GRNN-Based Data Prediction for Freight Volume Analysis
- Login to Download
- 1 Credits
Resource Overview
This case study utilizes data.mat containing p and t datasets with 13 samples each, representing freight volume and related variables from 1996-2008. The first 12 samples serve as training data while the final sample is used for prediction, implementing a GRNN neural network with MATLAB's newgrnn function and radial basis function computation.
Detailed Documentation
The data.mat file in this case contains two datasets (p and t), each comprising 13 samples representing freight volume and associated variables from 1996 to 2008. We implement a GRNN neural network where the first 12 samples of p and t serve as training data, while the final sample functions as prediction data. This approach enables accurate freight volume forecasting through MATLAB's newgrnn function implementation, which calculates smoothness parameters using cross-validation and processes inputs through radial basis functions. The network architecture automatically determines optimal spread parameters, providing probabilistic predictions suitable for transportation analysis and decision-making workflows.
- Login to Download
- 1 Credits