##Multi-channel support vector regression (SVR) Channel data is combined with an aggregated kernel, computed as:

K(x_i, x_j) = Sum_c [ k^c(x_i^c, x_j^c) / A^c ]

where k^c(x_i^c, x_j^c) is the kernel similarity between observation x_i and x_i with respect to the c-th channel using kernel function k^c, and A^c is the mean value of similarities for the c-th channel.

Requires:
numpy
scikit-learn (for the underlying SVR model)