I have X/Y data which if graphed would create somewhat circular graphs. Each circle is made up of approximately 40 data points, however I've recorded, say, 10 passes around the circle. What I need is a method of regressing this data into one singlular pattern which is the best fit through all 400 data points. NOTE: These circles can have somewhat sharp corners like squares or rectangles in some areas, and "dips" in others.
I'm currently doing this by breaking the data up into various segments, and regressing each segment, which works OK. I'm hoping to find something better.
## Deliverables
Here's a sample data file. The first column is the data point number. The 2nd and 3rd columns are the data which make up the "Circular" data. Ignore the other columns.
In this case you will see it is more like "rectangular" data. If you graph this, you will definitely see the trends of the 6 to 10 times around the circle. What I'm looking for is data at a length from the origin, say, every 1 degree of rotation around the circle. This would be like polar coordinates.
The origin would be the average of all the data, approximately the center of the data circle.
* * *This broadcast message was sent to all bidders on Friday Jun 17, 2011 6:25:00 PM:
I've now added an example data file, plus added some details of what I would expect as an output.
* * *This broadcast message was sent to all bidders on Friday Jun 17, 2011 8:25:10 PM:
My new comments are under the "Description" tab, under the "Detailed Requirements" heading for the original project description. Farther down the "Description" page you can obtain the zip file of actual data.