issues
search
6758-Project
/
hockey
0
stars
0
forks
source link
Feature Engineering I by *Nov 9 EOD* (10%)
#24
Closed
TimkLee
closed
2 years ago
TimkLee
commented
2 years ago
[x] Acquire all of the raw play-by-play data for the 2015/16 season all the way to the 2019/20 season (inclusive).
[x] Set aside all of the 2019/20 data as your final test set.
[x] Use the 2015/16 - 2018/19 regular season data to create the training and validation sets.
[x] Create a tidied dataset for each SHOT event
[x] Create figures (Shot counts histogram binned by distance, shot counts histogram binned by angle, 2D historgram distance vs angle)
[x] More figures (Goal rate vs distance, goal rate vs angle)
[x] Another histogram (goals only binned by distance, separate empty new and non-empty net events)
TimkLee
commented
2 years ago
Notes:
Until Part 7, any reference to the “dataset” will exclusively refer to the 2015/16 - 2018/19 data.
Can approximate the net as a single point
Goal rate, i.e. #goals / (#no_goals + #goals)
Rare event, higher possibility of incorrectly labelled.