Location Data Explained: Harnessing the Power of Places

Location Data Explained: Harnessing the Power of Places


By Tim Ludolph | Data Product Manager

As the number of smartphone users continues to boom worldwide, the need to understand the world around us—from information and events to the people and locations creating them—has become a critical commodity for businesses, leaders and users alike. Premise leverages its network of nearly 3M contributors in over 100 countries to gather sentiment data (e.g., area safety, event impact, personal opinion), demographic data (e.g., language, religion, ethnicity), and location data through our app.  

Premise collects sentiment and demographic data through recurring surveys on a range of topics—anything from current events (COVID’s impact, election security, etc.) to more generalized information (availability of goods and services, satisfaction with political leaders and security forces, the effectiveness of health care, etc.). However, it’s locational information that allows our customers to contextualize these responses and better understand the environments they come from.  

Definitions and Displays

Before delving into the many ways geospatial or location data helps our customers, we must first understand what it is. At its simplest, location data is information tied to a specific spot on the earth—this can be a point on a map, a road or border, or an entire region. It can refer to a person, place or thing, and represent both tangible and invisible elements of each—from the nearest supermarket or hospital’s location to the surge in orange juice prices or the spread of disease and people’s feelings regarding both.

This breadth and diversity in uses of location data is possible because each of these previous examples (and many others) include information that can be plotted on a map. To do this, you can either use raster or vector formatting to visualize the information. Raster represents the data as a series of cells or pixels, while vector presents it as a point, line or polygon. 

Raster representations map things like terrain features based on satellite imagery or population density based on demographic data—things that don’t have clearly defined, static boundaries but require very high granularity in the display. Heatmaps are popular examples of these, like the below image showing cellular signal strength in Mali.

On the other hand, we use vector representations for things with clearly defined borders—roads, waterways and buildings, or cities, states and countries. Even individual humans, such as those responding to one of our sentiment surveys. These vectors are displayed using a series of points, lines or shapes on the map, depending on the dataset. The display below provides an example of Premise’s health site data mapped in the Philippines.

Targeted Observations

To visualize the data, we must first collect it, which relies on our contributors completing tasks through our smartphone app. Some of these tasks are general observation (i.e. “record the location of a grocery store”). While others are directed observations (i.e. “record the location of a nearby Whole Foods”). Each of these helps build our reference layers of places, which currently hold over 700,000 locations from over 45 different categories worldwide—from health care facilities, schools and banks to critical infrastructure like bridges, cell towers and power plants. 

What sets our location data apart is that each submission also contains contextual information about the site (quality of service, availability of goods, safety, etc.) and user-captured images. This gives our customers an expansive knowledge of the physical locations in their environment when taken as a whole. 

Take the below example from a park in Mexico—not only do we identify its location, but capture key characteristics that potential visitors might be interested in, such as available amenities, personal security, and facility safety (such as broken or unsafe features). This richness in detail gives users all the information they need in one regularly updated place—a sort of “living” map of their area.

Separately, each of the 70M submissions we’ve gathered to date on sentiment and demographics also contains location information, which can be useful for our customers to visualize. Whether it’s the location of São Paulo sentiment on security restrictions, Bengali speakers in the eastern part of India, or 45 to 55-year-old women in Morocco who want better access to education—each of these are potentially useful populations for our customers to engage with and display on a map. For example, the map below shows how we can visualize the primary ethnicity breakdown in the Philippines at two geospatial levels—the state/province (L1) and county/district (L2) levels.

While other organizations may collect similar types of information, their data at its best often reflects a slice in time and quickly becomes outdated. Think of political polling by media outlets, the current U.S. census, or even Google Street View—those capture perspectives, populations and places that are continually changing and incredibly costly to keep up to date. (One of the reasons the census occurs only once a decade and Street View is updated once a year or less in most cases.) Premise’s ongoing data collection allows customers to fill the gaps in other datasets and maintain a more current view of life on the ground.  

Numerous Uses

As we’ve touched on so far, location data has a wide range of potential applications, from aid organizations to businesses across the commercial spectrum. Consider a few more examples—if there’s unrest or an outbreak of a disease in a region, security and aid organizations might need to understand the physical terrain, the local population, the thoughts and feelings of those people, and the presence of key buildings and facilities. These elements each have location data associated with them and are more valuable if examined on a map before taking action.

Similar scenarios are prevalent in the business world—say you’re trying to launch a new soft drink in an area. You would want to understand the local population, their buying habits, and the businesses you would need to stock and market your product. Looking at this information on a map will help our customers develop a plan of action. The flexibility, granularity and volume of location data we provide to our customers is nearly limitless and allows them to easily visualize the information to make timely, effective decisions.

Quality Controls

None of the above scenarios is possible if the data driving those decisions is faulty. Premise is always working to ensure our location data is of the highest possible quality. Our data passes through a series of automated and manual quality control checks, rejecting submissions where users have committed fraud or completed a task improperly. This system of checks allows us to purge the most fraudulent information so we can focus on refining the massive amount of high-quality information that remains.

We employ several algorithms to progressively improve the data and highlight our confidence in it. For example, our “Place Harmonizer” algorithm takes a location’s coordinates, name, and photos to distill hundreds of contributor submissions of a facility into a single high-confidence point. This process runs on a recurring basis as the machine learning models reevaluate and refine those high-confidence points with every new submission. 

Through the course of this, we also develop a confidence score for each location. We base this on a simple concept, the more contributors who say a facility is in a certain place and the more recent their responses are, the higher the probability the place actually exists. Using statistics we translate this into a quantitative measure so users can easily evaluate areas of interest and make informed decisions. We can also re-energize collection requirements if our confidence in a location/region drops.  

Putting it All Together

As is hopefully evident by now, location information buttresses almost every decision organizations have to make today. The need for access to high quality, current data that can be viewed in a way that makes sense is crucial. Premise’s ability to help organizations tackle this problem at scale and provide easily digestible visualizations—from the national level to the counties within them—allows organizations to operate with the speed and agility they need to keep up in today’s marketplace. 

Let us help you get started—email us for more information at info@premise.com

About Tim Ludolph

Tim Ludolph is Premise’s data product manager, ensuring our collection, analysis and visualization efforts provide insights from our global contributor network that have maximum impact on our customers’ daily operations. He has over fourteen years experience working with numerous U.S. government agencies and is a master’s graduate of American University’s School of International Service.