Aggregating features is a way to summarize them into logical groups using statistical methods. ArcGIS AllSource supports visualization of aggregated data for exploration and analysis purposes.
You can aggregate multiple features or you can aggregate multiple records (observations) for a feature or group of related features. Examples include the aggregation of individual census data in different levels of census boundaries, symbolizing weather stations using total highest temperature recorded at each station, the distribution of a disease outbreak across a region over time, or average volume of traffic along freeways during rush hour.
There are geoprocessing tools such as Aggregate Points that produce static aggregation. As data changes or your analyzing parameters (such as a time interval) change, you must rerun the tools to create new output.
Dynamic aggregation
Dynamic aggregation is responsive to changes in your data and to changes in scale. There are three main ways that dynamic aggregation is used. Dense arrangements of features can be aggregated into polygonal containers called bins or into grouped containers called clusters, or collected values can be aggregated to related features.
Aggregate features into bins
An example of dynamic aggregation into bins is tracking the concentration and spread of the Ebola virus. Typically, any reported incidence is recorded as a point feature. At large scales, you want to see each instance as an individual point. When you are zoomed out to smaller scales, the conglomeration of overlapping points prevents any comprehension of patterns in the data. So, at smaller scales, it is convenient to aggregate the features into bins, with each bin symbolized based on the number of Ebola cases within each bin.
See Aggregate features into bins for more information.
Aggregate features into clusters
Dynamic aggregation into clusters is an alternative method to visualize features, especially with smaller-scale maps.
An example to consider is a geocoded feature layer of home addresses for hospital patients in a town. Clustering identifies trends of where patients are located. You can use a definition query to determine whether patients in certain groups are most concentrated in certain areas in the town. You can also use mode summary statistics with unique values symbology to identify which groups are prevalent in which areas. Some patient address locations may not be within the cluster radius, so they remain unclustered and drawn with the layer primary symbology. You can adjust the size of the cluster radius, the extent of the map, and the map scale to achieve results that best fit your need.
See Aggregate features into clusters for more information.
Aggregate values into related features
Some data that is collected over time does not vary in location. For example, weather data such as temperature, precipitation, and wind speed is collected multiple times a day from a static location. This data is usually stored in a nonspatial attribute table. Symbolizing each static location with a summarized temporal classification of the data, such as monthly total rainfall or weekly mean temperature, can reveal more useful patterns than viewing the individual records.
A second example aggregates values into existing polygons. Although aggregating features into equal bins produces a map showing concentration of Ebola in a region, maps that show concentrations of outbreaks within actual political or administrative boundaries are necessary to respond appropriately with medicine and medical resources.
See Aggregate values into related features for more information.
Aggregation considerations
Both feature binning and feature clustering methods achieve similar goals but are visually and behaviorally different. Consider the following when choosing which aggregation method to apply.
Feature binning is the more predictable approach to feature aggregation when compared to feature clustering. The alignment of the bins is consistent, and the point features they represent fall within the bounds of their bin. This improves data interpretation and reduces data noise. Clusters may dynamically change location as you pan and zoom around the map, depending on the centroid of their represented features. The exact location of a cluster's features is not always clear.
However, feature binning obscures much of the map while clustering allows other features or the basemap to remain partially visible. For example, if a point feature is not clustered, it continues to be drawn as a singular point feature. With feature binning, a single point is drawn as a bin.
Consider aggregating point features into clusters to see trends in the location and arrangement of features. With feature clustering, the clusters update dynamically depending on the map's scale and extent. Clusters support additional symbology types, such as unique values, unclassed colors, and proportional symbology.
Using heat map symbology to show densely populated features is another way to visualize dense point information. Aggregating features into bins or clusters may better represent the data for sparsely distributed groups of points and may be preferable for multiscale maps in which the level of detail frequently changes or requires insets.