Notebook: KNN regression

memphis-iis / datawhys-content-notebooks-python

Content for DataWhys in the form of JupyterLab notebooks (.ipynb files)

Apache License 2.0

8 stars 2 forks source link

Notebook: KNN regression #7

Closed aolney closed 4 years ago

aolney commented 4 years ago

See the spreadsheet for details

Content	Programming
DB	TS

Ideas/prereqs: Regression vs classification, regression and prediction, causality and experiments, Functions, estimating functions, supervised vs unsupervised,

Direct link https://jupyter.olney.ai/hub/user-redirect/git-pull?repo=https%3A%2F%2Fgithub.com%2Fmemphis-iis%2Fdatawhys-content-notebooks&subPath=KNN-regression.ipynb&app=lab

aolney commented 4 years ago

v1 with https://github.com/memphis-iis/datawhys-content-notebooks/pull/39

Added pipeline and scaling
Cut loop plots put kept loop rmse for optimal k
Loop plots are beautiful and should be brought in again later

aolney commented 4 years ago

v2 with https://github.com/memphis-iis/datawhys-content-notebooks/pull/40

Partly put the loop plots back in, with the following differences
- Not in loop
- Removed need to sort dataframe by using markers

So this shows them the difference b/w observed and predicted using different color dots.

This gives them exposure to putting multiple layers on the same plot, which I think is really important and achievable for them at this stage :smile_cat:

aolney commented 4 years ago

PS version with https://github.com/memphis-iis/datawhys-content-notebooks/pull/41

Plots for increasing K back in, though without some of the pretty axis labels
Changed mpg (mtcars) so ? is nan
Added mpg-nona with all nan rows removed to avoid complexity

aolney commented 4 years ago

closing for now to clear the board; will reopen for discussion at future date