PLenz

As a geographer turned DS it is my duty to remind CS people that all geo methods are compromises. No system, s2, h3, mgrs, even lat lon are perfect for every project and you must think carefully about what you are doing and pick your geo representation based on your givens and druthers.

Physically training the models isn't the hard part. It knowing why you using x method, how to interpret the result, and what dials to turn that's the hard part. And there's no automating that.

I just want to say that my brother and I played _the_hell_ out of Streets. So thanks for being a small part of our childhoods.

That's what locals are, the last mile (or hundred mile) trains that get made up out of cars that have moved long distances already. Railroads hate thise trains. The money is in unit trains, all one commodity going from one place to one other place over a very long distance - the exact opposite profile of locals. Railroads started making this transition in the 60s and 70s as they shed branches and secondary mains and excellerated into the 80s and 90s by spining off shortlines that would focus on local traffic.

Exactly,slack is the space give to each car if the train to move before the next car has to. Starting cars rolling is the most energy expenssive thing the motive power has to do. So once you take the slack out on a single car it's pretty cheap to keep it rolling (Issac Newton to the rescue!) - then on to the next car. So a train starts rolling one car at a time as the slack gets taken out. Crazy efficent.

Trains are already crazy efficent both fuel and operational people wise - in fact this negates one of railroading great efficencies, slack (that 2 mile freight train already starts one car at a time). Also subtracting, cars are going to need to bunch up anyway at passing sidings to let cars go by opposite direction. Besides, the real cost is railroading is MOE and MOW (including land taxes here) and this solves neither of those issues. Could be interesting for locals and LCL traffic but those haven't been big parts of railroading in decades.

"Resource Limit Is Reached" which is actually quite on-brand

Your algo can easily be parallelized by increasing the number of people firing.

Only wild if you assume that everyone has the same likelihood of having proximity to such an event. But people involved in smallpox research have a much increased chance because they have (physical) proximity to the places this can happen and those that find them have (secondary) proximity to those that have higher (physical) proximity. And Gates funds lots of these researchers so he has a very, very high likelihood of being withing a few steps of relation to something like this. Think centrality from graph theory. Gates is a very central vertex.

The last mile problem here is the jump to people's brains and it isn't easily solved by technical means. It goes beyond UX or UI and is really an empathy issue.

The great thing about mandates is that they work. Worse is better as applied to human behavior.

This stretch is owned by another rail network (ConnDot) that prioritizes other passenger trains (Metro North and Shoreline East)

Near | Passadena, CA / Remote | Machine Learning Engineer | Full-time

I'm hiring a Machine Learning Engineer to join the Data Science team at Near. This is a new, remote-friendly role that help the department to develop solutions that are more scalable at a faster pace. You won’t be just an implementer of solutions, but a full-fledged part of developing those solutions. Our Data Science team is a highly collaborative group that works closely with other Product, Engineering, and Analytics Teams to develop novel algorithmic approaches for contextualizing location data, evaluation of data from potential business partners, as well as managing data optimization from advertising data sources, and analysis of bidding data from mobile app monetization. We are the champions and guardians of Near’s data. Ours is often greenfield work - we work on the problems no one knows how to solve yet. You will be writing code for machine learning and big-data processing solutions being developed by our team as well as being a resource to help our Data Scientists write more scalable, more maintainable, more performant code themselves using better processes like code reviews. You’ll also be a translator of ‘Data Scientist’ speak to ‘Engineer’ speak and help bridge the gap between the two teams. The perfect candidate for this role will be skilled with Java, Python, or Scala.

https://apply.workable.com/nearco/j/51E57DE565/

Nothing new under the sun. Take a look at Central Park in New York. Almost everything you is is constructed. They dug the ponds and built up the hills. They didn't have geofoam but if they did they would have used it. Look a Tunnel View in Yosemite. It's there because NPS manages the trees below the view to frame it. Hell, the very idea that there is urban space and nature space is something we humans invented.

Perhaps because I only work at companies where the data is the product but I have a hard time imagining a company without a Data Science Department reporting into the CEO

I've done email heavy companies and slack heavy ones and I infinitely prefer Email heavy any day of the week. Slack is just so needy and in your face - people can 'see' you typing(!). Email gives you a little distance. So now that I'm at a slack heavy operation I regularly log out of slack for a few hours at a time. I'll get to your question when _I'm_ ready for it.

This is just reinventing unions but without the employee's ability to input via collective bargaining.

NASA used contractors to build the equipment last time too

Ha! I see our brains are in sync!

Flashing back to the Glomar Explorer/Project Azorian: https://en.wikipedia.org/wiki/Project_Azorian

The biggest issue I've had with bridging the DS<>Product gap is that our (DS) work is not linear and often not incremental. Sometime we go backwards and sometime we have to build for a really long time only to discover that this way doesn't work and we'll have to do something else and that there is usually no stopgap. ML-based solutions often don't work at all until they work.

At the beginning of my career I worked in AEC on the planning side. It was well understood that whatever the Architects had designed would be entirely redone by engineers afterwards and then by on-site engineers and then by tradespeople after that in the implementation. No one really understands what's going on in a reasonably-sized building.

I would consider moving to my own metal and using hadoop.

Census is in the Constitution as well

The best tool is one you already know

The equivlent in North America (in terms of being a once popular, now nearly forgotten fruit) is the Pawpaw: https://en.m.wikipedia.org/wiki/Asimina_triloba

I still do my NLP preprocessing in perl for this exact reason

Whenever you read in Regency to Victorian books that so-and-so has an income some pounds per year it is usually referring to these.

Yes, that's EXACTLY how it should work. Lottery

NYT writes for the Newspaper - and that means column inches. Articles get a space on the paper and are written so that they can be cut at different places to fit that space without major edits.