Note: We at Locale.ai are committed to making the geospatial industry more robust and mainstream. This is part 1 of the series “Ask an Expert” where we engage with geospatial experts 1:1 on interesting ideas, discuss geospatial techniques and tools and talk about the industry!

Introduction

Sayan Ranu is an assistant professor in the Department of Computer Science and Engineering at IIT Delhi. His research interests include Spatio-temporal data analytics, graph indexing, and mining, and bioinformatics. Prior to joining IIT Delhi, he spent close to three years as an Assistant Professor at IIT Madras and a year and a half in the role of a Research Scientist at IBM Research.

As a data scientist, what were the challenges you faced while working with geospatial data?

Today, there exist plenty of tools and libraries in the market to analyze traditional forms of data such as high-dimensional points, images, graphs, etc. But unfortunately, there doesn’t exist a tool to store, analyze, and visualize large-scale and constantly streaming geospatial data.

Consequently, we are often required to spend a significant amount of time and effort building internal tools for spatio-temporal data for which we could have used off-the-shelf tools, especially for datasets from more traditional domains. These tools, often are built off as one-off solutions inside companies and are not very scalable and repeatable to use.

How is movement geospatial analytics different and what’s challenging about it?

Location is not only about a point on a map. It is about a line. It is about movement.

Computing metrics for static assets involves plotting points on a map and calculating metrics for that. Movement using geospatial data is extremely challenging since it is dynamic across two dimensions, space and time. That’s what movement analytics concerns itself with — how do we visualize, analyze and optimize how things move on the ground.

Developing an intuitive understanding of this data involves mining correlations and periodicities across both space and time, which is a computationally challenging task. Doing it in real-time with joins becomes even more complex!

Why do you think movement (geospatial) analytics is difficult to do on BI tools like Tableau or PowerBI?

As we discussed before, movement analytics is all about finding patterns with both location and time. Traditional BI (statistical) tools like Tableau or PowerBI are not equipped to visualize and analyze patterns that are hidden within the movement of geospatial entities. They can only provide information on the static component such as a heat-map of “current” locations, etc.

To give an example, if you want to know the revenue of a branch location, a static map with branch locations and their respective metrics shall work well.

But, movement, paths, aggregations of points become imperative in order to answer questions such as — How do people move? Where do people go from this origin? Where do long-distance trips happen?

What kind of business decisions or optimizations can be done?

Geospatial data in its raw form is not of much value. It becomes valuable when we derive actionable insights from it. Obtaining actionable insights requires us to understand data by mining patterns such as periodicity, anomalies, etc., and then use them for predictive modeling.

This can help in various applications that have anything to do with ground operations — ranging from predicting and matching the demand and supply of assets, selecting routes for ride-sharing or salespeople, selecting the locations for doing offline marketing, calculating ETAs, traffic, finding geo-patterns in business metrics such as order frequency, delays, cancellations, order value and revenue, affluence and so on.

Frequency of rides: Thicker lines indicate higher frequency of rides in the route.

What is the one piece of advice that you will give to companies that have moving assets on the ground?

For companies that have moving assets on the ground — collecting location data is a must. Once they collect this data, they need to understand what’s happening on the ground and how they can optimize this. Visibility is the first step.

Once you get visibility, invest in building predictive models to optimize ground operations and improving unit economics. After all, the ability to predict the future is key to optimizing your business strategies!

Endnotes:

We are Locale.ai, a company focused on geospatial analytics. As geospatial data scientists ourselves, we’ve lived through the pain of dealing with geospatial data at scale first hand. The existing analytics and BI products proved to be futile in our daily workflows.

Over time, we had built our own internal tools to handle high-frequency geospatial data. We then realized data scientists around the globe face similar problems. As a result, businesses are struggling to attain operational efficiency using their location data.

With Locale, we’re committed to making this data accessible to every business with moving assets on the ground!

Read Similar:

A Guide to Kickstart into the Geospatial World.
A collection of the best data sources, open source tools and packages to get started.
Making Location-Based Experimentation a part of our DNA
What can we learn about experiments on the ground from web-based experimentation?

Thanks for reading! Let me your thoughts on these questions. You can reach out to me on LinkedIn and Twitter.