Hello everyone,
This is my very first post and I am so glad there is a topic for time series data.
I wanted to build a predictive model for Football (Soccer) data. I first started by looking at Premier League data from ‘https://datahub.io/sports-data/english-premier-league/datapackage.json’ .
Then I added the ‘add_datepart’ columns and replaced the categorical targets to integers.
Followed by a walk forward train, valid and test split as it is Time series.
Got this as a result:
Beyond this point I am stuck with how to use the data block api to build my data bunch. As well as, chosing the right layer size and final activation function to narrow down my result to either a 1(Home Win),2(Away Win), 3 (Draw). I attempted to follow the rossmann procedure.
Any help or pointers would be very appreciated. I can most likely help you with another topic so I hope it is worth your time helping me. I am also available to communicate elsewhere like Skype.
Thank you,
Ethan