Maybe do a few more videos on some other classifiers + provide some short example code. Make optional.
Student comments:
More in-class activities while fitting the classifiers would have been helpful rather than one at the end.
Introduce more algorithm types
Introduce a little to the math behind each algorithm (consider but if we do this, make it optional)
It would have been interesting to learn about some of the other ML classifiers that the labs of the instructors and TAs use but not necessarily implement them with this dataset.
maybe directing towards other types of common classifiers--a random tree was the most intuitive predictor for me and that's why I went with it
Have a guide on the prelab or the homework on all the possible approaches we could have used (ie a summary of all the machine learning algorithms we've learned for the past five sessions)
I think some exploration of the classifier parameters beyond C or feature length/depth would be nice, though there may not be time. I do know that a diagram of how decision tree models would also be useful; I had to explain to a few of my table-mates what we were manipulating when we set max_features and max_depth.
I would like some more background information on how exactly some of the classifiers break down the data into categories and what some of the more frequently-tweaked parameters are, from the instructors specifically. I would also like to be able to plot the data and my models in some way for a visual representation of what exactly I was doing as I changed parameters to create the model. I had trouble understanding some of the documentation that I was looking up for SVMs et cetera, since I feel that coding is a bit of a language barrier. I'm new to python, and many of the support documents and forums use considerable amounts of jargon with which I was unfamiliar. I think spending a fairer bit of time working on this than just over the weekend would have made for a better shot at getting a good model. Perhaps doing this over two class periods would have given us a better shot at succeeding. Also I probably should have gone to office hours for more support.
Maybe some hints as to certain classifiers we should try
I would have liked to spend more time on types of algorithms
Maybe for the ML modules in general - more in depth description on the different type of classifiers and how to adjust those parameters (ie what each parameters mean and how to adjust accordingly)
The exercise was pretty good, I would have enjoyed a video going into the different classifier algorithms.
Maybe do a few more videos on some other classifiers + provide some short example code. Make optional.
Student comments: