Table 2 Model characteristics.
From: Using machine learning to uncover the relation between age and life satisfaction
Basic characteristics | ||||
Type of neural net | Feedforward neural network | |||
Loss function | MSE | |||
Training characteristics | ||||
Batch size | 1 | |||
Number of epochs | 20 | |||
Optimization | Adam optimization | |||
Learning rate | 0.0001 | |||
Initial weight distribution | \({w}_{i}^{l}\sim N(0, \frac{1}{\sqrt{31}})\) | |||
Layer characteristics | ||||
Layer | Input (1) | Hidden (2) | Hidden (3) | Output (4) |
|---|---|---|---|---|
Features/nodes | 31 (Including 16 cohort dummies and age) | 31 | 31 | 1 |
Activation | Input | Leaky-ReLU | Leaky-ReLU | ReLU |
Dropout rate | 0 | 0 | 0 | 0 |
Performance | ||||
Training error | 1.80 | |||
Test error | 1.95 | |||
