Hi,
Multiset canonical correlation analysis (MCCA) can indeed be used to address the problem of comparing accelerometer data from different locations, taking into account the different orientations.
To implement this you can use the "canoncorr" function, which is a built-in MATLAB function. This function calculates the canonical correlations between two sets of variables.
After loading and preprocessing your data to the required standards, you could use the following code
[A, B, r] = canoncorr(X, Y)
The outputs, "A" and "B" are the canonical coefficients for X and Y, respectively. These matrices contain the weights for linearly combining the variables to obtain the canonical variates and "r" is a vector containing the canonical correlations between the sets. The elements of "r" represent the strength of the correlations.
