vincenzocoia
commented
4 years ago This is looking great.
Some notes about setting up the worksheet to be structured for the autograder.
- nbgrader has its limitations, in that we can only test objects that the students create. So, this means that we have no way of knowing whether or not they've typed
3 + 6into the code chunk. - Maybe the
3 + 6question was not intended to be autograded. But either way, I think it would be useful to label each task with a question number, so that it breaks the worksheet into bite sized pieces that students can easily count their successes and failures with. - Maybe it would be useful, at the very start of each worksheet, to indicate how they will be graded on the worksheet. Something like "To get full marks for this participation worksheet, you must successfully answer at least 5 autograded questions."
- Answers need to be placed in between
### BEGIN SOLUTIONand### END SOLUTION-- when nbgrader creates the student-facing version of the worksheet, it detects these tags and removes everything between them. Anything outside of those tags will remain visible in the student-facing version. - A code chunk for testing should come immediately after the code chunk for the answer. The below example has the tests tucked away in a separate R script under the function
test_1.0(), but Team Sonja can put the tests using thetestthatpackage directly in that chunk.
Here is an example of a question from STAT 201, followed by a few comments:
**Question 1.0**
Which of the following is an example of an inferential question:
A. How many candy stores are there in Kitsilano?
B. Does the average sleep duration of individuals who consume melatonin 30 minutes before bed differ from those who do not?
C. If I study 28 hours, what score will I obtain on this exam?
D. Does owning a houseplant lead to the adoption of a puppy?
_Assign your answer to an object called `answer1.0`. Your answer should be a single character surrounded by quotes._
```{r}
# answer1.0 <- ...
### BEGIN SOLUTION
answer1.0 <- "B"
### END SOLUTION
```
```{r}
test_1.0()
```
Some additional comments:
I think it's important to credit others' work - but I wanted to consult you on how is best to do this.
Probably the best way to do this is to add a section near the top of the worksheet, entitled something like "Attribution", and essentially just put a description similar to what you describe in this pull request.
Going forward, I think that it may be easier if I learn how to knit the files to Jupyter notebooks and implement nbGrader
I don't think you need to do all of it. I would suggest seeing whether Team Sonja can help with writing the tests. We can provide guidance for what they should be testing, like "Object should equal 9", or "Object should contain a geom_point() layer, and have an aesthetic mapping of gdpPerCap to x".
iciarfernandez
Here is a draft of the first worksheet! I have yet to add the section that will correspond to Class 3: R Markdown & Reproducibility; so far I have only designed the first part corresponding to Class 1. I would like to acknowledge that although part of it is completely created from scratch, I have also taken some sections from the current guidebook & adapted them, as well as used the swirl() R Programming and the "A very (short) introduction to R" document by Claudia Brauer as inspiration (not copy-paste but did make some similar exercises adapting them to the workflow of the worksheet!). I hope this is okay, and I would also like to note it somewhere in the worksheet (maybe a "References" section at the end?) because I think it's important to credit others' work - but I wanted to consult you on how is best to do this. I look forward to hearing your thoughts!
Going forward, I think that it may be easier if I learn how to knit the files to Jupyter notebooks and implement nbGrader, I tried to make notes throughout for Team Sonja where the nbGrader sections would be, so hopefully that's enough but please feel free to reach out if anything needs clarification (and same goes for anyone working in Team Sonja). Thank you!