7
$\begingroup$

From the Data Transforms and Smoothing guide there is a link to the Linear and Nonlinear Filters guide. Do any of these filter functions perform a LOESS or LOWESS filter on set of paired data? Searches for "loess" and "lowess" come up empty in Mathematica help but things sometimes go by another name. I really prefer the LOWESS filter over the LOESS filter.

If not, is there another function (maybe an image function) or would this have to be coded from scratch?

Some data to demonstrate the smoothing on:

dat = {{2.5388, 160}, {1.791, 130}, {2.88673, 140}, {2.05391, 130}, {2.1987,
   130}, {2.22434, 140}, {2.53868, 170}, {2.59566, 150}, {1.68293, 
  130}, {2.32455, 140}, {2.70714, 160}, {2.29725, 140}, {2.26859, 
  140}, {2.59289, 160}, {2.12666, 140}, {2.20377, 130}, {3.14684, 
  150}, {2.34561, 130}, {1.68741, 120}, {2.51823, 130}, {2.8703, 
  160}, {2.94802, 180}, {3.10839, 180}, {3.39548, 170}, {1.38768, 
  81}, {1.76145, 97}, {1.88736, 97}, {1.92727, 102}, {1.8698, 
  102}, {1.6378, 104}, {1.85618, 112}, {1.8413, 112}, {1.83661, 
  119}, {2.1003, 122}, {2.18675, 124}, {2.20111, 127}, {2.23716, 
  127}, {2.65555, 132}, {2.55666, 132}, {2.54618, 137}, {2.7703, 
  142}, {3.06095, 145}, {2.73968, 147}, {2.6985, 147}, {2.93909, 
  152}, {2.839, 152}, {2.9723, 155}, {2.79205, 157}, {2.79859, 
  160}, {3.25364, 168}, {2.72511, 168}, {2.4127, 170}, {3.29402, 
  170}, {3.28745, 170}, {3.06602, 172}, {3.12907, 170}, {3.54607, 
  180}, {2.31484, 140}, {2.8195, 170}, {2.77789, 150}, {2.655, 
  170}, {2.79148, 170}, {1.29638, 78}, {1.47898, 84}, {3.12846, 
  182}, {3.04629, 179}}
$\endgroup$
  • 1
    $\begingroup$ An implementation of LOESS is available in this answer by @Rahul. $\endgroup$ – MarcoB Feb 10 '16 at 23:25
  • $\begingroup$ @MarcoB Thanks, that is interesting. After reading all the answers there I am thinking that I may be able to do something with MovingMap and LeastSquares for LOESS. I really want LOWESS though so I'll see what answers are posted. $\endgroup$ – Edmund Feb 10 '16 at 23:43
  • $\begingroup$ Edmund, could you spell out the difference between LOWESS and LOESS for me, as well as for the good of the question? $\endgroup$ – MarcoB Feb 11 '16 at 0:41
  • 1
    $\begingroup$ @MarcoB Both use a moving window from left to right across the data. The width of the window varies based on the density of the points in a neighbourhood of the window. At each view a regression is performed and the fitted value of the mid-point of the window is taken as the smoothed point. LOESS does a basic linear regression for each view. LOWESS uses a weighting for the points in the view with points in the centre of the window having greater weight than points near the edge. $\endgroup$ – Edmund Feb 11 '16 at 0:58
  • 2
    $\begingroup$ It seems that everyone has a different idea of what the difference between loess and lowess is (and none of them agree with you): en.wikipedia.org/wiki/Local_regression, stat.ethz.ch/pipermail/bioconductor/2003-September/002337.html, mathworks.com/help/curvefit/lowess-smoothing.html, stats.stackexchange.com/q/161069 $\endgroup$ – user484 Feb 11 '16 at 2:06
7
$\begingroup$

I am afraid that this is not really an answer, but a collection of bookmarks for future reference, since this question is bound to come up in searches about LOESS and LOWESS on this site. Here are a few implementations found searching the web:

| improve this answer | |
$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.