How to automate drawing a polyline from 2 intersecting datasets based on proximity?

Question

I've been stuck on a situation recently where my company's data has to be integrated with another.
Basically there are two point shapefiles comprising of >50000 points, and as shown in the image below, one data set lies between the other spatially.
The output I need would be around <100 polylines based on a unique identifier within the two datasets.
The problem occurs because once they are merged, the FID number (which I would usually use on the Line type within the points to line tool) no longer has any spatial use as my own dataset is at the top of the attribute table and the other at the bottom (no longer in spatial order).

Can anyone think of a solution to this?
A polyline tool based on proximity would be perfect for this situation but I am unaware of one existing... (I've already thought about sorting by northing/easting, but that creates problems when the data does not go in a continuous orientation)

Answer 1

Providing you can differentiate between group of points to form a polyline (distance tolerance?), the following works:

1. Find point for which total of distances to other points in the group is maximum. This makes a first point of target polyline (first point in the array)
2. Iterate through the rest of the points to find nearest, add to the array, etc.
3. Convert array of points into polyline.

The solution works fine for lines shown in your example, i.e. with the ends far apart and no sharp bends. To handle corners I usually split points into groups at the corner.

Answer 2

add a field, [draworder] calculate the fid into it.
Then use it to draw the polylines.
BTW ETTOOLs has a good tool for this.