Five models are currently being evaluated. But the neural nets was left out due to performance issues -- it takes too long to go on production for short forecasting horizons (i.e. 30 ~ 120 minutes). However, it would be interesting to see how it compares in accuracy overall.
Five models are currently being evaluated. But the neural nets was left out due to performance issues -- it takes too long to go on production for short forecasting horizons (i.e. 30 ~ 120 minutes). However, it would be interesting to see how it compares in accuracy overall.