How to get target data for neural networking toolbox?

I am a student and new to the Neural Networking Toolbox on Matlab. I am building a road monitoring system which distinguishes between different road conditions, being speed bumps, potholes, roughness and good road. I have collected 3D accelerometer data (x,y,z) using my smartphone by driving over speedbumps, into potholes, on rough roads and good roads. I have calculated features, min, max ,mean, median, variance, skewness and kurtosis on each axis of the data. I now need to train a neural network to predict the given road condition but I am having some confusion with my input and target target. I think my input data would be the features calculated for each road condition but how do I get my target data? I know that the target that must consist of 1s and 0s to give the correct output. Can anyone give some advice?

Answer 1

[Greg Heath]

Use class indices 1:4 for the classes

 speedbumps, potholes, roughness and goodroad.

Then use the ind2vec function to obtain your target matrix

 target = ind2vec([ 4 1 3 2 2 1 3 4 ])

Then get the estimated classindices from new data

 output = net(newinput);

 estimatedclassindices = vec2in(output)

 help patternnet
 doc patternnet

Hope this helps.

Thank you for formally accepting my answer

Greg