Unsupervised learning portion of the project

SBuxton-IDM / EquityMilestone

A repository for the MADS Milestone II project related to digital equity

0 stars 1 forks source link

Unsupervised learning portion of the project #3

Open SBuxton-IDM opened 9 months ago

SBuxton-IDM commented 9 months ago

This issue is meant to organize several different issues relating to the unsupervised learning portion of the project. As the project progresses I may add more issues/checkboxes

[x] Conduct PCA on existing dataset
[x] Compare to what we think are the 5 key features (Download mbps, latency, jitter, upload etc.)
[x] Create a visualization (like heat map or biplot) for the results
[x] Define a variable in the notebook that contains only the key features (which we will use to train our model)

SBuxton-IDM commented 9 months ago

Blocked by: https://github.com/SBuxton-IDM/EquityMilestone/issues/5

yolanda707 commented 9 months ago

PCA maybe not enough, I am not sure the relationship of data is linear or non-linear, I plan to use UMAP, PCA and t-SNE to check at the same time. But still waiting for the one complete dataset to work on, now we just have data in 4 different folders which is not working for the project.

yolanda707 commented 9 months ago

Have uploaded the UL procession file for checking the steps of unsupervised learning deduction dimension and k-means to get the clustering. Hope to have your review comments here.

SBuxton-IDM commented 9 months ago

Have uploaded the UL procession file for checking the steps of unsupervised learning deduction dimension and k-means to get the clustering. Hope to have your review comments here.

Looks great! I'm assuming the stuff after a certain line (mentioning food) is for another project, but this fulfills all the requirements detailed in the issue, I especially like the added 3d point plot. I will make a PR on Saturday to make this presentation-ready and add you as a reviewer (just getting rid of the extra cells and adding some markdown cells for legibility).