Home
Project Admins:
Demixed Principal Component Analysis (dPCA) is a new data exploration technique. Just like Principal Component Analysis (PCA), dPCA searches for a subspace that captures a high amount of information about q data set. However, often your data points have labels like time, stimulus presented, reward achieved, etc. In contrast to PCA, that completely ignores these labels, dPCA benefits from the labels and tries to find components that capture variance due to only a small subsets of the labels thereby facilitating the interpretation of the components.
A paper about this technique is just about to be published. Python and Matlab code will be published soon.
