How to create an information dashboard for visualizing useful data
One of the outcomes of the current COVID-19 pandemic is the fact that many things or activities are now positioned, counted, measured, tracked or labeled. Data collection under these circumstances is important because analytics can then be performed to determine whether we are overcoming the novel coronavirus or not. It’s also important to publish the collected data so that subject matter experts and interested individuals can look for patterns or categorizations in the data. One of the best ways to make near-real-time data available for public consumption on the web is via an ArcGIS dashboard. Read this blog post to see how easy it can be to collect, condition and publish data in a dashboard.
If you look at COVID-19 websites around the world, the majority have maps showing the locations of high rates of infection, but also there are often other graphs or displays showing infections, deaths or testing counts over time. Esri Canada has a COVID-19 resource hub that provides a number of links to health-related dashboards. These dashboards are very professional looking and are easy to understand and manipulate. However, I was wondering how difficult it would be to create and publish a dashboard myself, so I gave it a try to see how easy it really was. Here are my experiences.
So, the first thing you need for a dashboard is some data to display and analyze. I didn’t want to use any of the existing COVID-19 datasets that we already have, as they probably had already been conditioned for use in a web map and dashboard. I wanted to use a raw data file so that I could start from first principles. As I mentioned in last month’s SDI blog, for data elements to be really useful, both location and time are needed for each data point.
There was a dataset that I had often wanted to review that was available from the federal government’s Department of Fisheries and Oceans Canada (DFO). The table contains the average monthly water level for each of the great lakes from 1918 to 2019. This is a sizeable data set and perfectly suited for this dashboard demonstration as it is truly raw data.
Partial table of Great Lakes historical mean water level by month for 101 years.
The first thing you might notice at this government website is that this data is not downloadable, so you have to copy and paste the tables using a process often called “screen scraping”. I selected and copied the appropriate tables from the DFO website and pasted the scraped water level data into an Excel spreadsheet for further processing. For the dashboard, I wanted to have a water level value for each month for the 101 years of available data. I also wanted to have a different set of data including location for each Great Lake. Using some scripts and manual editing in Excel, I created a .csv format spreadsheet for ingest into ArcGIS Pro.
Each water body was identified in each record along with the latitude and longitude of the centre of the lake, which was manually determined. The month and year were combined into a single date field using the 15th day of each month. The water level for each month was included from the raw data. The same water level data was applied to both Lake Michigan and Lake Huron as this is how the data was initially provided by DFO. I presume this is because both these lakes are directly connected and always have the same water level.
Detail from Excel .csv file with the average monthly water level for each of the Great Lakes. The resulting spreadsheet had 7272 records (101 years x 12 months x 6 Great Lakes).
The next step was to load the .csv file into ArcGIS Pro using the XY table to point tool. After the map and water level data was verified as correct, the map was published to ArcGIS Online as a web map.
Water level data displayed in a map and showing the attribute table within ArcGIS Pro.
The published web map on ArcGIS Online was opened, the Create Web App function was selected and then the Dashboards function was selected.
The next process is to create and customize the dashboard to suit your requirements. Here is my resulting dashboard, which is available for viewing here.
Dashboards are very accessible because they are published on the web. Also, they are very useful for providing detailed data by presenting location-based information and analytics using intuitive and interactive data visualizations on a single screen. They are also easy to implement and are a popular way of providing visualizations within organizational SDIs or directly to the public.
The simple Great Lakes dashboard shown above took one to two days overall to create, but it does not contain much metadata and still needs polishing. The largest amount of work was conditioning the initial raw data for ingest into ArcGIS Pro. Once you have an initial simple dashboard set up, you can then use any number of dashboard configuration tools to customize your dashboard as you like. If you haven’t created your own dashboard yet, you should locate your favourite raw data file and turn it into a dashboard for visualizing your spatial data on the Web.
To learn more about how to communicate your spatial data using Esri’s different application builders—including ArcGIS Dashboards—and which builder to choose based on your intended audience, view our webinar, “The Fantastic Four Apps: Which Builder Is Best?” here.