Space Time Hotspot analysis

[AnthoniaOnyeahialam]

hello,
I have malaria data at hospital level for 20 hospitals on a monthly basis over a period of 10 years. I will like to do some hotspots in space and time but I haven't a clue of where to start even with arranging my data because I have the data separately from the georeferenced hospital database. If I succeed in doing the database, where can I start with the hotspot analysis? is there a possibility to do it in space and time with point data? Txs

[origudes]

Probably the best example which is addressing your need is the following link:San Francisco Crime Mapping
http://www.arcgis.com/home/item.html?id=06b67414932343868896c44c7ccda2b5

I think that will help.

Cheers Ori

[LaurenScott]

Hi,
You have 20 hospitals, but lots of data over time at each hospital, is that correct?  I ask because for most statistics it is important to have at least 30 observations in order to trust your results. 

We have lots of resources to help you learn more about hot spot analysis in ArcGIS.  Please check out the short videos, free web seminars, and tutorials at www.esriurl.com/spatialstats

At present, the most straightforward way to do a space time hot spot analysis is to create time snapshots of your data, run hot spot analysis on each time frame, then present the results via animation or small map multiples (creating a map for each time period and presenting them together).  Alternatively, you could create a custom spatial weights matrix file that links features based on both spatial and temporal proximity.  This would be a bit of work.  At 10.1 (the next release of the ArcGIS software), our Generate Spatial Weights Matrxi tool will allow you to create those space-time  relationships very easily and then use the resultant spatial weights matrix file to perform space-time hot spot analysis.  If you can get into the beta program for ArcGIS 10.1, you could get access to that functionality now (??)

If you are interested in constructing the space-time spatial weights matrix manually, here is some information about the file format.  You would want features that were within the same space window (like at the same or a nearby hospital) AND within the same time window (malaria count values within 3 days of each other, for example) to be assigned a weighting of 1... otherwise the weighting should be 0.  Once you create the space-time spatial weights matrix file, you would simply use it when you run the hot spot analysis tool.  Please let me know if you need additional information.
http://help.arcgis.com/en/arcgisdesktop/10.0/help/index.html#/Modeling_spatial_relationships/005p000...   <-- please scroll to the bottom of the document where it says Spatial Weights Matrix File (.swm)

I hope this is helpful,
Please let me know if I can provide additional information.
Lauren

Lauren M Scott, PhD
Esri
Geoprocessing and Analysis, Spatial Statistics

[PhilipGlasner1]

Hi Lauren,

where can I download the Temporal Collect Events tool that you have been using in this video: What's New in Spatial Statistics: Space-Time Cluster Analysis | Esri Video‌?

Thanks

Philip

[LaurenScott]

Hi Philip,

I'm attaching a zip file with the sample tool... it is pretty rough and has only been tested with 10.1 sp1.  This question comes up often, so please allow me to answer this question more broadly here:

You want to run a space time hot spot analysis on your event data (crime, disease, traffic accidents) where you don't have an attribute/field to analyze.  In other words you just want to know where in space and time you have a statistically large number of events.  Some people have guessed that they need to use the Collect Events tool.  Unfortunately when Collect Events combines points that are near in space, you lose the temporal component of your data (it will combine two points that are near each other even if they have date stamps that are very far apart).  What we need here is a tool that aggregates based on space AND time.  You want to be able to set a distance threshold (500 meters, for example) and a time threshold (something like 5 days) and have the tool aggregate only those points that are within both those thresholds.  The attached sample script will do that. 

Alternative approaches if the attached script doesn't work for you:

1) Create a model tool that iterates through time and selects only those features/events that meet your time requirement... run collect events on the selection set... use the Add Field tool to give the result a DATE field and Calc the value to be a date within the time period selected (like if you want to combine events within two days, you would want to calc the new date field to either the first or second day date for each record output from Collect Events).  Then merge all the results into a single file... then create the spatial weights matrix for the merged data using Generate Spatial Weight Matrix, then run hot spot analysis.

2) If you know python, you can try to debug the attached to make it work for whatever version of ArcGIS you are using.

3) You can sign up for the ArcGIS 10.3 beta program and use the space-time pattern mining tool in ArcGIS 10.3 ?? The new Emerging Hot Spot Analysis tool allows you to run space-time hot spot analysis on event data.

I hope this helps!

Best wishes,

Lauren

[PhilipGlasner1]

Hi Lauren,

thank you for your prompt reply. In the meanwhile, I already found the tool that was included in the training course ARC3, version 2.0.

(Anyway, how can users get access to the attached zip file on GeoNet?)

I already tested the Emerging Hot Spot Analysis tool and I think it is very useful in space-time crime analysis. Do you have any further reference on how the categories are calculated?

Thank you,

Philip

[LaurenScott]

Well I am still getting familiar with GeoNet too, but this is what I just tried (and it worked for me):

1) Click on the attachment

2) A pop up is displayed and one of the options is to save.

3) You should see a little blue down arrow when you say okay.  If you click on it, you can get the zip file.

If that doesn't work for you I will find someone who can tell us a better solution.

Here is some more information about the Emerging Hot Spot Analysis categories:

*************** Category Definitions ********************

Hot spots that are statistically significant for the last time step interval:

    New: only the most recent time step interval is hot

    Persistent: at least 90% of the time step intervals are hot, with no trend up or down

    Intensifying: at least 90% of the time step intervals are hot, and becoming hotter over time

    Diminishing: at least 90% of the time step intervals are hot, and becoming less hot over time

    Consecutive: an uninterrupted run of hot time step intervals, comprised of less than 90% of all intervals

    Sporadic: some of the time step intervals are hot

    Oscillating: some of the time step intervals are hot, some are cold

Hot spots that are not statistically significant for the
last time step interval:

    Historic: at least 90% of the time step intervals are hot, but the most recent time step interval is not

Cold spots that are statistically significant for the last time step interval:

    New: only the most recent time step interval is cold

    Persistent: at least 90% of the time step intervals are cold, with no trend up or down

    Intensifying: at least 90% of the time step intervals are cold, and becoming colder over time

    Diminishing: at least 90% of the time step intervals are cold, and becoming less cold over time

    Consecutive: an uninterrupted run of cold time step intervals, comprised of less than 90% of all intervals

    Sporadic: some of the time step intervals are cold

    Oscillating: some of the time step intervals are cold, some are hot

Cold spots that are not statistically significant for the last time step interval:

    Historic:at least 90% of the time step intervals are cold, but the most recent time step interval is not

We won't be able to get it done for the first release, unfortunately, but in a future release you will be able to select the categories you are interested in and also modify how each category is defined (i.e., right now a persistent hot spot is one where 90% of the time step intervals are statistically significant hot spots and there are no statistically significant cold spots... you might want to change it to 80%, for example).

Beta 5 will have better cell size and time interval defaults when you don't provide anything for those parameters, and messages defining categories.

I hope this helps.

Best wishes,

Lauren

[origudes]

Thanks Lauren

always your answer are  useful and helpful.

btw, I have managed to download the zip file with

the temporal collect event tool, but it did not work.

i guess I will focus in using the emerging hotspot tool for now.

regards Ori

[LaurenScott]

Just curious... what version of Desktop ArcGIS are you using?  We will be slammed until the 10.3 release is done, but if we can, we'll try to get the temporal collect events sample script working for 10.2.2.

Thanks!

Lauren

[PhilipGlasner1]

I zipped the Temporal Collect Events toolbox from ARC3 training course which works fine with me using 10.2.2.

Dropbox - TemporalCollectEvent.zip‌

philip