2018 Esri Partner Conference and Developer Summit – Part 2

The timing worked superbly, like the best Swiss clockwork: A few days before winter made a comeback in Switzerland, I sat in a plane to Los Angeles. Nevermind that California also had slightly cooler temperatures than usual – it was definitely preferable over the polar cold air masses that firmly occupied Switzerland. Even the place names felt evocative: Santa Cruz, Big Sur, and San Francisco. For two weeks I would cruise California, before making my way back to L.A. and then Palm Springs in order to attend the 2018 Esri Partner Conference and Developer Summit together with my colleague, Nicole Sulzberger. In what follows, we describe what we learned during the two Esri events: the latest news about developments at Esri.

Part 1 of this review has been published last week.

The Science of Where

As described previously, The Science of Where is still Esri’s tagline. Esri aims to apply the science of where to help answering spatial questions with:

  • increased efficiency to save resources
  • better analysis to actually understand what is going on, and
  • better communication to foster good decisions

Many of the recent developments shown during the Partner Conference and the Developer Summit can be linked very well to at least one, often several, of these three promises.


Select Highlights (continued from Part 1)


The big news of Esri in terms of data analysis was quite a mouthful: Esri Geo AI Data Science Virtual Machine (DSVM) on Microsoft Azure. That’s „GeoAI DSVM“ for short.  What is behind this? Geo AI DSVM is a virtual machine in the Microsoft Azure cloud that combines ArcGIS Pro and a plethora of Microsoft data science toolkits. It’s part of Microsoft’s „AI for Earth“ project. The VM contains pre-configured installations of, for example, Python, R, VisualStudio, RStudio, Microsoft Powershell, various Python and R packages, Power BI, and a Jupyter Notebook Server. So there is a lot of things that allow you to dive into GIS-supported data science in a scalable cloud environment. In order to use the GeoAI DSVM you need to have an ArcGIS Pro license and Azure VM usage charges apply. An overview of the GeoAI DSVM can be found in the Microsoft Azure Marketplace. On Github, Esri offers an example of a pixel-level landcover classification using Deep Learning with Microsoft’s Cognitive Toolkit, that can be used in conjunction with the Geo AI DSVM.

Geo AI DSVM was a big part of Joseph Sirosh’s (Corporate Vice President in the AI Research group at Microsoft) keynote address:


Jupyter Notebooks

Throughout the conference, various data science and machine learning examples were highlighted, and often demonstrated with Jupyter Notebooks – basically an interactive Python environment in your browser that lends itself ideally for making data analysis workflows more transparent and reproducible through integration of code, documentation, and output. Jupyter Notebooks can also be used with the Python API for ArcGIS for, e.g., Portal administration, however, if you are so inclined. If you do data analysis in Jupyter using, e.g. arcpy, results are by default temporary but can be persisted onto a Portal or locally. Esri offers http://notebooks.esri.com for testing Jupyter Notebooks.

One example that was shown using Jupyter was the extraction of SAM sites from orthoimagery using a neural network:

A planned feature for ArcGIS Portal is the integration of Jupyter Notebooks. You will be able to share your Jupyter Notebooks with your colleagues on your ArcGIS Portal.

And Other Things Python

In other Python news, we found an emphasis on ArcGIS Enterprise and Online automation using Python, specifically the ArcGIS API for Python for communicating with a web GIS. Example tasks that can be done through this pythonic API were the creation of groups and user accounts, the assignment of accounts to groups, and of content to users, cloning a portal, re-assignment of content, creation of reports about content, as well as publishing new and pruning old content. The plenary session had an Automation with Python slot that highlights some of the key developments around these topics.

Secondly, Python in ArcGIS Pro was a big topic and also part of the plenary session. Some of the key things to know: ArcGIS Pro comes with Python version 3, rather than 2.7 like ArcGIS 10.x. Further, the Python installation is conda-based. (Ana)conda is a widely used Python package and virtual environment manager that should make the lives of Python developers easier. Thanks to the conda-based installation, many relevant Python packages are pre-installed, for example the whole SciPy stack (this includes pandas). There have been numerous other improvements, big and small, of the Python developer experience, for example for those of you who like to work in Microsoft VisualStudio.

If you want to know more about these topics, check out the videos and the above links: Automation with Python and  Python in ArcGIS Pro.

Exploratory Data Analysis with Insights for ArcGIS

Insights, the data exploration solution by Esri, was highlighted throughout the event (earlier versions of Insights have been shown in previous events). This tool allows to carry out data analysis using a drag-and-drop interface that lets the user build a collection of „cards“ that can contain maps, charts, or tables. Users can interact with different cards using the linked view paradigm where features in a card are highlighted based on a user interaction in another card.

ArcGIS Insights (source: Esri)

Insights further allows joining data dynamically (not sure to what data set size this stays performant) and the analysis that a user builds is represented in a graphical model that can be shared with other users. Since December 2017, Insights is newly available also in ArcGIS Online (previously it was part of ArcGIS Enterprise): To perform analysis in Insights for ArcGIS, users need to purchase a subscription, in addition to an ArcGIS Online Level 2 named user license. A Level 1 named user license for ArcGIS Online provides you view-only access to Insights.


Also on the Table

There was much, much more on the plate: improvements around the performance of the GeoEvent Server, the Spatiotemporal Big Data Store and the GeoAnalytics Server, for example, but also in deployment with Docker and Kubernetes, UX and UI, data in the Living Atlas, as well as IoT and real-time applications.


And Where Do We Go From Here?

In our opinion, it was rightly emphasised in the plenary session during the conference: the future lies in

  • connecting separate software systems,
  • expanding collaboration between individuals, teams, departments, and organizations,
  • integrating all kinds of data in common views, be they interactive plots and visualizations, feature layers, maps or web scenes,
  • and adding powerful exploration and analysis of data.

In the perspective of Esri, these ingredients enable a new scale in the trajectory of GIS (if you still want to call it that): GIS will turn into a system of systems.

However, this process doesn’t happen by itself but requires careful thinking and designing.

If any of these piqued your interest, please get in touch with us. We are happy to think along with you and assist in designing tomorrow’s workflows, systems and tools!


Part 1 of this review has been published last week.


Ralph Straumann

Ralph Straumann

Ralph Straumann (Dr. sc. nat.) hat an der Universität Zürich Geographie mit Vertiefung in GIS, Wirtschaftsgeographie und Politologie studiert. Seit 2010 arbeitet er bei EBP Informatik als Senior Consultant. Seit 2018 leitet er das Tätigkeitsfeld Datascience. Er berät Kunden bei strategischen Fragen, zu Geschäftsprozessen und Organisation sowie bezüglich Quellen, Modellierung, Workflows und Analyse mit verschiedenartigen Daten im Schnittbereich zwischen IT/GIS und Anwendungsfeldern wie Verkehr und Raumplanung. Mail: ralph.straumann@ebp.ch Ralph Straumann auf: LinkedIn Xing GitHub

Das könnte Dich auch interessieren...

1 Antwort

  1. 9. April 2018

    […] → Click through to Part 2 of this review. […]