ArcGIS Blog

Analytics

ArcGIS Pro

Enhancements to Bin Size Evaluation in ArcGIS Pro 3.7: Analysis Field and Time Interval

By Martha Bass and Alberto Nieto

We often aggregate points into bins in a grid to understand spatial patterns. But the bin size we select plays an important role in the resulting maps and the stories they tell. The same point data can tell very different stories at different aggregation scales.

Three maps side-by-side showing point data aggregated to hex bins of different sizes over the DC region, from small bins on the left to larger bins on the right.
Bin size impacts the patterns we see and the results of our analysis.

This is often described as part of the Modifiable Areal Unit Problem (MAUP); a problem we often warn about but seldomly give tools to help with!

 

So how do we select an appropriate bin size?

The Evaluate Bin Sizes for Point Aggregation tool was released to help bring confidence to this decision. By testing a series of bin sizes, the tool finds bins that balance summarizing enough while still preserving the spatial point patterns in the data.

A tool diagram showing a point dataset with an arrow to two evaluation charts, and then another arrow to the resulting aggregated dataset.
The tool balances summarization with point pattern preservation.

However, this raises new questions:

  • What bin size should I use if points come with timestamps?
  • What bin size should I use if points come with values of interest?

In ArcGIS Pro 3.7, the tool now supports two enhancements to help with these questions. Let’s see how these work.

 

What bin size should I use if points come with timestamps?

When points have timestamps, you can create spatiotemporal bins to analyze trends and create forecasts. But the same challenges described in the MAUP play a part when considering time: the bin size determines the trends we find and the forecasts we would create.

So, what bin size should we use?

You can now use the Time Field and Time Interval parameters to find bin sizes that properly account for sparser patterns when the data occurs in time intervals. For example:

  • You want to analyze weekly trends of 311 service requests across the city.
  • You set the Time Field parameter to the field containing the date and time of the request.
  • You set the Time Interval parameter to 1 week.
  • The tool splits the request points into weekly intervals before finding appropriate bin sizes.
Tool dialog box with Additional Time Options section highlighted.

The output layer is time-enabled and includes a set of features for each time interval, so you can see the aggregation patterns change over time.

Animated gif showing a time-enabled hex bin aggregation layer over DC stepping through each time interval, revealing different spatial aggregation patterns over time.
311 incident points in DC are aggregated to week-long time intervals, revealing changing patterns over space and time.

The resulting bin size can be used as the bin size for the Create Space Time Cube by Aggregating Points tool before continuing on with your analysis.

Map of hex bins over DC ranging in color from blues to reds with different Emerging Hot Spot Analysis patterns identified in a legend.
Aggregating data into a space-time cube enables pattern identification (using Emerging Hot Spot Analysis in this case) over space and time.

What bin size should I use if points come with values of interest?

Points often have values that we need to analyze, and we can also aggregate point values to bins on a grid. For example, instead of just looking at restaurant point locations, you analyze the average sales revenue for restaurant point locations across each bin.

We face similar MAUP challenges when aggregating values, but there’s a new consideration: a bin mean is only a good representation of an area if the point values are not drastically different from the mean. As bin sizes get larger and larger, the mean often becomes a poorer representation of the area inside.

So how do we find a bin size that summarizes just enough before we lose too much information?

You can now set an Analysis Field, allowing the tool’s bin size recommendation to be driven by averaged attribute values instead of point counts.

Tool dialog box with Analysis Field parameter highlighted.

This is useful for cases where the value itself is the focus – such as business sales revenue, pollutant concentrations, or incident response times – and where you want the chosen bin size to prioritize preserving spatial patterns of those values. When using an Analysis Field, you can choose between combining coincident points (averaging the values of collocated points and treating them as a single point) or handling the points separately.

Two side-by-side maps show points symbolized by house price on the left and those same points aggregated to hex bins on the right, symbolized by the mean house price within each hex bin.
Aggregating point data into hex bins using mean house price summarizes local variation and reveals underlying spatial patterns.

At the end of the day, choosing a bin size is still an analytical decision. But with the Evaluate Bin Sizes for Point Aggregation tool, you can make a more informed decision by better understanding how scale affects your data. Whether you’re evaluating point counts, analysis field values, or either of these across time, the tool can guide you toward bin sizes that balance preserving spatial patterns with summarization. We encourage you to try out these new capabilities with your own data to see how thoughtful aggregation choices can support, rather than obscure, the stories your data are telling.

 

Resources

Share this article

Leave a Reply

Related articles