Notices by Alan-Tactical Nuclear 🐧 (alan@mastodon.technology)

The hard truth

TIL: in Excel if you refer to a cell formatted as a percentage (so for instance you have 0.05 and you see it as 5%) the resulting value gets multiplied by 100.

This is one of the cases where the line between bug and feature ahould be clearer 😓

I'm announcing panthera: a #Clojure wrapper around pandas and numpy for all your data sciencey work!

The library is still alpha, but there's already a lot of stuff to have fun with. So give it a run, if you like it star it and any kind of feedback is accepted!!!

https://github.com/alanmarazzi/panthera

Unexpected data leaking
https://m.xkcd.com/2169/

It turns out that cats are even weirder than we thought

http://flip.it/5NdwJ6

A weekend in the middle of nothing it's what I need after moving to one of the busiest cities in Italy

@bthall @lnxw37d2 @bthall I worked in this realm, I suggest to learn Spark early on

@bthall because race bias usually is closely tied to income, and this is tied to where you live.

As an example: an African-American kid growing up in a white, wealthy neighborhood is more likely to be similar to his neighbors than to other African-Americans living in a poor hood

@yogthos and then let's think what happens when we apply that model to EVERYTHING

Guess what, the Sun (yes, that huge burning thing that gives us energy and was our only source of light for MILLENIA) doesn't kill https://www.outsideonline.com/2380751/sunscreen-sun-exposure-skin-cancer-science

When evaluating a #machinelearning/#datascience problem always account time for every step:
- framing the problem
- defining data sources
- ETL
- Data analysis
- Modeling
- Evaluation
- Deployment
If you account only modeling you're saying: "Building this app took 30 minutes!" when that is just the compile/build time

Many articles about AI issues are popping out everywhere: https://towardsdatascience.com/when-automation-bites-back-8541b061d5ee

I'd say: Winter is coming!