Add noise to discrete time series

Hi,

I have a long-term time series composed by discrete labels,the values are 1,2,3,4.

Now, I want to add noise to these time series changing in randomly position the value of the label.

I need to pass the percentage of the data to change (for ex. 5% or 10%), evaluate randomly the positions to change and change the value in the positions selected.

Example:

input: [11122323444442343211] % 20 elements

Add noise for 10 % = 2 elements...

select randomly positions of 2 elements and change the value

output: [11142323441442343211]

Thanks in advance.

Andrea

Answer 1

[Image Analyst]

Try this rather straightforward approach using randperm():

inputValues = [1,1,1,2,2,3,2,3,4,4,4,4,4,2,3,4,3,2,1,1] % 20 elements
% Find out how many elements 10% of our input vector is.
elementsToChange = round(0.1 * length(inputValues))
% Randomly choose that many locations to change the value of.
indexesToChange = randperm(length(inputValues), elementsToChange)
outputValues = inputValues; % Initialize
% Get new values for the indexes that we want to change.
newValues = randi(4, 1, length(indexesToChange))
% Assign those values to the locations we selected.
outputValues(indexesToChange) = newValues

Let me know if there is anything you don't follow or understand.