waltherg

Great write-up, thank you. Do you have rough measures for what constitutes high/mid/low- dimensional data? And how do you use XGBoost et al for multi-step forecasting, I.e. in scenarios where you want to predict multiple time steps in the future?

OP pointed out that meetings and projects have happened remotely anyways for a long time - an experience I share. The peers you describe aren’t in the same place as yourself even if you’re both in some office.

Yes, I read this and was thinking therapy from the start. I was never against therapy but finally pulled through and my mood and outlook have improved dramatically over the past six months thanks to my weekly therapy sessions.

OP’s username seems somewhat German so I would cordially invite OP to make use of our pretty decent Krankenkasse system that makes it a relative breeze to get therapy sessions. Shop around if you don’t vibe with your therapist.

Come on mate what's with the pessimism. When did the rich upper class of Italy fail to deliver for the remaining 99% of Italy?

Oh wait ...

Gosh. Did Winterpeg for a year. That was an experience for sure. So long Windypeg.

In the south of Italy where I am right now there appears to be a pattern of pre-lunch and pre-dinner rush hour. You get some pretty decent cycling paths along the sea between 1 pm and 5 pm which turn into what can only be described as death-wish-routes outside that time window.

I thought the point of those ads wasn't that Brave make money but that you accrue those Brave tokens as a representation of your attention and get to send those tokens to publishers of your choice via the browser?

Also, I turned them off as these ads are quite annoying and have a "cheap feel" to them.

The value proof in the backend that convinces advertisers to throw more many at this? Mostly rule of thumb heuristics with a fair amount of overselling. So no, in most cases the telemetry just doesn't prove much.

I've seen the thesis that all this advertising revenue - even if poorly spent - subvents large portions of exciting research in deep learning etc. at the likes of Google and Facebook. So all that talent wouldn't entirely be lost to advertising.

This is a great point and way of looking at it: pandas and SQL live at opposite ends of the maturation level of data wrangling pipelines.

There's usually entire teams involved in preparing these sorts of decks - from content researchers via subject matter experts to designers. So not a single person needs to believe in it all at once.

I think a far better question is: Does whoever is paying for these decks to be prepared believe in them? These decks aren't (always) built for internal use.

So, essentially, you live in Italy where traffic signs and lights are nothing more than street art and some extra frills on the side of the road.

Mice are really into cheese - hence become a mouse to avoid Alzheimer's later in life.

I don't doubt that marketing works but always had a feeling that a lot of it is busy work (e.g. reorganizing AdWords campaign structures every couple of weeks for no apparent reason) and a lot of data-driven success stories probably boil down to right time right place.

Clean randomized experiments naturally provide good answers and are especially feasible in this space but performing those properly, especially with regard to attribution and user identification, presents a whole zoo of issues that are inherently different from what a lot of marketers seem to work on.

I'm intrigued by what you wrote about CNAME records. Could you elaborate on that briefly?

I usually have at least some broad, high level understanding of what tech outside my narrow scope looks like.

Reading the comments here I have zero clue what you're talking about.

Is this what my less technically inclined colleagues feel like when people talk about e.g. machine learning and Python?

German here. The latter statement doesn't really hold for us. Super patchy coverage on trains.

The - to my mind - most absurd version of this was a well-known Swedish fintech I interviewed with where the one barrier they had set up was one of those online IQ tests (rotating, mirroring etc. blobs on a grid).

I was interviewing for a senior data science role and the other points of contact I had in the process were surprisingly non-technical conversations with a VP of data and another senior data scientist.

Alas, my skills in rotating blobs on a grid failed me that day so that was that.

I've heard two C-level guys with two separate German SMEs say that the only part of their workforce that hadn't handled remote work well was middle management.

The people doing the actual work were happy WFH and simply getting stuff done. Senior management / C-level types were content seeing sales figures and general output from afar.

Middle management struggled because they had a hard time judging work estimates for tasks and whether people had their butts in seats etc.

Just an anecdote but thought that was intriguing.

Lots of loose ends still and nothing noteworthy to the extend that others are posting on here.

But I finally got around to reading up on a topic I've been curious about (causality modelling, causal inference, causal discovery) and started writing a little about it. The couple of interactions I gleaned from doing so have been very refreshing and are a great driver to continue down this path.

https://georg.io/the_causality-driven_company

https://twitter.com/GeorgRWalther/status/1271796549157339136

https://twitter.com/GeorgRWalther/status/1272193903639375872

Been there so many times.

I've been kicked off a project because I couldn't stop talking about training data (they already had) and machine learning algorithms (NLP, text classification) that we could implement right now to start automating a couple of internal processes that are currently pointlessly manual. Think moving incoming e-mail to appropriate downstream support channels.

Not enough magical AI / AGI in there.

The countless PowerPoint presentations built after my departure described processes along the lines of "the AI will detect when you're about to miss your connecting flight and book you into your favourite hotel with your favourite dinner pre-ordered".

Surprisingly, the project survived and now they collect training data and use machine learning for text classification.

Any chance at sharing what consulting company put the AI play idea in your CEO's head?

Hi to anyone reading this. This is my first attempt at describing applications of causal inference in the context of possible industrial applications.

I started this article because I often find commonly used examples from economics and medical research hard to relate to.

Both my understanding of this topic and my article likely need further work so I would appreciate any constructive criticism!

Since you’re here, what does the following property stand for in your second paper?

s ∈ AifM(s)/gec

I have seen similar examples of pseudo-distributed Hadoop Docker images that baked all HDFS and YARN daemons into the same container at runtime.

With this toy example I set out to learn more about Hadoop clusters by untangling the various components and move them to isolated containers that communicate via a Docker bridge network.

Hope anyone finds this useful!

Just in case anyone has time to help me out:

What's the difference between trades and quotes here in this chart? And how are the trader's order and purchases indicated?

Have you ever worked in scientific research yourself?

Scientists sell their story and their version of The Truth just as much as reporters sell their news stories. This image of the objective scientist that prevails outside of academia is mostly false.

If I remember correctly, you'll find more material to think about here (see researching what your boss wants you to, researching what funding agencies want you to, etc. etc.):

https://news.ycombinator.com/item?id=7763737

https://news.ycombinator.com/item?id=122106