Model v2 results #5

fallonfarmer · 2018-03-15T09:53:27Z

Run new_model file to get second version of predictions
- Cross validation accuracy = 0.75
Update join_outputs_to_slots so it can be run inline / notebook
Add new readable slots results for model v2 predictions
- Villanova is still overall
- More upsets than before
  - biggest changes: Duke over Kansas, Xavier over Gonzaga
- Tends to bring the probability differentials closer together (more close calls)

so it can easily be modified to create different versions

fallonfarmer added 7 commits March 15, 2018 05:24

Clean up file

8bdbe84

Move file vars up top

e69621d

so it can easily be modified to create different versions

Fix to run new model locally

f1ad935

Model v2 prediction outputs

6539d6e

Remove init so it's easier to run in kernel

e6227d2

Model 2 slots results

7e9b5ea

Fix it so it doesn't overwrite previous

a627783

fallonfarmer requested a review from tyler-ranalli March 15, 2018 09:53

fallonfarmer merged commit 899bf9f into master Mar 15, 2018

fallonfarmer deleted the run-model2-results branch March 15, 2018 09:54

Provide feedback