Adventures with R
The great Fela sang “Teacher no teach me nonsense” .
While taking a course in Machine Learning the other day, I came across this:
To find this kind of code on a course amazed even me, an R neophyte, as there is so much wrong with this function that is wrong and it really left me puzzled. What made it worse was that the course is being organised by a highly reputable software giant.
But please, if you’re loading (or more correctly, attaching) a package within a function, do not call
library(); rather make use of
require(). This blog by the author of the knitr package explains it better than I ever could.
View original post
teaching kids to code has had about three stages in the world:
- computer classes or labs in primary schools, in the 20th century
- computers for developing nations initiatives like one-laptop-per-child
- the contemporary rasp-pi/mobile app stage
each of these stages was driven by significant changes in hardware; the first by the advent of the home (and school) pc, the second by the development of a radically inexpensive “netbook,” and the third by both smartphones and single-board computers (sbcs).
i grew up in the first of these, and taught myself to code at home. the question “code what?” was not important then; asking it would be like the j.k. rowling interview where a reporter asked: “what should kids read while waiting for your next book to come out?” a stunned rowling replied, “other… books?”
with logo, the question is answered before you start: logo lends itself most easily to making graphical designs…
View original post 610 more words
Adventures with R
The good thing about sharing your knowledge with people is that you always find yourself strangely inspired. That’s what happened to me the other day in a meeting with colleagues where I were looking at some issues we were having with a reporting format we created in R Markdown. I was trying to explain to them in simple terms the little I knew about how R’s search path works and how it access objects along that path. I typed
search() and got this all too familiar output
Because of the size of my console window that day, I suddenly had an analogy for them.
I asked them to imagine that they were sitting at a desk with two drawers and suddenly started looking for a particular document. What would they do?
© mattbuck 2012
The first impulse would be to start from the tabletop first. This represents
View original post 70 more words
I found this easy-to-read blog on renaming git branches and just had to share…
The code may not render properly on this page, so I suggest you hop on over there and read it.
Multiple States Knowledge Base
If you have named a branch incorrectly AND pushed this to the remote repository follow these steps before any other developers get a chance to jump on you and give you shit for not correctly following naming conventions.
1. Rename your local branch.
If you are on the branch you want to rename:
If you are on a different branch:
2. Delete the old-name remote branch and push the new-name local branch.
3. Reset the upstream branch for the new-name local branch.
Switch to the branch and then:
View original post
In the introduction to his book R packages, Hadley Wickham provides a neat function for making sure that everything is set for writing your own R extensions, by simply running the
devtools::has_devel(), which, if all goes well, should evaluate to
This did not work out for me and I had to fix this problem on 2 different occasions so I felt I need to share this info in case there are others also stumped by this hurdle.
The fix I found – after a full sweaty day – was in this conversation on GitHub and I would like to break it down very quickly:
- Make sure you have installed Rtools from CRAN
- Make sure that Rtools/bin as well as Rtools/MinGW/x64/ are added to your system PATH (if you don’t know how, click here)
- In addition, it is recommended that you install LATEX (the link is also found on the Rtools page mentioned on No. 1)
- Run the following lines of code
install_github("hadley/devtools") # to get the latest 'pre-CRAN' package updates
has_devel() # output should be TRUE
Like I said, I had this problem on 2 different machines (Windows 7 and 10) and the same fix worked on both of them.