A question for a NN project

Submitted by jacob on Tue, 01/05/2010 - 12:31

Getting Started

Hi,

I'm trying to make a Neural network project. My project is MPG(MilePerGalon)
Auto.

The data concerns city-cycle fuel consumption in miles per gallon, to be predicted in terms of 3 multivalued discrete and 5 continuous attributes.

1. mpg: continuous
2. cylinders: multi-valued discrete
3. displacement: continuous
4. horsepower: continuous
5. weight: continuous
6. acceleration: continuous
7. model year: multi-valued discrete
8. origin: multi-valued discrete

An data example:
-18.0 8. 307.0 130.0 3504. 12.0 70. 1.;
-15.0 8. 350.0 165.0 3693. 11.5 70. 1.;
-18.0 8. 318.0 150.0 3436. 11.0 70. 1.;
-16.0 8. 304.0 150.0 3433. 12.0 70. 1.;

So how which algorthm should I use?

Submitted by jeffheaton on Tue, 01/05/2010 - 20:29.

This is not a classification problem, so I would not use a SOM. Perhaps an RBF. Its not temporal, so I don't think you need Elman or Jordan style networks. I would just start with a single layer feedforward network. First task is to get your input and output neurons. The output neuron will be miles per gallon. I would probably just divide by 100 to keep it in the 0-1 range. For example 30mpg would be .3, 18 miles per gallon would be .18. You could probably get a bit more precise to use more of the range, since I doubt you will have many cars with greater than 50 mpg. The Encog normalization class can actually analyze the input data and deduce the range.

I would assign a single input neuron to each of your continuous inputs.

The MVD inputs will need one input neuron for each of the discrete values.

Then create a training set with the input (your values above) and ideal (the mpg) and train it.

Jeff

Submitted by jacob on Thu, 01/07/2010 - 11:50.

Exception in thread "main" org.encog.neural.NeuralNetworkError: Error, must define radial functions for each neuron
at org.encog.neural.networks.layers.RadialBasisFunctionLayer.compute(Unknown Source)
at org.encog.neural.networks.logic.FeedforwardLogic.compute(Unknown Source)
at org.encog.neural.networks.logic.FeedforwardLogic.compute(Unknown Source)
at org.encog.neural.networks.BasicNetwork.compute(Unknown Source)
at org.encog.neural.networks.training.propagation.PropagationUtil.forwardPass(Unknown Source)
at org.encog.neural.networks.training.propagation.Propagation.iteration(Unknown Source)
at org.encog.examples.neural.xorbackprop.BENIMnnPROJEM.main(BENIMnnPROJEM.java:945)

How should i define radial functions?