2026_retreat_workshops

notes on the unfold workshop:

• the XX' tip is wrong & missleading, should be X'X, and probably better to write to do a qr of X'X - I dont fully understand how pinv and qr(X) interact in terms of parameters / performance
• The linearmodelfunction should be based on 0+event to remove any dummy-coding, then the beta vector could be [1,-1] and not show as much confusion about the ground truth