Average of Polls pulled from RCP but displayed with error The Shiny App Version is Live Here.. For the time being, this project has merged with the "Back of the Envelope" project for point-and-click regression analysis. The two projects may un-merge at a future time.
Eventual goals:
- use mixed-effects to correct for "House Effect" on polling error.
- or use GAM with a covariate of the house effect for each poll result
linear distance calculating module for "Whom should I vote for?"redraw compass map to place user and create a voronoi diagram to go with it?
TODO:
- General Election map (Polling)
Note that excel doc is structured to be copy/pasted into a jupyter notebook, not actually used for analysis directly becuase no IDE is available on a work computer.
Here, the House Effect of each firm is taken from fivethirtyeight's dataset (see their github repo here ). For Harris polling, the house effect is for both HarrisX and for Harvard Harris until I can figure out a discernment between the two. This is not meant to be a perfect statistical adjustment, just some additional variance to control for with the hope of quieting these effects. In this case, + is |+D| and - is |+R|.
