nav[aria-label="Primary Navigation"] { padding: 0; & ul { list-style: none; width: 100%; display: flex; flex-direction: row; justify-content: start; align-items: start; gap: 30px; padding: 0; & li { margin: 0; } & ul li { list-style: none; } } }

GeoStatistical Analysis on two shapefiles

PeterMerx1

Hi,

I have two shapefiles each containing a set of polygons. The polygons in both shapefiles contain pieces of land and all have a value. There's no relationship between the classes, I only want to search for a possible spatial relationship like "what happens to the values from set 1 if they are close to a polygon of set 2"?

How would you suggest to do this?

Sincerely, P

Najděte více příspěvků označených jako

Accepted answers

All comments

XanderBakker

Hi Peter Merx ,

I'll try to answer your questions and comment on other parts.

What I did was Generate a Near Table with a search radius of 3 km and a maximum of 5 features. Input is the Green Polygons and Near Features the Orange Polygons. As in your example my orange polygons also have a value which might have an impact so I exported it to SPSS and made a function (value/near_distance)/near_distance. I made it this way because I expect a positive relation between the values of green and the values of orange, and I expect that the closer they are to each other the higher the values of green will get.

Your suggestion of using the Generate Near Table, made me think that this could serve the analysis of mangrove and villages. So I did a test and the it is interesting because it allows me to view things from 2 different perspectives (at least in my case). If there is a demand for mangrove in a village to which mangrove forest they will go use that probability on impact to sum all those potential impacts from each village on the mangrove polygons. Now you mention that you exported the result to SPSS to make a function and do some analysis. I think most of the analysis surely could have been done in ArcGIS, but normally this is a choice depending on what one feels comfortable with.

It would be interesting if you could share a screenshot from your result to see what it looks like.

So I have now for each green polygon 1-5 values of the weight of the orange polygons. I then sum those values and divide them by the number of it. Then I have an averaged weight of the value x distance of the nearby orange polygons.Then I can use these values to perform a regression...You have comments or suggestions on this method?

Is it really a matter of the average of the weights from multiple orange polygons around the green ones or would a sum of those values be better? In other words, is there a cumulative effect?

@Xander Bakker: Is the approach you're using almost the same as creating a Heat Map with values as weight in ArcGIS Pro?

A heat map is normally a representation to indicate where a certain feature is more present. IMHO it is not really an exact analysis but more a visualization. To do a "real" analysis you could use hot spot analysis (An overview of the Mapping Clusters toolset—Help | ArcGIS Desktop ) or use one of the Density tools in Spatial Analyst (An overview of the Density toolset—Help | ArcGIS Desktop )

PeterMerx1

Hi Xander,

Thank you for your post. Very good understandable and also very interesting to see how you approach my question.

What I did was Generate a Near Table with a search radius of 3 km and a maximum of 5 features. Input is the Green Polygons and Near Features the Orange Polygons. As in your example my orange polygons also have a value which might have an impact so I exported it to SPSS and made a function (value/near_distance)/near_distance. I made it this way because I expect a positive relation between the values of green and the values of orange, and I expect that the closer they are to each other the higher the values of green will get.

So I have now for each green polygon 1-5 values of the weight of the orange polygons. I then sum those values and divide them by the number of it. Then I have an averaged weight of the value x distance of the nearby orange polygons.

Then I can use these values to perform a regression...You have comments or suggestions on this method?

@Xander Bakker: Is the approach you're using almost the same as creating a Heat Map with values as weight in ArcGIS Pro?

XanderBakker

Hi Peter Merx ,

You could use Near—Help | ArcGIS Desktop to calculate the distance to the nearest orange feature for each input green feature:

After that you can graph your prices against the NEAR_DIST field using a scatter plot:

...and investigate if there is a trend between the two data fields.

However, you probably should dive in a little deeper to see if the size of a orange polygon influences and perhaps there is a cumulative effect when you have multiple orange polygons nearby.

In my case of analyzing impact on mangrove (blue outline) due to nearby villages I had to include the cumulative effect of multiple nearby villages and taking into account the size of the villages (grey polygons):

To obtains this result I had to:

use Multiple Ring Buffer—Help | ArcGIS Desktop on different ranges of the villages
determine the potential impact due to distance and village size
Merge—Help | ArcGIS Desktop the 3 multiple ring buffer featureclasses into one
Intersect—Help | ArcGIS Desktop the merged rings with a single polygon covering the entire area (this will also intersect all the overlapping ring polygons, creating a huge amount of overlapping polygons)
Use Feature To Point—Help | ArcGIS Desktop to create label points inside each polygon
define a ID based on the X and Y coordinates (round those and concatenate them into a new field)
use Add Join—Help | ArcGIS Desktop to join those ID's to the polygons
use Dissolve—Help | ArcGIS Desktop to dissolve the small polygons applying a SUM on the potential impact value

This resulted in this:

But you can also choose to use a raster approach:

XanderBakker

I assume that Near—Help | ArcGIS Desktop would at least give you a distance to the nearest feature in the other featureclass. You could use Creating scatterplot graphs—Help | ArcGIS Desktop to graph the values against the distance and examine if there is a relationship, but keep in mind that relationships are often more complex and other factors are involved too.

PeterMerx1

Hi Dan,

Thanks for your response!

I included a picture. I want to know if there's a relationship between the value of the green polygons and nearbyness of the yellow polygons. Is this better understandable?

DanPatterson_Retired

what do you mean by 'what happens to the values' ?

what process are you studying?

are you just checking to see if there a co-location relationship? (ie if X is found, Y is found nearby, but if there is no Y, then there is no X)

An image would also go a long way to illustrating what pattern of attributes you are looking at

Připojte se k diskusi

Přihlášení členové mohou přispívat, sledovat aktualizace a další. Jste tu noví? Zaregistrujte si bezplatný účet.

Rychlé odkazy

Community Help
Find useful guides, FAQs, and documents to help you navigate and make the most of Esri Community.