This blog post is about the newly formed UniverCity Council, it's mission, objectives, and participation
Location data is an important source of real-world context and geo-insights for many industry verticals. Market research pundits are projecting USD $20 Billion will be spent on geo-insights by 2020. Our blog post this week revisits the fundamentals to ensure the best return for your location data investment.
What is a Location Signal ?
A location signal is a sequence of numbers that represents the geo-coordinates [ latitude and longitude ] for a point on the map. As an example, below map shows a marker for the Parson's School of Design located at 5th Avenue, New York. The geo-coordinates for Parson's are : [40.735316, -73.994583]
How precise is this blue marker? How accurate is the location for the place it represents? We've all heard that location data needs to be both accurate and precise in order to be actionable. What that does that really mean? What is the acceptable accuracy and precision for data signals used for location intelligence?
CITYDATA.ai has put together this cheat sheet to unravel the mysteries of location accuracy and precision, for the benefit of everyone in the industry.
How Precise are Location Signals?
Precision indicates the exactness of the location signal. The precision of the raw latitude and longitude is represented by the number of digits after the decimal point, ranging from 1 to 9. The table below lists the precision levels and explains the significance of each digit.
|Decimal Place||Queen's Distance||American Distance||Translation|
|1||10 kilometers||6.2 miles||Town or City|
|2||1 kilometer||0.62 miles||Neighborhood|
|3||100 meters||328 feet||City Block|
|4||10 meters||33 feet||Street Address|
|5||1 kilometer||0.62 miles||Neighborhood|
|5||1 meter||3 feet||Store Entrance|
|6||10 centimeters||4 inches||Drone Control|
|7||1 centimeter||1/2 an inch||Military Operations|
|8||1 millimeter||Paper Clip||Civil Engineering|
|9||0.1 millimeter||Strand of Hair||Likely fake data|
Turns out the blue marker for Parson's is actually very precise. When evaluating sources of location signals, it is important to verify the precision of the dataset to ensure that latitude and longitude values include at least 4 digits after the decimal point. In other words, the signals should be precise to at least 10 meters or 33 feet for the location dataset to be useful for geo-targeting and audience segmentation.
How Accurate are Location Signals ?
Accuracy indicates conformance with reality. Represented by a 2 digit number measured in meters, accuracy is the error margin in the reported latitude and longitude values relative to the ground truth. Most smartphone GPS receivers have an accuracy of 5 meters or approximately 15 feet. However the accuracy could worsen near buildings, structures, trees, or indoors. Accuracy could also fluctuate with cloud cover and weather conditions.
Accuracy can be horizontal or vertical. The accuracy reported in location datasets is almost always the horizontal accuracy. GPS receivers in mobile phones are not designed to capture vertical accuracy beyond 5 meters, rendering such measurement moot in dense urban or downtown settings.
It is not uncommon to see location datasets with accuracy ranging from 1 meter to 100 meters. Below is the horizontal accuracy distribution for one of our datasets comprised of approximately 2 billion location signals obtained from 20 million mobile devices in the US.
When considering location datasets, it is important to understand the accuracy distribution of geo-signals to ensure that the bulk of the data has an accuracy better than or equal to 10 meters. Any more that 20 meters would be unacceptable for deriving actionable intelligence and meaningful geo-insights from your location data.
Evaluating your location datasets
CITYDATA.ai has evolved a sophisticated set of tools and techniques for rapid analysis of large location datasets with Billions of data points. Reach out to us if you'd like us to evaluate your location data and provide a summary report for free.