RA_Fisher

So China will allow loads of Western journalists to investigate without hinderance? Excellent!

Access has been heavily restricted so far, so I’m eager to see the results of China’s new-found forthrightness on the issue.

If it’s as China says: show us.

It’s a very powerful way to align incentives. It results in a major reduction in political arguments and a major increase in productivity. It’s a long-term incentive, so it causes employees to take good risks for the customer even if they won’t be apparent today.

Stock is literal ownership and it eases “ownership problems” proportionate to the meaningfulness of the expected value of the stock to the employeee-owner. It’s a powerful tool for increasing teamwork.

ML are algos not models (estimators to be precise). This model could be fit by many of them. However, fitting the model directly with the linear model will be advantageous for computational reasons.

True, it’s the law of one price. However, privileges / capabilities aren’t evenly distributed. So, while remote does widen the labor market, wealthy countries tend to carry large shares.

It’s not a zero-sum game at all. Employees from wealthy countries can gain labor surplus through technology transfer a number of ways or even find business partners and employers.

Mutually beneficial exchange is the name of the game. It’s true that different folks achieve different relative success compared to their peers. That’s okay and good when it’s the product of honest just exchange.

Ah, but it also increases the pool of potential employers.

Sure, for inefficient businesses likely to have their lunch eaten by wiser competition.

This is a better link: https://rstudio.github.io/reticulate/

RStudio supports Python these days. I use it daily and it’s really enjoyable.

I think they ought to call it DataStudio bc it’s multilanguage.

https://rstudio.com/solutions/r-and-python/

One counter example can be diabetes. Healthy insulin usage often causes weight gain, and withholding it the opposite. It’s a human growth hormone. So it’s entirely possible that one could mistake a thin person suffering from diabulemia for being healthy, and an “overweight” diabetic as unhealthy.

Sure, but for most processes that’ll yield disappointing performance.

Any tool can be used without regard for the consequences, but knowing a tool’s statistical properties yields know-how about the consequences of its application. It’s often a matter of the costs of error / stakes of the most problem at hand. Cheap solutions are often the best for cheap problems.

Exactly, it’s a smuggled assumption. A standard deviation is the second moment of a normal distribution. A negative binomial is a much better model for count data: https://statwonk.com/poisson-gamma-negative-binomial.html

This is a nicely done article. One of the issues with using a z-score for count data is it assumes the data are normally distributed. This implies values < 0 are possible. However, since counts are lower-bounded at zero, they’re asymmetric unlike the normal. This means that practically this method would produce too many false positives for most applications. Whatever threshold is chosen, is likely to have too narrow an envelope between false negative and false positive to be reliable for count data. However, for normally distributed data, this would work perfectly.

That’s because most dental care is predictable: prophylaxis. Insuring against it is just adding an admin fee. Insurance is good for unpredictable things.

Incidents are unplanned investments. They’re a decision by leadership and sap the company’s prospects. Certainly negotiate. Maybe it’s your only job? Maybe it’s a 50% pay boost? Ideally you can show the company a better way: keep records of incidents, make statistical measurements to set expectations and then change from the reactive to proactive stance. The idea is to make things sufficiently reliable _before_ the customer experiences the incident (not after!). However, on the margin, if you’re not being sufficiently compensated, you might need to find wiser leadership.

Check out stan’s variational inference algos. They’re relatively fast (compared to MCMC) at the cost of being approximative.

This is a great breakdown of risk factors. It's smart to focus on the riskiest factors.

I'm so thankful for R, it's community and their great libraries! I've built a eight year (so far) career in data science using R to model data and perform experiments. I love R's functional programming style / dplyr which makes manipulating data a delight. ggplot2 is such a great plotting library, well worth the investment to learn. Then there's all the stats tools like glm, MASS, through brms for advanced Bayesian analysis (https://github.com/paul-buerkner/brms#brms). With R and Python, it's a great time to be a statistician-programmer!

I recommend folks looking to start with R check out: https://r4ds.had.co.nz/

There are a lot of really fantastic packages, too. Many are the only implementation of a certain stats tool in the world.

Folks get jealous of success.

Yikes! Talk about theft of labor. A 25% sales tax means one in four consumptive dollars is taken away. Ouch.

Dr. Kurz wrote a really great companion that uses brms, a fantastic wrapper that makes stan easier to pickup. I highly recommend it. https://bookdown.org/ajkurz/Statistical_Rethinking_recoded/

I prefer the raw logs. One reason is that Google performs "statistical improvement" like "sessonizing." This act destroys the value of point process data. It wouldn't be found in a stats textbook because the operation destroys valuable information. Also, the concept of Visitors and Users that GA uses isn't transparent to me.

Another fun fact. Since the tracking happens on the client side, there's potentially a ton of truncated data that GA simply misses. Backend server instruments don't suffer the same way.

We have to be neutral. /sarcasm The heavy-handed opinion of HN mods drove me away from this community. It's not about protecting folks or creating a strong community. Creating those requires cultivated judgement not neutrality.

It's not, it's government censorship.

R is plenty fast. Commonly it's calling C++. When I run a multilevel model, it's C++ I'm waiting on, not R. The tidyverse + ggplot2 + statistical tools + interactive nature makes working in R really productive.

One of the great things about capitalism is that it has some redundancy and resiliency. Competition may also lead to new features (markets), etc. On the other hand, this is duplicative and wasteful. Contrast the duplication (redundancy) of the capitalist system to the one producer model of communism. Less duplication but less resiliency and less ability to find new features (market). I think when a market is young and returns are high that it makes sense for many producers. However as a market matures, we can shed some of the duplication and allocate it to other new markets. If the monopolist restricts quantity sold? Easy, it increases the tears for a new entrant. The system manages itself in this way (in reality no system at all, it's decentralized coordination).

Considering the huge flows into SF, Vancouver and Sydney real estate, that doesn't seem likely. They even institute capital controls to prevent capital flight. That's a sign that the printer is running at a high speed!

The assertion that the gov has the equilibrium interest rate (price of money) set too high is unsubstantiated in the article. We've just been through quantitative easing, so it seems unlikely we're above equilibrium. If only we'd let the market choose!

That's a great article! I read about 3/4 and I'll come back for the rest. Your arguement is the ying to my yang.

Where I depart is: wages adjust much more slowly than assets. Wages are negotiated over long spans of time vs. liquid assets. Wage holders think they're negotiating for real value X but it's eroded. Just look at baby boomers. Their wage flows are ~ fixed. Sure many get the benefit of asset appreciation but the ones that don't (get word) are crushed.

Sure but China inflated its currency, that's a redistribution to asset holders. The acomplishments are admirable all the same, but the accelerated climb is probably boosted by inflation. I predict a correction in next year's list.