now I'm wicked late for counselling :/

@bea Yeah that's fair. Like I'm not really sure how the indexing works, which is weird when I feel exceptionally comfortable with R's take on indexes.

Sometimes I can just be like
Where A == TRUE & B == FALSE, and sometimes it's very unclear what it's looking for.

Pandas feels so out of place for Python when Python isn't isn't vectorised in most of its operations.

@patcoet I like using IDEs and when EVERYTHING HAS GREEN LINES I'm just annoyed.

tell me what's broken, not just "YOU NEED TO BE x + y NOT x+Y"

@patcoet The style guide is legit why I hate it but I'm also aware that that toot got a fair amount of attention and I don't want actual arguments :')

@bea You're not wrong. It does its work well. It is less brick-wall than R.

(Okay I've been using Python for about 2 hours now and I don't hate it. I'm still not forgiving it for all that syntax shit about for loops, but I'm begrudingly starting to like it)

It is so freaking cold today. I'm trying to do things, but I also just wanna nap.

Apparently Trump's UK visit has been set for February? Like a load of weird papers who are usually full of shit are reporting it, like the Indie and Sun.

I should mention I'm actually off work today, I don't get quite this much freedom to dick about at work. :')

Huh. I've honestly been using caret quite a lot in ML because I'm pretty lazy when it comes to using someone's algorithm, and I really like adaptive resampling.

and like I'm looking at the xgboost documentation and it's like I was just failing myself the last few months. THERE'S SO MUCH STUFF

@balrogboogie I mean, if that's the cost to be platform agnostic? :(

@LottieVixen definitely do clojure! There's a book called Clojure for the Brave and True that's legit really good :)

@LottieVixen mate setting it up is the most painful bit. It's not awful to write in though, just a bit unusual.

I don't get why it's so much more popular than clojure?

It also has nulls, ick :(