Here's a wrong answer😅 ------ I'm glad you asked! Buckle up for a wild ride on the cutting edges of business intelligence and dataviz ! Here's a 16 tweet primer with all the deets you need to level up your BI adventure! /1 - ThreadSky

rshean.bsky.social • 9 days ago

Here's a wrong answer😅
------

I'm glad you asked! Buckle up for a wild ride on the cutting edges of business intelligence and dataviz ! Here's a 16 tweet primer with all the deets you need to level up your BI adventure! /1

Comments

rshean.bsky.social•9 days ago

Meet #PowerBI, Microsoft's new low/no-code business solution that still somehow requires learning 2 different scripting languages, M and DAX, that were (probably) ported over from excel. With power bi you get to do data ELT with enhanced 💪💪excel formulas and create reactivate functions with DAX /2

rshean.bsky.social•9 days ago

People say that M is sooooo much faster than #rstats* or Python for data ELT, but I really wonder how that benchmark was calculated. /3

*Footnote: R has its own special weird hashtag ❤️❤️❤️

rshean.bsky.social•9 days ago

Inefficient python/R is indeed pretty slow but there are so many different ways to optimise it to get orders of magnitude faster. Even if M is innately and categorically faster than python or R, (doubt it), it's not like those languages are optimized for speed! #Developertime💪computertime😴 /4

rshean.bsky.social•9 days ago

The solution for slow code, might not be power BI, it might be fast code. 🤔💡🤯 It's possible to rewrite the slow bits of an ELT pipeline in Rust or C++, or even better just use one of the many rust/c++ tools ppl have already built for R/python /5

rshean.bsky.social•9 days ago

or even even better realize you don't actually need ALL the data ALL the time and switch to using a database and only loading subsets into working memory /6

rshean.bsky.social•9 days ago

Anyways, dashboards! The DAX combined with the point and click interface let you create stunning, ermh, acceptable, interactive visuals and dashboards in an only partially reproducible or version controlable way. /7

randvegan.bsky.social•2 days ago

Who in the world says M is faster the R? In no universe have I ever seen this

rshean.bsky.social•2 days ago

Also like R and python were never designed for speed. 😅They were designed for usability and then later some really cool people added stuff to make them faster.

If I wanted to put in the extra effort to learn a new language for speed, it wouldn't be M! It'd be a compiled language like c++ or rust

rshean.bsky.social•2 days ago

I've seen some things on reddit and quora saying that M is optimized for large data transformations!

I've never run any speed tests, but my guess is they compared M to less than perfectly written/structured R. R speed can vary quite a bit depending on how it's written and which packages are used

randvegan.bsky.social•2 days ago

these tests must not have included joins -- because that is where M falls apart.

rshean.bsky.social•2 days ago

Guess not! 😅

mpj.fff.dev•9 days ago

Oh this was so good, thank you dearly

rshean.bsky.social•9 days ago

Always happy to share the joy of power bi with others 😌😆

I didn't realize that bluesky sort of hides post after about ten in a row, but the thread keeps going past
/11 😅

mpj.fff.dev•8 days ago

Haha, I sort of love these quirks, takes me back to the early days of Twitter (I don’t think most people know that @replies wasn’t part of it initially and grew as a phenomena, not a feature

Comments

Posting Rules

Reply