I would choose instant thou.
948 karma · joined July 30, 2022
I would choose instant thou.
And furthermore the models are always chosen from a predefined hypothesis set, so there can never be truly automated specification.
Did you come up with this yourself, or did you copy it from somewhere else? Because I would like to know more. This sounds like a very good approach.
> I'm thinking what the the most optimal way might be
From my experience - don't even go looking for "most optimal ways" in this journey. The simpler, the more unstructured - the better. Otherwise maintaining the whole setup will take more than give.
If I add the time I spend on NH over the week, some weeks it can probably reach 3-5 hours. That's 3-5 hours reading about something, checking comments, sometimes replying. But there is no substantial, measurable gain that would justify spending this much time.
If instead I spend the same amount of time reading a book, meditating, socialising, working, or even staring through the window - all of these would be time better spent, in my assessment.
I think this is overblowing the similarities to matrices. Matrices have elements all of the same type, while data.frames mix numbers, characters, factors, etc. You certainly cannot transpose a data.frame and still have a data.frame that makes sense. Multiplying rows would not make sense either, since within one row you will have different types of data. Unless you have a data.frame that is all numeric, but in that case one should probably be using a matrix in the first place.
1. code for performing a scientific simulation or analysis (a script).
2. code for solving a specific problem generally (a program).
There are different "best practices" for the two situations above. And the article primarily talks about applying "best practices" from the 2nd scenario to the 1st. Of course they don't apply.For instance, the shown example filters for "Bug" but seems to filter for rows that contain "Bug" in any of the columns. Can I specify to filter for "Bug" only looking at a certain column?
At the same time - is filtering by numbers implemented, like filtering for rows that have a numeric value above X in a certain column?
If not these should be on TODO list - those are very common operations for .csv type data.
For instance, about epigenetic clocks, there are a lot of them now. There is GrimAge for mortality, FitAge for fitness, and dozens of different clocks for chronological age. I cannot know for sure, but I am almost sure Bryan tried at least a few. Why did they select to showcase only this one? Is it because they liked the results the most?
Same for all other markers. All of them are "optimal" "above 95%", etc. Is there no marker that is not so great and can still be improved? Also showing the history of the measurements (how they fluctuate during years Bryan is on the protocol) would be wonderful. Or measuring the same markers for a different person, who is above him in the leaderboard, but not going through the protocol.
I want to repeat that I love what he is doing. But for some reason the website gives a marketing vibe. This is our protocol, everything is optimal, here is our olive oil. Which is a bit of a weird look, given the lengths Bryan Johnson is going through.
But on the other hand, I get a sense that the public side of his results are overblown. For example, my field is epigenetics, so I had a look at his "epigenetic clock" results. He uses "DunedinPACE" to track the rate of aging and his result is supposedly 0.72 (which very roughly estimates that he ages 0.72 years per a single astronomical year).
However, what is not mentioned, is that this result, while impressive, is not so extraordinary. He is number 6 in his own online leaderboard [1]. And the people who beat him at this metric don't do anything fancy to get better numbers than him [2]. Why not mention things like that along with all the optimism?
[1]: https://rejuvenationolympics.com/leaderboard/#absolute
[2]: https://fortune.com/well/2023/11/04/longevity-women-biohacke...
Is this about that circadian rhythm line? It look true, if not under-stated. Shift between nights and days was one of the major constant things throughout the evolution of life. And the influence of this rhythm goes deeper than just physiology - it reaches a molecular level. Almost every cell in the body has a mechanism for keeping internal time of the day [1]. And they are kept in synch through suprachiasmatic nucleus, mainly via the light through the retina, changes in temperature, and food intake [2].
It is so prevalent, that even small things, like your sense of smell or which nasal side is more active and which one is congested falls under circadian regulation [3].
[1]: https://en.wikipedia.org/wiki/Circadian_clock
"Politics has nothing to do with what a person personally advocates for." - Norman Finkelstein
An example he often gives is advocating for a world without borders. You can advocate for it, and create a group of like minded peiple and write blogs about it, but it will be like creating a cult. Politics, according to him, starts with what the people want. It has to have some concrete proposals and practical momentum that are accepted (or at least not resisted) by the general public.
1. "I think your view about the difference between Frequentist and Bayesian methods is biased"
2. "I think your view about the difference between Frequentist and Bayesian methods is wrong".
3. Any other suggestion?I think your view about the difference between Frequentist and Bayesian methods is wrong. There is this rant I like from Larry Wasserman [1] on the subject:
My opinions have shifted a bit. [...] Bayes-Frequentist debate still matters. And people — including many statisticians — are still confused about the distinction. I thought the basic Bayes-Frequentist debate was behind us. A year and a half of blogging (as well as reading other blogs) convinced me I was wrong here too. And this still does matter.
My emphasis on high-dimensional models is germane, however. In our world of high-dimensional, complex models I can’t see how anyone can interpret the output of a Bayesian analysis in any meaningful way.
I wish people were clearer about what Bayes is/is not and what frequentist inference is/is not. Bayes is the analysis of subjective beliefs but provides no frequency guarantees. Frequentist inference is about making procedures that have frequency guarantees but makes no pretense of representing anyone’s beliefs. In the high dimensional world, you have to choose: objective frequency guarantees or subjective beliefs. Choose whichever you prefer, but you can’t have both. I don’t care which one people pick; I just wish they would be clear about what they are giving up when they make their choice.
[...]
Of course, one can embrace objective Bayesian inference. If this means “Bayesian procedures with good frequentist properties” then I am all for it. But this is just frequentist inference in Bayesian clothing."
[1]: https://errorstatistics.com/2013/12/27/deconstructing-larry-...- "The fate of empires and search for survival"
- "The population cycle drives human history"
- "Our fragile intellect"
To me, paradoxically, it reached a point where my subconscious associates light or poor typography with serious material, and pretty web pages with empty ad-ridden content:
Serious vibe: http://fmwww.bc.edu/repec/bocode/t/transint.html
Intermediate vibe: https://drewdevault.com/2023/06/30/Social-and-parasocial-med...
Empty content vibe: https://www.theverge.com/2023/7/12/23792382/google-notebookl...
> When I try to get this point across about techniques like the PCA, I like to show that the measurement units strongly affect the inference.
In such a case the problem is not with PCA but with application. PCA is just a rotation of the original coordinate system that projects the data on new axes which are aligned with the directions of highest variability. It is not the job of PCA to parse out the origin of that variability (is it because of different units, or different effects).
> Really, if your conclusions change depending on whether you measure in inches or centimeters, there’s something wrong with the analysis!
To get a statistical distance one should: subtract the mean if the measurements differ in origin; divide by standard deviation if the measurements differ in scale; rotate (or equivalently compute Mahalanobis distance) if the measurements are dependant (co-vary). The PCA itself is closely related to Mahalanobis distance: Euclidian distance on PCA-transformed data should be equivalent to Mahalanobis distance on the original data. So, saying that something is wrong with PCA because it doesn't take units of measurement into account is close to saying that something is wrong with dividing by standard deviation because it doesn't subtract the mean.
[1]: https://www.amazon.com/Reign-Quantity-Signs-Times/dp/0900588...
> Since there is at least one algorithm (WESP) that is NP (verifiable in polynomial time) but not P (solvable in polynomial time), P cannot be the same set as NP.
The gist of it: you program procedurally without thinking too much about the code and only add "compressions" when the need for them arises. The compressions can take various forms like encapsulating data types within objects, behaviour within functions, extracting often-repeated operations into separate libraries and then loading and referencing them when needed, etc.