This is a simple Mathematica code module for performing LOESS (locally weighted polynomial regression) smoothing on a set of bivariate {{x1,y1}..} scatter plot data.

Usage¶ ↑

In[17]:= (*load some sample data*)

data = {{0.55782, 
  18.6365}, {2.02173, 103.496}, {2.57733, 150.354}, {3.41403, 
  190.51}, {4.30141, 208.701}}

Out[17]= {{0.55782, 18.6365}, {2.02173, 103.496}, {2.57733, 
 150.354}, {3.41403, 190.51}, {4.30141, 208.701}}

In[23]:= (* Evaluate the LOESS approximation for each data point *)

xs = data[[All, 1]];
Table[Loess[#, data, 1, 0.5] &[x], {x, xs}]

Out[24]= {18.6365, 103.496, 150.354, 190.51, 208.701}

For more information about LOESS, see:

en.wikipedia.org/wiki/Local_regression www.itl.nist.gov/div898/handbook/pmd/section1/pmd144.htm

This software is licensed under The MIT License (MIT). See LICENSE.

Files¶ ↑

• Loess Demonstration.nb

  • A Mathematica notebook demonstrating the usage of the Loess.m module

• Loess.m

  • A Mathematica code module with the Loess implementation.

• test_1.csv

• test_1_expected.csv

  • Test data used to demonstrate and validate the Loess impl.