ESRI User Conference 2023 – Tools for Decision Support

Jul 14, 2023

In search of the promised land
I don't know if the 18,000 attendees at this year's ESRI User Conference makes it their biggest conference ever, but it certainly seemed like the number and diversity of sessions, products, teams, vendors, and use-cases was at an all time high. I selected sessions based around the broad goal of developing decision support tools and systems. Providing decision support for ag and natural resource managers is one of primary functions of cooperative extension, and a role that geospatial technology in particular is well-suited for. One might argue that all maps help people think about the world around them and function day-to-day, but GIS tools and data can also be leveraged to aid specific decision making contexts that require more than a simple map.

ESRI has developed a wide range of products and services that can be brought together to support a range of decision support contexts. Some of these are "no code" tools to develop web apps relatively easily. A good example of this is ArcGIS Dashboards, which allows you to bring together with relative ease a range of data sources with a modest number of transformations, layouts, and widgets. These are good options for relatively simple use-cases and if your data are already in the ArcGIS ecosystem.

However I actually get more excited about the building blocks of these systems, which to its credit ESRI also makes available. You can think of these as legos that can be combined in different configurations to support specific decision contexts. They include tools for data importing, pre-processing, analysis, and hosting, as well as delivery through a custom web interface. Below are a few highlights from my week at the ESRI conference, themed around the main components of a decision support system.

Data Pipelines. ArcGIS Data Pipelines is a fairly new tool (still in beta) for ArcGIS Online that allows you to create a data processing pipeline that imports vector or tabular data, apply some transformations, and (re)publish it as a Feature Service. A typical use case would be ingesting data from a non-ESRI platform (think open-source data portals commonly used by municipalities), clean it up, and republish it on AGOL (or Enterprise). Previously you would have to do this manually or write Python scripts for this kind of thing. Being able to import and clean up data through a flow-chart style interface is a pretty cool functionality, and particularly useful for decision support systems which often require bringing together disparate data sources under one umbrella. ArcGIS Data Pipelines exist 100% on AGOL, which is convenient but presents some limitations on what you can do. Right now all the import "jobs" have to be executed manually from the flow chart style editor, but it will support scheduling when it comes out of beta this fall. Pro Tip: although ESRI isn't charging an extra license fee to use Data Pipelines, credits are charged according to how much time you spend in the Editors, not the actual amount of data processed. So be sure to close the Editor when you're done with it!

Data Processing with Python. In terms of pre-processing and analyzing data, ESRI has a strong set of options including both "point-and-click" solutions and scripting tools. The scripting languages provide the most flexibility, and within these Python is far-and-away the most developed in the ESRI universe. ESRI has two primary packages for Python, including the venerable "arcpy" package and the recently rebranded "ArcGIS Map SDK for Python". The former is recommended if you're working locally (i.e., on your laptop), while that later is recommended if you're working in a cloud environment. Both have hundreds of functions for all kinds of data manipulations, as well as other functions for administering ArcGIS resources and interfacing with their various products. I'm a relatively newbie to working with spatial data in Python, so I attended a number of Python sessions that covered both the big picture as well as specific workflows. I look forward to working with these more and using them in projects.

Constructing Composite Indices. One of my favorite sessions this year was on constructing composite indices. Indices are commonly used in decision support systems to simplify decisions involving lots of different types of data - think about a system designed to allocate resources based on a multi-dimensional construct like "vulnerability", "conservation value", or "acquisition value". These are hard problems to wrap your head around, because they inherently involve multiple criteria which are often apples and oranges. In IGIS we bump into this need all the time. ArcGIS Pro now has a "Create Composite Index" data processing tool. But what I really appreciated about the workshop was the robust discussion of best practices, and how construction of a 'good' index is by no means a technical problem. Bias is introduced at every corner whether you like it or not, including how you normalize the individual layers, and how you combine layers into an single composite index. ESRI has some smart people thinking about this, and I would recommend everyone who creates or uses Composite Indices to read their recent whitepaper Creating Composite Indices Using ArcGIS: Best Practices.

Building Front Ends. Many people think about web-based decision support systems in terms of the 'front end', or web app. Although the magic of decision support is almost always grounded on the data or model, a well-designed web app is effectively the gatekeeper for sharing the value of research with users. ESRI has developed a number of "preset" web app tools, that combine a modest number of layouts and capabilities. These includes everything from simple Dashboards to the more flexible "Instant" and 'Experience Builder' web apps. These are all "no-code" solutions, which work well enough for many use-cases. Where more custom functionality is needed, ESRI continues to develop and document the underlying components. Of note this year is a reorganization of the JavaScript and Python libraries, which is not a game changer in itself but makes it a little easier to find what you need to construct a dataset or web app. If you like the look of ESRI's web apps, ESRI has also made many of the web app 'controls' (like a button or color selector) in their Calcite Design Framework available to web developers. But the most exciting announcement this year was that many of the underlying components of ESRI's 'out-of-the-box' web apps (like a map viewer or legend) will soon be released as open source web components. This will dramatically lower the bar for creating custom web apps scratch, and enable people like myself who don't have the time or patience to fine-tune web GIS JavaScript commands for a standard UI/UX.


All-in-all, attending the ESRI User Conference was an enriching experience. In addition to learning a lot of technical info, I met the ESRI developers who build these tools, other users like myself who are trying to figure out how to put the pieces together, and vendors who add value to ESRI's ecosystem. In IGIS, we like to say a 5 minute consultation can save you 5 hours of frustration. With the info and contacts I picked up at the User Conference, I feel like I made a big step forward that will save me a lot of time and effort applying these tools to upcoming projects.


By Andrew J Lyons
Author - Program Coordinator