How do I arrange and input my data for a time series problem?

Hi,

I have data for 62 households. I have 9 parameters (eg. floor area, number of occupants, income etc) and the dependent variable is energy consumption (kWh). I know energy consumption of the houses every hour (for a 24 hour period) and want to build a neural network model to predict energy consumption over a 24 hour period for other houses based on parameter inputs (test set).

Do I suppose: inputs = 9 x 62 matrix (ie. 9 variables and 62 cases) target = 24 x 62 matrix (ie. the energy consumption every 24 hours for the 62 houses observed)

What neural network is better suited - time series or fitting a function (or other)?

If anyone can help a MATLAB virgin I would be grateful!

Thanks in advance.

Regards, David

Answer 1

[Greg Heath]

This is a fitting problem.

 help fitnet

 doc fitnet

 Search

 fitnet

 greg fitnet {;>)

Hope this helps