How to calculate linear regression between two time series and remove?

I want to calculate the following between two time series:

R^2, slope, intercept , p, q,

and if possible q^hat

I know how to get r^2 for the whole timeseries but I would like to get this as a timeseries of coefficients.

I need the slope, P and the intercept in order to calculate q^hat (if it is not possible to get directly using a function) so that I can remove the relationship trend between two time series from one timeseries, like what is done using FERRET here: http://ferret.pmel.noaa.gov/static/FAQ/analysis/regressions.html

So far, I only know of 'corrcoef' which I can use to get R^2 and 'regression' to get the slope.

Can anyone help me?

thanks, Michael

Answer 1

[Star Strider]

The easiest way is to use polyfit and polyval:

b = polyfit(p, q, 1);   % Choose Degree ‘1’ for Linear Regression, b(1)=slope, b(2)=intercept
qhat = polyval(b, p);

Using corrcoef will give you the ‘R’ values, and you can get their statistical significance if you ask for ‘P’ as well.

[Michael]

great! thanks

[Star Strider]

My pleasure!