Thank you for this very useful tool. I have used it succesfully to study time trends of various pollutants. However, for one pollutant I always get the error Runtime error 13: Type mismatch. I could not identify any difference with the other datasets that ran fine. Any thoughts?

Here’s the data:
2/13/2001 140
2/26/2001 129
3/5/2001 80
11/13/2001 23
3/8/2002 3
11/9/2002 138
12/18/2002 24
2/12/2003 110
2/3/2004 169
2/19/2004 12.2
2/27/2004 232
10/19/2004 28.1
10/28/2004 39
12/6/2004 4.89
1/10/2005 83.8
10/19/2005 4.14
11/10/2005 13.8
2/21/2006 23.8
3/1/2006 26.2
12/11/2006 7.4
2/23/2007 5.8
4/21/2007 2.9
9/24/2007 5
12/19/2007 5.7
1/24/2008 14.6
11/26/2008 26.3
12/16/2008 11.3
2/7/2009 2.6
3/5/2009 2.8
10/14/2009 21
12/8/2009 20
2/6/2010 24
10/7/2010 20
10/31/2010 47
2/17/2011 7.7

The program works okay on your data, but there are a couple important features in your data.

1. Your data should be sorted in order of increasing X values.

2. Your data seems to show two regions. There is a steep increase from 0 to to mid-50s, followed by a gradual increase from mid-50s to upper-60s.

There are only about 50 points in the initial steep region. Using the default setting of 0.3 for alpha means each output point uses 300 input points for the calculations, so the LOESS calculations don’t show much of the initial region. In fact, even using N=50 doesn’t show the full drop to zero at the start of the data.

You may want to investigate spline fitting for your data.

Fantastic Job;

I was looking for something like this for ages and i got so excited when i saw your examples undertaken by LOESS, however, it didnt work for my data. Maybe i’m doing something wrong during introducing X,Y. I’ll be grateful if yo can look at my data.

How can i send you my file?