Tarq0n

34 ms·

451.
by Tarq0n 7y ago
F1 score is harder to misrepresent, and relates more closely to business objectives. MSE is mostly for regression. Log-loss is very useful for training because it represents a useful signal for how well your model is progressing in fitting
452.
by Tarq0n 7y ago
Licensing and processing are entirely orthogonal rights. The former does not grant the latter under the GDPR framework.
453.
by Tarq0n 7y ago
People certainly can behave ethically.
454.
by Tarq0n 7y ago
I think many people have a problem with mathematical notation, not maths itself. Mathematical notation is optimized for terseness and suitability for use on a whiteboard. Code has made great leaps in terms of readability and expressiveness
455.
by Tarq0n 7y ago
How is that at all in the purview of an intelligence agency?
456.
by Tarq0n 7y ago
Hypothesis generation is definitely also a common goal of scientific work. There are many situations in which you don't have a good prior to base an experiment on.
457.
by Tarq0n 7y ago
Also you don't have the direct feedback loop of success or failure in the market obviously. Funding isn't directly related to performance either.
458.
by Tarq0n 7y ago
One thing about the public sector is every task of a public body is important or mandatory to some degree. This means that most strategies are necessarily overly broad and that they don't have the freedom to pivot away from or depriori
459.
by Tarq0n 7y ago
That sounds like a feature to me.
460.
by Tarq0n 7y ago
VSCode's solution is awesome. Rmarkdown works the same way, and it makes a lot of sense. Let code live as plaintext, it's native format, and only render results at runtime. Trying to preserve results in the file just makes things
461.
by Tarq0n 7y ago
Is an anti join not feasible in this scenario?
462.
by Tarq0n 7y ago
Pagedown[1] leverages paged.js to create reproducible scientific documents. It's still a little rough but I prefer it to fiddling with LaTeX templates. [1] https://github.com/rstudio/pagedown
463.
by Tarq0n 7y ago
These angle brackets are how R handles printing some Unicode to stdout on windows. In memory it should be regular Unicode though. The mistake they made is a classic R footgun: the fileEncoding argument to write.table() controls the encoding
464.
by Tarq0n 7y ago
Google assistant requires a fair amount of data collection turned on to function as well. GDPR enforcement for this stuff will hopefully happen soon. As far as I can tell this stuff is illegal in the EU.
465.
by Tarq0n 7y ago
I've used simmer (r library) and SimPy (python library) for this purpose and they were both excellent.
466.
by Tarq0n 7y ago
Could easily be type I error. That's the problem with null hypothesis testing. You can't confidently say that the intervention caused the observed change without replication.
467.
by Tarq0n 7y ago
I don't really see the issue with a regular carbon tax? Whether is redistribution or government spending, both can have positive economic impacts.
468.
by Tarq0n 7y ago
Agenda-setting in politics seems to require some amount of crying wolf, paradoxically.
469.
by Tarq0n 7y ago
The residuals of this model look really bad though, and that should tell you that there's insufficient predictive value. I'd at a minimum use a piecewise fir, or simply not fit a model without further predictors.
470.
by Tarq0n 7y ago
One interesting aspect to Simpson's paradox it that it's not straightforward which perspective is the more valid one, by groups or combined. One example in which the combined perspective is the valid one is election results (it is
471.
by Tarq0n 7y ago
That's part of it, but drawing a linear fit through the data shown in the article is outrageously misleading in itself, and should be rejected outright.
472.
by Tarq0n 7y ago
The baselines they compare to are only other neural network models, rather than models that actually state of the art for time series. This makes the author's abstract rather disingenuous.
473.
by Tarq0n 7y ago
These are not mutually exclusive categories.
474.
by Tarq0n 7y ago
Not quite. It's a fact of statistics that sometimes you'll sample extraordinary evidence for a hypothesis that's false. Check out this paper for some examples: http://www.stat.columbia.edu/~gelman/researc
475.
by Tarq0n 7y ago
Regardless of the merits of his actions, 10 years in prison seems extreme here. It's this normal for IP theft?
476.
by Tarq0n 7y ago
The Netherlands has mandatory private insurance and it works well for them. The insurers are heavily regulated though.
477.
by Tarq0n 7y ago
Considering the very high dimensionality of the space, it's not as obvious as you think. Consider the noise inherent in the word 2vec negative sampling method as well. Another word could very well end up closer to "king".
478.
by Tarq0n 7y ago
I suspect it's not incidental so much as part of a PR drive around the new moon mission.
479.
by Tarq0n 7y ago
The problem is this only works if everyone is using tabs. As long as there's significant groups of people using spaces you'll need tooling to convert it to your needs anyway. I see more point in making this tooling easier to use a
480.
by Tarq0n 7y ago
It doesn't make a difference. Rstudio's products aren't integrated with their libraries in particular.

More ›