Using a CNN trained on a sports betting dataset, I tried…
0 reactions
Using a CNN trained on a sports betting dataset, I tried my hand at predicting who will win the Super Bowl.
The CNN (Multi-modal, time series) trains using data from the NFLVerse github.
Code pulls player-specific data points from the dataset as seen in the chart.
It then groups the data by team/season/week.
A python script then lets the user define how many weeks to train from and what teams to use (you could use the entire NFL, or just two teams)
The CNN is trained based on these parameters and runs several epochs to ID what the most significant/weighted values are that have causal relevance to winning.
#education
source
The CNN (Multi-modal, time series) trains using data from the NFLVerse github.
Code pulls player-specific data points from the dataset as seen in the chart.
It then groups the data by team/season/week.
A python script then lets the user define how many weeks to train from and what teams to use (you could use the entire NFL, or just two teams)
The CNN is trained based on these parameters and runs several epochs to ID what the most significant/weighted values are that have causal relevance to winning.
#education
source
| 4,000 mlx | |
| 4 | |
| 0 |