Only update on this model:
Sorry for lack of pictures.
It is now pointing towards a R+4% win.
Overall I think the model was too sensitive to the early results (which were early votes, skewed democrat)
For those interested in the #TN7 special election and the results of it, I have built a model which will (hopefully) accurately predict the final results as they come.
To start, my base prediction for the results is approximately:
🔴 Matt Van Epps : 51% (+4%)
🔵Aftyn Behn : 47%
1
96
Early on, the model thought Behn could win given early results from Montgomery. When she began losing that county, the prediction began to shift more towards +4 R.
1
48
For those interested in the #TN7 special election and the results of it, I have built a model which will (hopefully) accurately predict the final results as they come.
To start, my base prediction for the results is approximately:
🔴 Matt Van Epps : 51% (+4%)
🔵Aftyn Behn : 47%
However, I have done the occasional test using "biased" multivariate hypergeometric (hence, Republican or Democrat votes are more likely to be counted first). It does skew the model, but only if it is extreme. It will mostly come down to if mail-ins/early vote/e-vote affect this.
1
58
If you have any questions about this, please ask! I am still working on making this thing work, and whether the model succeeds or fails, it'll help me build better ones for future elections.
If you're interested, follow. I will be updating the model throughout tomorrow. #TN7
80
Being pulled into a spot to get 2 sausage mcmuffins at McDonalds is embarrassing.
Have drive times gone too far?
1
1
79
Do you remember when you joined Twitter? I do! #MyTwitterAnniversary
2
68












