Repeat and Near Repeat Analysis

Calculate Prediction Zones

Summary

Identify areas at risk of repeat and near repeat incidents by specifying the space and time range of influence of past incidents. Optionally, append the resulting polygons to an existing feature service.

Usage

  • The Input Features must be a point feature class or shape file with a date field, representing the locations of incidents.
  • The tool will use all points with date values in the Temporal Range of Influence preceding the Initial Processing Date. Features with date values outside this range, and features with null or invalid date values will not be processed.
  • Use the values TODAY and YESTERDAY to have the tool calculate the Initial Processing Date value at run time based on the current date and time. Alternatively, specify a date using the format yyyy-mm-dd. The dates in this field may be displayed in any format.
  • This tool will honor a selected set of features. When the Input Features contain a selection, only the selected features will be considered when building the prediction zones.
  • Incidents that occur closer in time to the Initial Processing Date will have more influence over the prediction zones than older incidents. Areas nearer to incidents will be considered to have a greater risk of future incidents, up to the Spatial Range of Influence. Use the Spatial Half Distance and Temporal Half Life parameters to modify the rate of decay of influence.
  • The tool also requires an existing polygon feature class as input. Polygons representing the Prediction Zones will be appended to this feature class. The necessary fields will be added by the tool if they do not already exist. Existing features in this feature class will not be deleted or modified. The following fields are populated as output:
    • CREATEDATE: Text field populated with the date and time the features are created.
    • STARTDATE: The Initial Processing Date, or the calculated date if TODAY or YESTERDAY is the input value.
    • MOSTRECENT: This field will be True for the most recently created features and False for all features created previously.
    • TIMEBAND: The Temporal Range of Influence used to generate the features.
    • SPACEBAND: The Spatial Range of Influence used to generate the features.
    • RISKRANGE: The classification of risk for the area defined by that polygon. Higher values have a higher level of risk and lower values have a lower value of risk based on the location and dates of nearby incidents.
  • This tool creates up to three outputs:
    1. A raster representation of the levels of risk over the area covered by the incidents.
    2. Polygons generated by binning the raster values into a number of Risk Ranges are appended to the Output Prediction Zones Feature Class as described above.
    3. Optionally, the output polygon features may be appended to an ArcGIS Online, ArcGIS for Server or ArcGIS Enterprise service. This service must already exist with the same schema as the Output Prediction Zones Feature Class, including any fields added by the tool.
  • While storing the output raster in a geodatabase is supported, it is recommended to store the raster in a folder rather than a geodatabase, especially if this tool will be run as a scheduled task.
  • All distances are calculated using geodesic measurements.
  • This tool requires the Spatial Analyst extension.

