Currently, the smooth piecewise cubic fit easily gets into a fit of oscillatory behavior.
To fix this, I should implement TIkhonov regularization instead of simple least squares, i.e. do a least-squares that penalizes coefficients size as well as data error.
Currently, the smooth piecewise cubic fit easily gets into a fit of oscillatory behavior.
To fix this, I should implement TIkhonov regularization instead of simple least squares, i.e. do a least-squares that penalizes coefficients size as well as data error.