Smoothing by applying defined functions

  • Applying a defined function (e.g. linear, exponential, sigmoidal) is likely more intepretable than a general-purpose smoother

  • But how do you know if the function you chose is a good fit? Model statistics (e.g. r-squared)?