How can I create a data set with multiple random plots?

I am trying to develop a random time series data set to use in training a neural network. I would like to mention that I am very new to Matlab.

This is the equation I'm working with: test = µ + r(t)σ + gt

where µ (mean value) = 3 t (time in seconds) = 1:1:20 σ (noise level) = 5 g (magnitude of gradient trend) = 2 r = normally distributed random number

This equation should show an upward trend in the data, correct? I would like for this equation to plot multiple y values over my x values (time).

When I plot this data using r=rand, it just plots a linear graph.

Can you help me find out what I'm doing wrong? Thanks

[Greg Heath]

In general, whenever you ask a question:

1. Please post the exact executable code, not just your interpretation.

2. If applicable, use the code in the help and doc documentation. For example, see

 help timedelaynet

and/or

 doc timedelaynet

3. If additional data is needed, use MATLAB data obtained from

 help nndatasets

 and/or

 doc nndatasets

4. This will save a lot of time and misunderstanding.

Hope this is helpful.

Greg

Answer 1

[Ali Ridha Ali]

mu=3; % mean value
s=1; % stepsize >> decrease it to get more points
P=1:s:20; % time period
sigma=5; % noise level
g=2; % magnitude of gradient trend

for t=1:length(P)
    y(t) = mu + sigma*rand + g*t;
end

plot(P,y);xlabel('time in seconds');ylabel('y');grid;

If you run the above code, you will get you randomly generated plot; such as the following generated below (it will variate based on the value of rand)

Also, because your function is very simple, so you can do that without using for-loop as follows:

mu=3; % mean value
s=1; % stepsize >> decrease it to get more points
P=1:s:20; % time period
sigma=5; % noise level
g=2; % magnitude of gradient trend
RND=rand(1,length(P)); % to generate a vector of random numbers equal to the time period length

y = mu + sigma*RND + g*P;

plot(P,y);xlabel('time in seconds');ylabel('y');grid;

I hope this will help

[ampaul]

Thank you. That is exactly what I was looking for.