Feedback from days 4 through 8

As of April 29 2013

Copied from original feedback1 etherpad.

THINGS YOU LIKE BEST ABOUT THE COURSE SO FAR

It starts with raw data, and works through every step of preprocessing, leaving no detail uncovered. The instructors emphasize the logic of each step, something often overlooked by people who use “point and click,” black-box software packages. +3

I’ve heard Matthew lecture on fMRI analysis several times in the past, and I have to say, he has become a master at conveying his very important messages clearly and concisely, to beginners and experts alike. It’s a privelege to have him around. Of course, JB and Paul are outstanding as well. +2

The homeworks are really well-designed. They’re quick and easy to complete, but illuminate issues that would otherwise be easy to miss. +1

I liked the in-class assignments (the small functions we were asked to complete during class). We could have one or two of these in every class.+1

The things we learn are cumulative. I like that we cover topics once and then cover them again in later stages or later classes. Seeing things more than once really helps to gain a grasp of how they work, and lets them become integrated with the new material. +1

THINGS YOU LIKE LEAST ABOUT THE COURSE SO FAR

The focus on minute details – a blessing of the course – can at times become its curse, when lengthy, confusing discussions about obscure coding or math issues seem to detract from the big picture. Getting to the bottom of those details is without doubt important, but our time is limited, and attention could be better allotted to give priority to the most important things. +2

– I find that I often get lost on these details as well, and most go over my head... as a beginner (with a very basic understanding of Python, that is), I feel happy when I understand about 50% of what we’ve covered. I may be alone in this... +1

It may be useful to assign some reading before class introducing the issues to be dealt with next. For instance, the MRC webpages put together by Matthew and others are a good intro and may help to explain the problem we are trying to solve for anyone who doesn’t have much fMRI experience. Then, before going through the code, we could go over how we will approach the problem conceptually (or go over the statistics before they are implemented in the code). This way, even if people get lost in the code, they will be able to take away a big picture lesson on how to deal with common imaging issues in the future. +1

STUFF YOU’D LIKE TO HEAR MORE ABOUT

I don’t know where we’re headed, but it’s really important to me that, by the end of this course, I know how to write all my own code to move from raw data to completed analysis. As long as the course provides pre-written notebooks to get from the beginning to the end, I should be OK – whatever I don’t understand by the end of the course I’ll review after it’s over. It would be great if the instructors could point us to good reading material about the issues they aren’t able to delve deep into. +3

I like it when we approach a problem as a class, rather than just reading code that has already been written (though that is definitely useful as well). +1

It would be nice to achieve a couple “big picture” items per class in order to keep the necessary attention to minute detail in context.+1

It would be helpful if any variables defined in the notebook were explained explicitly. For the most part this is done well, but one notebook from the 19th was rather confusing.

I’d like to learn a bit more about the pros/cons of various pre-processing steps. For instance, how can we test the best parameters to coregister the images we may have?

ANY OTHER COMMENTS?

Great job, guys! +1

I hope this course becomes a regular offering. It really is valuable, particularly learning how to break out of the black box and work with data directly.

Please offer this course again for those of us who could not make it or would like to review the material a second time. It also might be good to offer a separate intro python lessons beforehand (and have pre-requisite intro python homework) so that those unfamilar can learn the basics and those familar don’t waste their time.

I think this course might be better divided into two distinct sections: 1) A Python basics section that helps everyone feel competent with running basic Python code and using the notebook 2) A fMRI analysis section that gets into the higher level topics. I think this format could help with attrition.+2

I’m really sorry I’ve dropped off - I enjoy the learning the nuts and bolts, but have had conflicts the last several weeks... if I’m too far behind I’d love the chance to try again another semester!