nicoschuster01
commented
6 months ago Avi and I discussed our options and we decided that we'll do the dashboard as it seems like the best idea. Thank you for the feedback! The proposal will be adjusted accordingly.
Hi @nicoschuster01 @acg425
The writing of the proposal is clear and crisp and formatted (oddly?) well. It's legitimately quite clean. You should make all your readmes this 🔥
I know you're referring to the PaySim dataset, one for which a million public notebooks and code exists. I assume this is where you became versed in many of the terms in your write up, as we haven't used all these terms or used slightly different ones. (Or from another class?)
So, if you continue down the path of "estimate a few models, put into a (nice) ipynb file, and make that a website", that's a thing you can do very nicely, but your grade will have an upper limit because it's not project-level ambitious and I will have a hard time knowing if your work is original or comes from one of the many notebooks on this.
My suggestion: You make a streamlit dashboard. The dashboard will take as inputs possible choices users might make in estimating a model on this dataset, and show the performance of that set of choices, along with some diagnostic outputs. A second page on the streamlit dashboard can show a "menu" of choices and how different combinations of choices leads to different performances.
This dashboard idea, if executed well, can get a high score, and could be very cool. It's also VERY doable! To pull it off, start by making a streamlit repo (mine) but have the app.py file just open the data and do some basic output (a table or a plot). This would be just to get you started. And then go from there.
If you don't take the dashboard route, your report needs to do some extra stuff of some kind (I'll leave the possibilities open for you to imagine and choose).
Please let me know below what you choose, and then update the proposal as necessary.
Cheers, Don