search

UBC-MDS / DSCI522_309

Workflows Analysis Project for DSCI 522
Other
1 stars 3 forks source link

Analysis plan needs clarification #28

Open ksedivyhaley opened 4 years ago

ksedivyhaley commented 4 years ago

I'm finding your analysis plan difficult to follow.

You spend several bullet points discussing preprocessing, but then end on the very vague "Decide a methodology for outlier treatment & feature selection" - milestone 1 was intended to give you the time to decide on your methodology so you can focus on implementing it for milestone 2. You then skip to a specific list of outputs with no explanation of why these are appropriate.

I've asked Firas and Tiffany to have a look as I think they may be able to provide more specific methodological feedback - I want to make sure you're using an appropriate analysis for Milestone 2.

ksedivyhaley commented 4 years ago

Related issue:

Proposal could use more detail on the outputs and what they will illustrate about your research question (as mentioned in lab, don't assume the reader has taken the same courses as you - even if your audience are data scientists there's some variation in background).

I wouldn't worry too much about fixing your proposal at this point, but make sure it's clearly conveyed in your final report!

ksedivyhaley commented 4 years ago

(One final thing to keep in mind re: proposal and reports - please use minimal point form as described in the "Project organization and documentation expectations"!)

ksedivyhaley commented 4 years ago

Firas has indicated that the parts of your analysis plan that you do mention should have included more detail to indicate not just what you are doing generally, but how. For example, how will you standardize or normalize, how will you choose how to merge the categories, and how will you tackle class imbalance?