Hi Seyed Aref,
It is my understanding that the Mu value in a fitnet regression model is related to the training performance of the neural network. A decreasing Mu value means that the network is slowly converging towards the optimal solution, and a value that is far from the target indicates that the network has not yet achieved the desired level of accuracy.
While a high R and a low MSE may indicate good prediction performance, it is important to note that the training performance of the network affects its ability to generalize to new data. Therefore, it is important to monitor the Mu value and ensure that it is close to the target value, especially if you plan to deploy the model in real-world applications.
The following are the suggestions that you can follow to improve the performance of the model,
- Increasing the number of training iterations can help the network to converge towards the optimal solution. You can try increasing the maximum number of iterations in the fitnet options.
- The learning rate determines the step size of the weight update during training. A high learning rate can cause the network to overshoot the optimal solution, while a low learning rate can cause slow convergence. You can try adjusting the learning rate using the fitnet options.
- It is possible that the current network architecture is not suitable for the given problem. You can try experimenting with different network architectures, such as changing the number of hidden layers or the number of nodes per layer.
