search

Aakash282 / 1ia

data driven sports intelligence platform
0 stars 0 forks source link

2006-2008? Data has problems #9

Closed edwardgarza closed 9 years ago

edwardgarza commented 9 years ago

When using these years as the training set, we get strange predictions

Aakash282 commented 9 years ago

Also, when we use these years as the test set, we get predictions that are large magnitude: like 4289757 as a score difference. Definitely need to investigate this further.

Aakash282 commented 9 years ago

@edwardgarza @2017ssahu I know you guys were looking into this - was there any type of final word on this?

Aakash282 commented 9 years ago

@edwardgarza so i guess something was just weird in the code. we never figured out what was going on. idk what happened lol

2017ssahu commented 9 years ago

@Aakash282 does it still not work for these years?

Aakash282 commented 9 years ago

i think we can close the issue for now. we aren't seeing any weird behavior. it just as easily could have been strange behavior from our models.