jlamberts

I'd agree with you if this applied to primary residences, but it seems like this only applies to secondary residences? I find it hard to reconcile "middle class" with "has a second home in NYC"

Re: existing model, for recsys, as long as the product already exists you have some baseline available, even if it's not very good. Anything from "alphabetical order" to "random order" to "most popular" (a reasonable starting point for a lot of cases) is a baseline model.

I agree that a randomized control is extremely valuable, but more as a way to collect unbiased data than a way to validate that you're outperforming random: it's pretty difficult to do worse than random in most recommendation problems. A more palatable way to introduce some randomness is by showing a random item in a specific position with some probability, rather than showing totally random items for a given user/session. This has the advantage of not ruining the experience for an unlucky user when they get a page of things totally unrelated to their interests.

Yes, it's just really nice to be able to push a button and have it hold at the right temperature; sort of like how boiling water on the stove or in the microwave and using an electric kettle are functionally equivalent but the kettle is way more convenient.

Guessing the hygenics depend on your local water supply, but some varieties of tea (and coffee actually) want to be brewed at below-boiling

A little desktop app that lets me upload transaction csvs from my bank and figure out how much I need to split with my partner. Mostly because I always forget to charge her for utilities or flight bookings and I hate going through the bank UI. Might also expand it with some simple subscription auditing logic.

Also, a dramatic anime intro (complete with cheesy AI generated theme song and video) starring our foster kittens. It's been interesting to learn about some of the techniques needed for consistency, how to storyboard, etc.

I don't see how their analogy is stupid. Aeropress and french press are pretty similar from an "enthusiast coffee device" perspective. Lots of room for variability in grind size, coffee choice, and specific brewing technique with both methods.

Might have been k-nearest-neighbors rather than k-means. Knn can be used for "recommended because you bought X" or "users like you also bought X" type recommendations that relate user to user or item to item.

K-means could potentially be helpful to group together common users/items if e.g. you're memory constrained and don't want to give each user a fully unique embedding entry so that's also possible.

With free games (which make up a pretty large portion of competitive online games), cheaters who get banned usually just make or buy new accounts.

As part of the Intuit purchase Credit Karma had to sell off their tax service, which Square bought. I'm guessing that's what OP is referring to.

I have also recently been playing around with Julia coming from numpy/pandas. I found that 1.6.1 felt way smoother on startup and import than 1.5.x. Might be worth trying to upgrade if you haven't already. At the very least it's encouraging that it seems like they're making progress in that direction.

This is actually the idea behind Direct Primary Care[1]; basically the idea is that by not taking insurance the overall cost of care goes down due to the decreased overhead around the claims process.

My dad is a PCP who switched his practice over to that model several years ago. From what he's told me he prefers it a lot over the traditional model specifically because it allows him to have a subscription based model rather than charging per-visit. That means that his revenues are more consistent but also has the benefit of aligning the patients' and doctors' incentives more directly: the doctor makes more money when their patients are healthy than when they are sick so they're incentivized to do a better job.

It also meshes well with telemedicine and chat-based consultations. He was offering both of those options to his patients well before covid because it tends to save both doctor and patient a lot of time, and there's no concern over "how much do I bill for telemedicine vs a normal visit, and how do I bill insurance for it."

[1] https://en.wikipedia.org/wiki/Direct_primary_care

I've heard "the tallest blade of grass gets cut first" in the US, although Google says it's either a Russian or Assyrian proverb originally.

I was also curious. Using an out-of-the-box weighted random forest, I got an f-score of ~.85 using a 75:25 stratified train-test split.

To encourage active participation, I have created a private repo where all future edits and entries will be made starting November 2019 after which this project will be deprecated. To join indefinitely you have to submit two unique open source python ML or DS projects to d.snow@firmai.org (unique to this list only, it needen't be your own). Please submit them now so that I can put you on the list. How to know if your contribution is unique: search the Google Sheet file for the respective URL.