Parameters

Parameter Explanation Data Type
Input Features Feature class containing points representing the location of incidents and from which prediction zones will be calculated. The feature class must have a date field (in date format) and all features must have date values. The feature class must use a projected coordinate system. Feature Layer
Date Field The field in the input feature class containing the date on which each incident occurred. If the date of the incident spans a range (for example, date information is recorded in two fields representing from and to dates), choose the date field you consider to be most appropriate. This will usually be the date at the beginning of the date range. Values in the date field are used to calculate the level of future risk based on a combination of when the incident occurred, the Initial Processing Date and the temporal decay in risk (determined from the Temporal Range of Influence and the Temporal Half Life). Field
Initial Processing Date The date to use from the Input Feature Class for creating the prediction zones. The value TODAY automatically uses the current date as the Initial Processing Date. The value YESTERDAY automatically uses the date before today as the Initial Processing Date. The prediction zones will be created using incidents that took place between and including the Initial Processing Date and the Temporal Range of Influence value. Alternative dates (in other words, dates before yesterday) can be used as the Initial Processing Date but must be entered in the format yyyy-mm-dd. String
Spatial Range of Influence This value determines the size of the prediction zones around each incident (in the units of the coordinate system of the Input Feature Class). The value refers to the maximum distance from an incident that the incident is estimated to have on influencing the risk of future incidents taking place. Additionally, this value can be based on the area around an incident that is considered to be practical for resource targeting and deployment (for example, the area to which additional patrols are targeted). Double
Spatial Half Distance The distance from an incident at which the risk is estimated to be half that of another incident taking place (a near-repeat). This value is used to calculate the exponential rate of decay in the spatial risk of a near-repeat incident and must be lower than the value for the Spatial Range of Influence. Distance units are the units of the coordinate system of the input feature class. A useful initial value to apply is half the value of the Spatial Range of Influence. Double
Temporal Range of Influence This value refers to the number of days before the Initial Processing Date that incidents are estimated to have an influence on the risk of future incidents taking place. More recent incidents have the greatest level of influence on future risk. Incidents that took place between the Initial Processing Date and the value for the Temporal Range of Influence are used for creating prediction zones. For example, if this value was set to 3, only incidents for the Initial Processing Date and the two days prior would be used for creating prediction zones. Incidents that took place 3 days ago would have less of an influence on the risk of future incidents than those that took place on the Initial Processing Date. Double
Temporal Half Life The number of days before the Initial Processing Date on which the risk is estimated to be half that of further incidents taking place immediately after the Initial Processing Date (in other words, incidents that took place on the Initial Processing Date are considered to have the greatest influence on the risk of future incidents taking place. Incidents before the Initial Processing Date have less of an influence on future risk). This value is used to calculate the exponential rate of decay in the temporal risk of a repeat and near-repeat incident and must be lower than the value for the Temporal Range of Influence. A useful initial value to apply is half the value of the Temporal Range of Influence. Double
Risk Calculation Method Choose the method for calculating the predictive risk.
  • 'CUMULATIVE' (default) creates prediction zones where the value of each cell is the sum of any prediction zone cells that overlap;
  • 'MAXIMUM' creates prediction zones where the value of each cell is the maximum value of any prediction zone cells that overlap.
 String
Output Prediction Zones Raster Location Workspace where the output prediction raster will be saved. Workspace or Raster Catalog
Output Prediction Zones Feature Class Output feature class to which the polygons representing the prediction zones will be appended. The output feature class must already have been created and be a polygon feature class. New polygons and associated attribute information (described below) are appended to the existing output feature class. The following fields will be added to the output feature class if they do not already exist (in other words, at the first time of processing prediction zones):
  • grid code (Long Integer): Prediction zone classification of the polygon (a number between 1 and the Number of Risk Ranges). Higher values represent polygons of highest risk.
  • Id (Long Integer): feature identifier
  • CREATEDATE (Date): The date and time the prediction zones were created. This can be used to identify previous sets of prediction zones.
  • MOSTRECENT (String): TRUE for the most recently generated set of prediction zones, FALSE for all previous sets of prediction zones.
 Feature Class
Number of Risk Ranges Integer value representing the number of ranges that will be used in the polygon version of the output for representing the variation in predicted risk (for example, if this value was set to 2, two Risk Ranges would determine areas of primary risk from areas of secondary risk). Long
Publish Polygons (Optional) Enable this option to add the prediction zone polygons to a polygon feature service previously published. Boolean
Publishing Environment (Optional)

Choose the prediction zone service publication environment:

  • ARCGIS_ONLINE
  • ARCGIS_PORTAL
  • ARCGIS_SERVER

This environment must already contain a service with a schema that matches the prediction zones polygon feature class, including the four additional fields.

 String
Username (Optional) Administrative username for the service. This could be either an organization administrator or the owner of the service. String
Password (Optional) Password corresponding to the administrative username. String Hidden
Server URL (Optional) URL to the data store hosting the prediction zones service. For ArcGIS Online organizations, this will likely be the URL of your organization, for example [prefix].maps.arcgis.com String
Polygon Service URL (Optional) URL to the rest endpoint of the prediction zones layer. When working with hosted feature layer in ArcGIS Online, Portal for ArcGIS, or ArcGIS Enterprise this URL can be found on the services' item detail page. String
Top