How to store time series data for use in patternnet?

Hi, I am trying to find the best way to store my data for use with patternnet in the Neural Network Toolbox (if this is even possible). I am using 300 data sets of 39 points in time. These 300 data sets are then evenly split into 3 categories (data with increasing trend, data with decreasing trend, stable data). My objective is to train the neural network to be able to classify new data into one of the three categories. So far, I have one 1x39 2-D numerical array of time points. Then, I have three 100x39 2-D numerical arrays of frequencies that correspond to each time point as a sequence. What is the best way to store this data so it can be run through patternnet? I can't seem to find any examples of this.

Answer 1

[Greg Heath]

A neural net is not necessary:

Typically, linear trends are removed before applying the FFT.

Therefore, your question can be answered before applying a FFT.

Hope this helps.

*Thank you for formally accepting my answer^

Greg

[ampaul]

Greg,

Isn't important that linear trends remain in my data as my classifications are based around them?

Thanks