| title |
|---|
What could I improve next time? |
Some way to summarize the "best practices" or top thoughts at the end of the each class was appreciated (we sometimes did this through the whiteboard summaries, etc). (+2)
Sometimes the classes felt like they ended on the note of... well, things are complex! Without necessarily a sense of how to move forward on it (if we were embedded within an organization, etc). (+2)
I think that the range of topics covered provided a pretty good overview of data science, what it is, the roles of data scientists in various industries, typical work responsibilities and workflows, and what tools exist. (+1)
It would be nice to get Hadley's thoughts on each of the sets of readings for maybe 10 - 15 minutes sometime in class. (+3)
Definitely more experts. I wish there was more insight from those who understand and know the field well (Hadley, Eduardo, Etc) (+2)
I really liked the presentation by Ed from Facebook (+9), and it would've been cool to have other presenters or maybe other speaking events outside of class if we want to spend more time actually discussing the readings in class. (or to be able to ask the paper / blog authors questions)
- One thing that was great about Ed's presentation was that he gave a lot of concrete examples (e.g. Examples of specific questions he asked when interviewing at Facebook; Concrete example of what a career path at Facebook looks like)
One possibility is to have students bring in a relevant article that they liked and share it with the class.
Hearing more expert opinions would be great! Whether it's Hadley speaking on his thoughts, bringing in more industry speakers (the FB lecture was great!), or having Stanford faculty speak. (+1 having current faculty who are thinking about these topics would've been great to have in class to hear opinions from)
Less google docs.
The DevOps and software engineering sessions had a lot of overlapping content.
- Also felt like content in many weeks was overlapping
I would have liked a working definition for "data science" at the beginning of the quarter, so that even if we disagree on what it is, we can be referring to the same thing throughout the quarter.
- Might have been interesting to get our naive definitions during class one; and compare that to a definition at the end of the class
I would appreciate if we could spend more time on data structure, such as relational databases. (+1 seems data engineering practices are important to know)
Case studies of really well-done and poorly done analysis might have been nice rather than high-level overviews of process suggestions. (+4)
Did we discuss everything in the course overview? I feel like we didn't spend any time on visualization for explanation and exploration, which would have been great to cover with Hadley.
I think a group project would have been fun.
I found the various activities to be very useful for idea generation, but they were less effective at distilling and refining ideas. It would be useful to have a few activities where the final product is some nice, compact artifact. (+4)
- I actually wouldn't want this. I don't know if it's useful to create a compact artifact for an entire class of people, and it seems like it would take a lot of time to distill the thoughts of 20 different people into a single thing. Maybe instead of having us write the responses before each class, we could write them after, so that we could distill our own ideas after the discussion.
I would have liked more concrete discussions. Perhaps the in-class prompts could be made more specific.
I liked some of the interactive class discussions, but would have liked more of an "intro" and "summary" from Hadley (+3). I still feel that I got the range of opinions and thoughts from people not at the center of some of the debates. In this context, I think that the learning curve plateaued for me and I got more out of the readings than the class itself.
I found the response papers to be quite helpful as a part of the organization of the course, and was happy with them being quite nebulous, and not really that highly criticized in terms of assessment. They helped me synthesize my thoughts in ways that I would not have done if I were only reading the papers. (+4)
I've saved or bookmarked some of the readings for future reference as I thought that some were very good at discussing and summarizing a particular aspect.
It might have been useful to turn the weekly responses in earlier, so you could tailor discussion based on students' interest. Maybe just have a discussion board (+1) on which we all respond? Then we may have avoided some of the less fruitful discussions.
For the response papers, perhaps the default guideline should be to pick just one article and discuss it, rather than going through all the articles.
More clear feedback on responses would have been helpful to know how to improve from a check to a check plus. But might suggest re-thinking the weekly response paper in favor of discussion board / question posting option.
- I preferred the response paper to an online discussion board. (+2)
It would be great if we could turn in the reflections online both for (a) greater ease of submission/assessment and (b) we could view each other's reviews. (+3) (+3 mainly printing was such a pain)
Would much rather post questions / comments to a group discussion board or Canvas than write response papers. (+2)
A concrete question prompt rather than an open-ended reflection would have been nice). (I actually liked the open-ended reflection. I may have enjoyed the responses less if I couldn't focus on a tangential topic of personal interest)
Annotated bibliography seems rushed? Maybe if we did it in two stages (develop idea, preliminary reading) earlier in the quarter, then wrote up a couple pages on it?
Drop the annotated bibliography or having a clear purpose for that. (+1 yes - I would have liked some feedback from other people about what kinds of questions would be of greater interest to not just me; want to make it useful for more than just a weird curiosity)
An option would be to have an opinion of a data analysis made by us in the past.
Working with a few examples of real code would have been interesting. Perhaps a peer code review or a presentation of some good code by Hadley. (+5) Code from Hadley!
One of my favourite moments of the whole class was early on when we traded our own practices for workflow, reading, etc. (+1) More of that would be been great (i.e. process things that are super helpful but rarely come up in class or with other students)
Perhaps some of the tools and workflow discussions could have been illustrated by talking through some real-life examples of projects that might come up for the people in the room. For example: "Imagine you're tasked with doing XYZ for a company that does ABC, how would you go about doing that?" (+1 a few more concrete discussions!)
Any thoughts on maintaining some sort of post-class community? (e.g. e-mail list, etc)?
It would have been cool to get ".rprofiles" of some of the data scientists Hadley knows, with more nuggets of personal experience mixed in. (+2)
A clear goal/expected learning outcomes for the course from the start.
Emphasise ambiguity and why I'm not talking
More history/context? Data science is very new - conventions less established.
Start with Tukey's data science + statistics.
More concrete data science process --- case studies + examples.
More sharing from discipline on experience.
Annotated bibliography should be multiple assessment.
Online discussion so time in class for other activities?