No transcript available for this video.

What's New in Spatial Statistics: Space-Time Cluster Analysis

Esri staff show how to identify space-time hotspots, detect spatial-temporal outliers, and group features using space-time constraints.

  • Recorded: Jul 24th, 2012
  • Runtime: 14:40
  • Views: 808
  • Published: Nov 2nd, 2012
  • Night Mode (Off)Automatically dim the web site while the video is playing. A few seconds after you start watching the video and stop moving your mouse, your screen will dim. You can auto save this option if you login.
  • HTML5 Video (Off) Play videos using HTML5 Video instead of flash. A modern web browser is required to view videos using HTML5.
Download VideoDownload this video to your computer.
<Embed>Customize the colors and use the HTML code to include this video on your own website
Start From:
Player Color:

Right-click on these links to download and save this video.

Comments  (8)

All Comments
To post a comment, you'll need to login.
If you don't have an Esri Global Login ID, please register here.
Thank you, My analysis is held up however at around the 6:30 minute mark ... If the ICOUNT field has records of say 15 different incidents, each having thier own distinct times per just one FID, then how can I add a time field per each FID if the times are variable in the ICOUNT field? Since you group collected events, you now have several records in the ICount field. I can't just pull over my time field becuase if I select an FID with multiple wieghted Icount's each of the records in the ICount field has its own distinct time.

Say I had 50,000 records and then I integer and then collect events. Now I have 7,000 FID's in the new shapefile. The records are all there, they are just weighted in the ICOUNT field. Problem is to do the temporal analysis, how do you add in the time field for each of these weighted records, when the original time field would have 50,000 records per the 50,000 FID's, not the new 7,000 FID's. I would really like to do a temporal analysis in Arcglobe of my dataset, but creating the unique ID with the time field is holding me up.

I start out with around 50,000 of the 54 NAICS records from 1990 - 2010 and once I integrate and collect events, I end up with around 7,000 FID fields. When I open the spatial weights matrix to add a unique ID, i only have ICOUNT and FID, but in your video you have a unique ID carrying over a time field. My question do I carry over my time field, which is the Year i.e. 1990 if i have say 15 different ICount's for the same FID in the collected events shapefile? Those 15 ICOUNT's could all be different years so just attaching one year to each FID in the table would be wrong.
Jschnitz 1 Day ago
You can find documentation about visualizing the output in 3D here: //005p00000056000000. There is a space before the double slashes (//) to ensure the whole link shows up. Just delete that when you copy and paste the link. Hope this helps!
lrosenshein 1 Week ago
You mention documentation for the 3d? Is that available?
Jschnitz 2 Weeks ago
@thivanka.siriperera  Thank you for providing the additional details. I have forwarded them to Lauren Scott. For further ***istance, please e-mail Lauren directly at She will be able to ***ist you further.
KarenJaffarian  (Staff Comment) 6 Months ago
@KarenJaffarian  Thanks for the quick response, but i still have some things unresolved would appreciate if you can help me resolve these things.

1) Point data that has a time field cannot be integrated with a tolerance before using [collect_events] tool, as each date is required to carry out space-time analysis if I'm not wrong

2) By using the [incremental_autocorrelation] the optimal distance threshold can be identified as the input for the [Getis-Ord Gi*] hotspot tool. Likewise isn't there a tool to find out the best time interval the most pronounce temporal clustering is happening to be provided as a input for creating a "spatial weights matix" for time window option.

3) After I created the spatial weights matrix I tried to collect the events to create [ICOUNT] field thinking it will carry the ***ociated [ICOUNT] values ***ociated to the spatial weights matrix. Then as u suggested I populated the FID field and carried out the hotspot analysis. But the in the processing messages it indicated that the id did not match with the "id" field that was provided as input for hotspot.

4) as a solution what I did was created a table using the swm file then added x coordinate as a field also added a x coordinate field in the collected events feature set and joined both the feature dataset with the table thinking the x coordinates should match with each others. When I did this I was able to go ahead with the time-space analysis

5) Is that method ok according to your best practices?

Sorry for asking so many questions. It would be great if you can clarify these doubtful areas to me soon.

Many Thanks
thivanka.siriperera 6 Months ago
  • Showing 1 to 5 of 8
  • 1
  • 2
  • »