R is overwhelmingly used in bioinformatics. There is nothing quite like bioconductor. Most new tools/methods (for ex, in the scRNA-seq) release R packages first.
Well I'd say conda is quite like bioconductor with the ease of installing relevant packages. scRNAseq has popular r packages like seurat but also popular python packages like scanpy.
Really depends on the application. For clean, concise and reproducible ad hoc statistical analytics and modelling, there isn't a better tool than tidyverse+tidymodels.
It's a classic case of the best tool for the job. I usually create simple stuff in R and then move to bigger datasets and production in py+spark.
Although I agree and don't like R very much, I believe ggplot is still the gold standard for creating top-quality visualizations. None of the python (or other language) clones are quite as good. For projects where the end goal is a complex or detailed graph or plot, it's sometimes worth trudging through R to achieve the best final result.
Yup, not a data scientist but often do data processing to analyze the outcome of experiments (drone-flight related). I'll use Python/Jupyter if there's a significant amount of clean-up that needs to happen, but R/ggplot is unbeatable if I'm trying to look at the data from different perspectives. As an example, I was trying to look at GPS data the other day and ggplot() + geom_point() + geom_density_2d() was an absolutely perfect way to better grok what was going on.
I've used Python since 1995, so I should be biased, but switching from Python to R is a huge productivity boost - like switching from Excel to Python. R is just years ahead.
This is because R borrows a lot of syntax from S. When R came out, statisticians were using S, so it was natural to make it like this. If they went another way, you'd get statisticians in mailing lists 20 years ago bemoaning how its so much not like familiar S, rather than regular old programmers 20 years later today who bemoan that R isn't like familiar python like what happens on HN whenever there is an R thread.
I mean compared to other languages these sorts of quirks might seem like big deals, but they rarely come up. You see that error, you copy paste and find a stack overflow thread explaining it, you know what to do next time and move on. R is certainly no C.
That book isn't so much about R weirdness. It's more about teaching data scientists to consider the implications of practices like copying a huge table in memory on every loop iteration.
>Ideally grad students would stop pouring in, free labor would dry up at universities, and they’ll have to raise grad student salaries to acceptable wages again. Seems unlikely given how badly people want to do anthropology PhDs, and that there’ll always be people who can afford to take a poorly paid position like that because their partner or parents are paying the bills.
I like how you phrased that.
One potential solution is some feedback loop telling students how many PhDs in anthropology we really need. We're currently producing way more PhDs in non-STEM disciplines than there are post doctoral academic positions (I mean that broadly; e.g. post-docs, tenure track, full time lectureships).
Instead of giving out 5 PhD spots, maybe a program can give out 1 PhD spot and pay that student 3x more. They'd save money and could focus on creating one really good professor, instead of 5 struggling lecturers.
The problem is that having a large body of grad students increases your research output, allowing you to get more grants and raise the prestige of your institution at a relatively low cost. Universities are incentivized to admit more grad students if it leads to these improved outcomes.
And more students running your labs/substituting you for teaching/etc. means more time to do non-teaching stuff, which is a net positive for most professors.
How sad was I to realize that I was the only one in my PhD program who had chosen that route because I was as passionate about teaching as I was about research!
Feynman:
If you're teaching a class, you can think about the elementary things that you know very well. These things are kind of fun and delightful. It doesn't do any harm to think them over again. Is there a better way to present them? Are there any new problems associated with them? Are there any new thoughts you can make about them? The elementary things are easy to think about; if you can't think of a new thought, no harm done; what you thought about it before is good enough for the class. If you do think of something new, you're rather pleased that you have a new way of looking at it.
The questions of the students are often the source of new research. They often ask profound questions that I've thought about at times and then given up on, so to speak, for a while. It wouldn't do me any harm to think about them again and see if I can go any further now. The students may not be able to see the thing I want to answer, or the subtleties I want to think about, but they remind me of a problem by asking questions in the neighborhood of that problem. It's not so easy to remind yourself of these things.
It's not just selfish. If the school admitted fewer anthropology grad students, the marginal rejected candidate would say "Just let me in, I don't need to be guaranteed a postdoctoral position, and I'll work for a pittance -- I just really want to do my PhD." How is not admitting them kinder when they're beating down the doors to be let in and "exploited"?
From knowing a bunch of STEM PhDs over the past decade, we're producing way more of those than "needed" too. They just have cushier basically-unrelated fallback options.
It's so strange to me that the solution here people propose is to limit the free choice of students that want to further their education. A PhD does not have to be a professorship job training program anymore than a bachelor's should be job training. A more educated workforce is a good thing in its own right.
The economics are already there for increased grad students, yet we want to artificially restrict it to prevent what? Sounds more like enforced elitism.
In data engineering, I think your responses to the 'entity resolution' problem are a good Dunning Kurger style litmus test.
If you don't know, entity resolution is the process of matching unique rows in two or more databases. Are these the same movies? Are these the same person?
Novice DE: Oh easy, just merge on the name.
Intermediate DE: OH GOD NO. <michael_scott_no.jpg>
