tylermw

Completely unrelated concept: unit testing is part of a regression testing framework while OFAT is a (almost always suboptimal) test strategy for designed experiments.

See:

https://en.m.wikipedia.org/wiki/One-factor-at-a-time_method

> If you collect 3 different data points changing each thing one at a time (original design, some number higher, some number lower) whilst keeping everything else constant (usually a good scientific approach) that's 320 possible combinations of changes!

There is an entire field of statistics (Design of Experiments) where one of the first lessons you learn on day one is how one-factor-at-a-time testing is one of the most inefficient ways you can test something. It’s usually only done out of ignorance to better methods by those with little to no formal statistical training.

An experiment designed by someone who is well versed in modern experimental design methods would not take billions of runs to optimize—a sequential design that first screens out factors to those that matter (basic Pareto principle) followed by a response surface design or a GP model surrogate to optimize the response would likely be on the order of hundreds (possibly thousands) of runs. This is basic industrial experimentation—see “Design and Analysis of Experiments” by Douglas C. Montgomery for a nice introductory textbook.

The helium used to fill balloons is no different than the helium used to cool MRIs. The latter has just been liquified.

Several of the maps were made by Twitter users @researchremora and @cstats1, using R and the rayshader package.

No problem! Nice to hear you found the writing and code clear and easy to understand!

Thank you!

Thanks, glad you enjoyed it!

I'm always really impressed with the clever techniques and methods to accurately map topographic surfaces (in this case, lunar) before computerization. Really interesting!

(and your comment reminded me to finally place my order for a Klein bottle!)

Opensubdiv would definitely be the robust, industry standard solution. However, R packages that are distributed on the CRAN have additional restrictions on required system libraries, so for portability I went with a bespoke implementation.

My personal favorite resource is "R for Data Science" by Hadley Wickham. It covers lots of nice data manipulation and visualization examples, and provides a good introduction to the tidyverse, which is a particular dialect of R that's well-suited for data analysis. It's available for free at:

https://r4ds.hadley.nz/

For more specialized analytical methods there are lots of textbooks out there that provide a deep dive into packages for a specific field (e.g. survival analysis, machine learning, time series), but for general data manipulation and visualization it's hard to beat R4DS.

I have not. I started using R due to its open source codebase and ability to audit and understand exactly what its doing under the hood—being able to see how statistical formulae were implemented in code was invaluable in understanding and interpreting a package's analytical output.

I remapped the UV coords based on the spherical projection of the mesh after subdividing, so there should be minimal distortion, especially compared the UV sphere. There is a slightly higher density of vertices where the edges of the cube used to be, but it's small compared to the UV sphere's extreme convergence at the poles.

R is hands-down the best language for data manipulation, analysis, and visualization: it's a language truly centered around treating data as a first class citizen. That focus does make some traditional programming workflows more error prone (helpful interactive data analysis features like vector recycling, flexible automatic type conversion, and non-standard evaluation provide lots of footguns), but the last decade of language improvements (stringsAsFactors = FALSE!) and R packaging ecosystem improvements have made the situation much nicer. The flexibility and lispy expressiveness of the language make it really fun to develop in, once you've gotten over the initial quirks.

Thanks for the kind words! You can also see the documentation/examples for {rayrender}, the pathtracer I wrote for R, here:

https://www.rayrender.net

And the rasterizer/mesh manipulation package {rayvertex} here:

https://www.rayvertex.com

Thanks! Me too (Obviously)!

> but aren't kid's <14 supposed to be supervised at all times by an adult

No. The age which supervision is required varies by jurisdiction.

> And 14+ years old that drowns in a pool for inability to swim is also something that should be blamable on the parents, not the pool owner.

There's no law stating you need to teach your kids to swim.

I believe they are using the percentages as part of the height of the bar chart! I thought I'd seen every way someone could do dataviz wrong (particularly with a bar chart), but this one is new to me.

Just wanted to say I recently converted my personal blog [1] from Wordpress to Quarto and it's been a joy to work in! The fantastic support for literate programming tools (code highlighting/annotation) and cross-referencing has taken away much of the friction to writing and releasing new posts.

[1] See profile

The word you’re looking for is “cheaper,” not “accessible.” Those 1x1m elevators certainly aren’t accessible for wheelchair users. ADA requirements require the following minimum dimensions for an elevator:

“The width of the elevator car is a minimum of 80 inches (2030 mm). The depth of the elevator car measured from the back wall to the elevator door is a minimum of 54 inches (1370 mm). The depth of the elevator car measured from the back wall to the control panel is a minimum of 51 inches (1291 mm).”

Agreed--the R base pipe and right assignment operator obviate the need for a pipe assignment operator.

As an added plus, they are one of the few natural predators of stink bugs!

I see no distinction there.

R’s data visualization capabilities are heads and shoulders above Matlab’s, and it’s free and open source.

> R is too specific and it lacks the vast ecosystem of Python

That's the first time I've ever heard this criticism of R: most staunch R critics will acknowledge R's vast breadth and depth of statistical libraries outclasses Python. Python has far more general use programming libraries (so deploying models is easier) and used to have a slight advantage in ML specific libraries, but there aren't major differences nowadays between the two ecosystems.

Many planes take off northwards, and immediately bank to follow the river.

Famously, see:

https://en.wikipedia.org/wiki/Air_Florida_Flight_90

R is not constantly breaking. R Core does a remarkable job ensuring backwards compatibility. There are only a few prominent examples of significant "breaking" behavior across decades of the language existing, and those can often be reverted by setting an option (e.g. `options(stringsAsFactors = TRUE)`). But backwards compatibility is the primary concern with any update to the R language or the packages maintained by R Core.

Now, if you're thinking about changes introduced by a specific user-contributed package breaking your analysis, that can indeed be a problem. But that can't be blamed on the R language. And the main user-contributed R statistics packages that have been around for decades (such as lme4 or survival) are mature and stable.

The modern "data science" solution, however, is to make your 3D animated rollercoaster in R! Much more efficient :)

Link: https://www.tylermw.com/datacoaster-tycoon/

Note: the 4th edition of this book will likely be out by the end of this year, and cover GPU pathtracing. (I believe the free web edition will be released a few months after the hardback)

The Early Release (still in-development) version of pbrt-v4 can be found here:

https://github.com/mmp/pbrt-v4

I used and extended this tutorial to write the rayvertex [1][2] package, which is a deferred parallel software rasterizer in R. Explicitly building the rendering pipeline really helped me understand the underlying data flow and rasterization techniques, which in turn made me understand GLSL much better. And extending the rasterizer with order-independent transparency, multicore rendering, antialiasing, and other niceties made me appreciate the work that goes into modern renderers.

Highly recommend.

[1] https://www.rayvertex.com/ [2] https://github.com/tylermorganwall/rayvertex

Check out the rayshader R package for lots of methods for hillshading terrain maps (original HN post mentioned here as well, but it's evolved well beyond that original post)

Website:

https://www.rayshader.com