Hi everybody,

I've just created a spider diagram consisting of lines departing from different points (representing desire lines of students) that arrive to other points (representing schools). I've divided the township area in various sub-areas and I need to trace a sigle line for all the students going to a particular school in that sub-area. I need to collect lines in a sigle line placed possibly in the middle of the spider diagram, creating a field that will count all the lines arriving to that determined point (the school). I tried the integrate tool, this does not merge lines but just make them coincident, if falling inside a given x,y tolerance.

This tool seems to work in some cases but I have more than 2700 line features (students) and usually the tool does not integrate well ( beacuse I need to integrate just lines going to the same school) or just erase some line features.

Does anyone know if there's a way to perform this task properly and in a reasonable amount of time?

Thanks for helping

I've just created a spider diagram consisting of lines departing from different points (representing desire lines of students) that arrive to other points (representing schools). I've divided the township area in various sub-areas and I need to trace a sigle line for all the students going to a particular school in that sub-area. I need to collect lines in a sigle line placed possibly in the middle of the spider diagram, creating a field that will count all the lines arriving to that determined point (the school). I tried the integrate tool, this does not merge lines but just make them coincident, if falling inside a given x,y tolerance.

This tool seems to work in some cases but I have more than 2700 line features (students) and usually the tool does not integrate well ( beacuse I need to integrate just lines going to the same school) or just erase some line features.

Does anyone know if there's a way to perform this task properly and in a reasonable amount of time?

Thanks for helping

Would it work to have all the students within a given subarea be represented as coming directly from the centroid of that subarea regardless of what school they attended using a straight line to each school? Those lines could easily be integrated into a count (you don't even need a line to create the count, just the subarea and school association to every student, which could be shown in a variety of locations and ways). Any example of an end product you are trying to emulate?

A centroid of a mutlipoint derived from a dissolve of all the students points that come from the same subarea and attend the same school could also be used to define origin points for students with similar characteristics. Such centroids would show some distribution of student clustering by their schools within each subarea. The main problem is that a multi-point centroid could fall outside of the actual sub-area, so more processing is needed to correct for that situation. But it give a more true "location" for the origin of the students being aggregated than Integrate will.

Fancier efforts could be made to deal with point density clustering of students within a sub-area to derive an origin location for the student group at the density centroid using statistical methods. For example, get the mean X values and Y values of all students who originate in the same subarea and attend the same school to get a student density centroid. Same problem as the multi-point as far as potential to fall outside of the real subarea, but probably the point derived by this method has a higher likelihood of falling in the actual subarea than a simple multi-point centroid.

I hope this gives you an idea of the approach, which is cluster the student origin points by subarea first before creating a line to the school. All two-point lines from one origin to one destination can be aggregated easily spatially or by other means and they would all represent the same generalization category to as "true" of a location representation as you can get.