Example of Proximity Analysis

daniel · December 8, 2011 7:27AM

I've created a report showing how far I have managed on this side project to create a specified radius which highlights all surrounding data points by a defined distance. The details are on Note in each block on the Data Manager page and gives us a demo where by selecting a point of interest you can also highlight all the postcodes within 0.5 km of it.

Possible application:

Suppose each POI is a school and as a soon-to-be parent you may wish to locate within half a mile of a school to ensure your child will be in proximity of a school. If you have a preference of a particular school ie at EC2A 1WS is a top London school then you can find a list of postcodes which falls into that radius. Then this may factor into the equation of where to buy a new home.

This sort of dilemma can easily be applied to setting up retail locations, tourists attractions, planning applications and distance from local transport.

However this has been completed only for a small area of Central London, to apply a similar technique across the whole of London or the UK it will create a massive number of permutations depending on the total number of locations and how many POIs there are. As such I am looking for suggestions and ideas to refine this process down.

I have been encouraged to put this onto the forums to see if others have a similar situation and get some help with this. It has been suggested to use scripting to condense the permutations to a single row to make it it much more manageable, but I have no background in this whatsoever so I appreciate any help I can get on this.

tjbate · December 9, 2011 9:30AM

Daniel - You may want to consider using two decimal slider variables {Trial Lat} and {Trial Long} to give the file a starting perspective by defining a single (variable) trial focal point within each file.

There are various POI databases available so instead of finding all postcodes in a certain distance you can try calculating the distance of all the POIs from the {Trial Lat} and {Trial Long} analytical location points, using Sorts and RANKs of the calculated distance (with threshold max distance values) to list the nearest POIs for any given trial location coordinates.

Using multiple views and Named Queries keyed to maximum distance for each type of POI, you should be able to create an interactive, iterative location finder that works by the analyst moving the trial location variable sliders in the Sidebar until the optimal zone in terms of distance to various different types of POIs is identified.

Remember to reset/re-callibrate the default, maximum and minimum values of the trial location variables after each iteration.

daniel · December 9, 2011 10:37AM

Any chance of a demo showing what you mean? I think I get the general idea to fix it for a single focal point and working from that point to find whats around it. Though Im not sure how that would look on the data side and how I would need to manipulate the data to do.

I think I would prefer to avoid using decimal lat/longs as people do not generally know what lat/long they are precisely interested in and is more likely to know a postcode more than anything else.

If the idea is to enter a postcode in and then have the software calculate what is in range and pull up the records then it may be workable. I would really like to see something demonstrating your suggestion so I can better understand it.

Thanks Thomas

benjamin · February 24, 2012 7:11AM

I cannot find the distance function in omniscope function list

benjamin · February 24, 2012 7:18AM

I have found it now in 2.7

Welcome!

Categories

Ideas Parade