From: Forecasting worldwide empty container availability with machine learning techniques
Hyperparameter | Destination prediction modell | Duration prediction modell |
---|---|---|
Number of layers | 3 | 3 |
Number of neurons | 750 | 450 |
Epochs | 10 | 30 |
Batch size | 1028 | 1024 |
Activation function | ReLu | ReLu |
Activation function (last layer) | Softmax | Softmax and NNelu |
Loss function | Categorical cross-entropy | Negative log-likelihood |
Optimizer | Adam | Adam |
Initialization | Xavier | Xavier |
Mixture model | – | 3 distributions (2 × moyal distribution, 1 × normal distribution) |