Project-specific Mathworks repository
"MATLAB for Data Processing and Visualization"
"Machine Learning with MATLAB"
In this work, following a systematic approach, a Nonlinear-Autoregressive (NARX) type model was developed in order to predict the behaviour of a vehicle in traffic, with the following performance in Test: R^2 0.90 MSE 9.47 e-3.
A method for the application of the obtained model has also been proposed in the report.
• "Traffic Data Analysis for Modelling and Prediction of Traffic Scenario.pdf"
This is an extended description of the work performed and the results obtained.
• "Code_NARXnet.m" This is the MATLAB file where it is possible to inspect what has been done.
• "Live_Code_NARXnet.mlx" This is the MATLAB Live Script file where it is possible to run a new training session.
• "Live_NN.mlx" This is the MATLAB Live Script file where it is possible to use the pre-trained network "NARXnet_TRAINED.mat".
• "NARXnet_TRAINED.mat" This is the TRAINED NET.
• "NN-IN.xlsx" This is the Inputs set used to train and test the net.
• "NN-TARG.csv" This is the Target set used to train and test the net.
• "X-Input.xlsx" This is the Inputs set used in "Live_NN.mlx".
• "Y-FeedBack.csv" This is the FeedBack set used in "Live_NN.mlx".
• The file "Untitled.slx" and "MyNeuralNetworkFunction.m" are the Simulink block and MATLAB code of the NARX network respectively.
If you want to use the pre-trained network
- Download the files: "X-Input.xlsx", "Y-FeedBack.csv", "NARXnet_TRAINED.mat", "Live_NN.mlx".
- Save the files in the same folder.
- Open "Live_NN.mlx" with MATLAB ( with the " Deep Learning Toolbox" installed ).
- Run and get the predictions!
If you want to train your NARX network with the datasets used in this work
- Download the files: "NN-IN.xlsx", "NN-TARG.csv", "Code_NARXnet.mlx".
- Save the files in the same folder.
- Open "Live_Code_NARXnet.mlx" with MATLAB ( with the " Deep Learning Toolbox" installed ).
- Run and train the network!