So the intent is to "lock out" anyone who doesn't contribute before November 2019 by keeping all new content in the private repo? Seems kind of sketchy, unless I'm misunderstanding. Maybe the intent behind releasing this publicly is just to build up the email list mentioned in the readme?

I tried a few: runbox, protonmail, and fastmail, and ended up staying with fastmail. The web interface isn't pretty but is functional and the service has been great so far. Very easy to set up with my own domain.

I graduated from the program in December. Overall, the program was absolutely worth it for me. Based on my experience I'd say that the online program isn't quite at the same level of instruction quality as an equivalent in-person program simply due to the number of students in each class. However, for me personally the lack of cost (both real cost and opportunity cost from lost work years when pursuing a fulltime MS) more than made up for it.

I haven't interviewed or applied for jobs since graduating, but it's worth noting that the degree itself is the same as the one awarded in-person. If you get involved in the community, the network is also surprisingly good; the slack channel has hundreds of students at all sorts of places.

For anyone who's interested in a more long-form, intuitive explanation of the fourier transform and how you can use it to multiply polynomials, I'd highly recommend Erik Demaine's lecture on the topic: https://www.youtube.com/watch?v=iTMn0Kt18tg.

In my experience, it's a combination of a couple of things. First off, the field is heavily regulated (in the US anyway) and the penalties for violating regulations like HIPAA are incredibly high. Second, the field is currently dominated by major players such as Epic, so its pretty important to be compatible with them, but they don't really have an incentive to open up their ecosystem since they have such a stranglehold on the market. Finally, a lot of medical folks have been burned by technology in the past, and, in my experience, often view a lot of the tech they have to use as an insurance and government mandated evil, rather than a way to make their lives better.

Not to say things can't be improved, but there are a lot of factors that make it more difficult than a traditional B2B or B2C product.

By the utilitarian argument, wouldn't the answer just be to stop eating meat and eat whatever we're feeding to livestock (or some other food grown in its place)? Meat is a pretty inefficient source of food compared to large scale agriculture.

I think the idea for the original product (I.e. non cypto) is that it's target the "little guy" with no commission on trades. So people feel like they're taking from the rich institutional investors.

I think they're saying that they're being taxed as if they were getting 4% interest on their savings, when in reality they're getting less than 1% because rates are low. So even though it's nominally a tax on interest earned, it's effectively a tax on value of assets held in savings over 21k.

Well, it's a resume service. Gotta include at least one JS framework to make it past HR.

I would love this as an individual when learning new languages on my own projects. I find it really hard to tell if I'm actually doing things the "right" way without talking to someone more experienced.

ML flows nicely into RL, although I've heard ML4T is a gentler intro if you have no experience in machine learning at all (I haven't taken it yet)

I'm currently in the program with a BS in Industrial Engineering. I highlighted the CS side of my job in my application, and had done a lot of learning on my own time between my BS and application. So it's definitely possible if you're interested.

We always used to go up to Torch Lake during summer when I was a kid. Collected a bunch of these. The easiest way to find them is to bring some water with you and go for a walk on a rocky road, and pour the water on likely rocks. The water makes the pattern much clearer and fewer people look for them on the roads.

I think in theory most people would agree that a benevolent dictator is a good thing both economically (monopolies) and politically (actual dictators). The problem is that human nature gets in the way in both cases, so you can't really trust a company or a leader to stay benevolent (or that their successor will be even if they are).

It's also possible that the heart benefits outweigh the side effects. A lot of medicine is just saying "this sucks but if you don't do it you'll die" and prescribing the least bad option.

Bayesian statistics says the opposite. By looking at the history of an entire city, you're giving yourself a highly uninformative prior. This means that it's very difficult to actually draw conclusions from the results. You have to reduce your population size first, and then perform your "test" (looking at search history) on the subset of the population who actually has a reasonable chance of committing the crime even without the search evidence.