Disqus - Latest Comments for hadley

Re: Winners Take All in the Dependency World (or Hell)

Hadley Wickham — Tue, 13 Nov 2018 12:13:40 -0000

What is the real problem here? Social media? OR BABIES?

Re: Winners Take All in the Dependency World (or Hell)

Hadley Wickham — Tue, 13 Nov 2018 09:20:53 -0000

I think I must disagree with your definition of marketing, because I think you do a lot of it! From my perspective, some of your recent marketing initiatives are:

* The bookdown website listing all the books created with it.
* The bookdown competition.
* Having a blog where you talk about your work (and deliberately not promoting it via twitter!).
* Making the bookdown and blogdown books available for free.

I think you discount the importance of marketing because you are so good at it.

Re: rstudio::conf(2019) diversity scholarships

Hadley Wickham — Mon, 10 Sep 2018 08:37:13 -0000

We would love to help more people, but given the limited number of international scholarships available we decided to focus on just a couple of regions to hopefully maximise our impact. We will definitely think about how to expand the program next year.

Re: rstudio::conf(2019) diversity scholarships

Hadley Wickham — Mon, 10 Sep 2018 08:36:19 -0000

Re: rstudio::conf(2019) diversity scholarships

Hadley Wickham — Wed, 22 Aug 2018 10:44:22 -0000

Yes, sorry - applications close Sept 15 and we'll aim to make decisions by Oct 1.

(If you are concerned about missing out on early bird pricing you can buy now and we'll refund you if you get a scholarship0

Re: rstudio::conf(2019) diversity scholarships

Hadley Wickham — Thu, 16 Aug 2018 17:26:28 -0000

Yes

Re: Help Needed: Use Sentence Styles in knitr's Documentation

Hadley Wickham — Sun, 31 Dec 2017 09:59:36 -0000

You might want to label the issue with "good first issue" or "help wanted" ( https://help.github.com/art... )

Re: Don't teach students the hard way first

Hadley Wickham — Sun, 24 Sep 2017 09:04:32 -0000

The thing I don't understand about your argument is that the tidyverse tends to be more consistent than base R. Take one of the motivations for tidy evaluation: unquoting.

There are three very commonly used base R functions that used NSE: $, rm, and library(). They each use a different technique to "unquote" (or switch to standard evaluation). There will be (once the conversion of ggplot2 is complete) only one way to unquote in the tidyverse.

No need to talk about quosures if you don't want to. And do(tidy(...)) is an older idiom that we've mostly deprecated in favour of new tooling around list columns and lapply/map.

Re: Don't teach students the hard way first

Hadley Wickham — Sun, 24 Sep 2017 09:01:29 -0000

THERE IS A SCIENTIFIC DISCIPLINE WHERE DATA COME PRETTY CLEAN?!!?

Re: Don't teach students the hard way first

Hadley Wickham — Sun, 24 Sep 2017 09:00:03 -0000

I take your point, and more data on which pedagogical techniques would of course be better, but it's not clear to me where the burden of proof lies. Should I have to prove that a collection of packages designed to work together is more effective to teach with? Or should you need to prove that a collection of functions that evolved over 10+ years is more effective?

Re: Teach the tidyverse to beginners

Hadley Wickham — Wed, 12 Jul 2017 11:58:01 -0000

I'd argue it's actually easier in the tidyverse, because (e.g.) tibbles print the class of each column so you can easily tell if you have a character vector or a factor.

Re: Teach the tidyverse to beginners

Hadley Wickham — Thu, 06 Jul 2017 13:21:02 -0000

Consistency of argument names and order is just one small part of the problem. I'm sure you can think of other areas where base R is inconsistent.

Re: Teach the tidyverse to beginners

Hadley Wickham — Wed, 05 Jul 2017 17:59:00 -0000

No one teaches the tidyverse without also (eventually) teaching functions, basic data types, control flow etc.

Re: Teach the tidyverse to beginners

Hadley Wickham — Wed, 05 Jul 2017 17:37:45 -0000

I think the best place to systematically introduce alternatives is in a second semester long course. I think a semester is just enough time to get students from knowing nothing about R to being competent data scientists, and it's better to go deep, not wide, in a first course.

Students will also start exploring on their own well before that. There are plenty of rich resources on the internet for them to supplement what they learn in class.

Re: Teach the tidyverse to beginners

Hadley Wickham — Wed, 05 Jul 2017 16:51:12 -0000

Yes, it's easy to miss how these small frictions add up to large pain. And once you've internalised the inconsistencies, it's hard to see why you'd bother with the tidyverse.

Re: Teach the tidyverse to beginners

Hadley Wickham — Wed, 05 Jul 2017 16:36:37 -0000

You don't need to memorise them, but if you don't you need to constantly look them up, which substantially lowers the fluidity of your analyses. I think if you're inside the tidyverse the chances of you correctly guessing the argument order or argument names is much higher.

I don't think you can argue that the inconsistency of base R makes it easier to learn!

Re: Teach the tidyverse to beginners

Hadley Wickham — Wed, 05 Jul 2017 16:31:48 -0000

You can now express that sort of thing fairly elegantly with dplyr:

var1 <- sym("pounds")
var2 <- sym("wt")

mtcars <- mutate(mtcars, !!var1 := !!var / 1000)

It's not quite as simple because it generalises to cover more cases.

Re: Teach the tidyverse to beginners

Hadley Wickham — Wed, 05 Jul 2017 16:30:12 -0000

I don't think it's fair to claim that there's "one syntax" for base R. For pretty much every syntax that the tidyverse uses, you can find some function in base R that uses a similar technique.

Re: Teach the tidyverse to beginners

Hadley Wickham — Wed, 05 Jul 2017 16:29:26 -0000

Have you used tidy eval? It's much better!

Re: Teach the tidyverse to beginners

Hadley Wickham — Wed, 05 Jul 2017 16:29:12 -0000

I can't imagine covering the full breadth of R in just one course. Maybe in 5 or 6! There's just so much to learn: apart from modelling (which typically needs several courses), there's a lot to cover about R as a programming language.

Re: Teach the tidyverse to beginners

Hadley Wickham — Wed, 05 Jul 2017 16:27:45 -0000

Base R is hard to learn because there are a lot of inconsistencies that you need to memorise to use it fluidly. For example, take the functions that provide the regular expression toolkit in base R: grepl, greprexp, gsub, and strsplit. Firstly, (apart from strsplit) the names are not evocative: who is Greg and what does he have to do with strings? Secondly, the order of the arguments is different: strsplit takes the character vector as the first argument, but all the other functions take the regular expression as the first argument. This type of inconsistency slows down learning because you have to rote learn many idiosyncracies.

Re: Teach the tidyverse to beginners

Hadley Wickham — Wed, 05 Jul 2017 16:24:54 -0000

I also agree that it's really important to cover those topics. I just wouldn't cover them first, because you're better of providing some immediately useful skill on day one to motivate learning the details later on.

Re: Teach the tidyverse to beginners

Hadley Wickham — Wed, 05 Jul 2017 16:24:01 -0000

Why do you think the tidyverse limits imagination?

Re: Measuring Gobbledygook – data science ish

Hadley Wickham — Tue, 29 Nov 2016 12:00:07 -0000

geom_freqpoly() ?

Re: Rebuilding Map Example With Apply Functions

Hadley Wickham — Wed, 27 Jan 2016 18:24:27 -0000

http://blog.rstudio.org/201...