For consumers today, data can be a great tool for building trust in your brand. If your company can deliver accurate data that helps users safeguard their health, you’ll be well-positioned to nurture trust and improve sales.
That’s why accuracy is the first priority for BreezoMeter’s air quality data. When you’re dealing with the health of today’s well-informed consumers, it has to be. How does BreezoMeter ensure accurate air quality readings down to the city block, all around the world?
There’s 3 parts to that answer.
#1 - Extended Data Collection
Official air quality monitoring stations is where most air quality data providers get their data. It’s an important layer, but for us, it’s only the first step.
We collect raw air quality data from close to ten thousand government monitoring stations worldwide, using measurements from a total of 41,000 sensors. As of today, this is more than any other air pollution data provider.
It is made possible because we’ve done the hard work of building relationships with countries around the world so that we can utilize accurate air quality data legally. And we’re continuing to add countries to our list, based on demand. We only use air quality measurements from official governmental stations to make sure we collect reliable data.
At every monitoring station there can be several times as many sensors, each of them used to detect different pollutants. That’s how we’re able to track 17 different types of pollutants, delivering specific data about what’s in the air so users can make smart health choices.
Beyond the large number of official monitoring stations measurements used in our air quality monitoring system, what makes BreezoMeter truly unique is the additional layers of data we gather.
With traffic heavily impacting air quality in cities, real-time traffic jam data, combined with machine learning techniques allow us to deliver hyperlocal and real-time pollution data around the world. But that’s not all, our data sources also include satellite data, weather and topography. We aim at taking into consideration all the elements impact the quality of our air into our calculations.
#2 - Proprietary Air Quality Algorithms
Our data sources might be the foundation for our air quality data, but it’s actually our algorithm and our QA & testing that make BreezoMeter data so accurate.
Even with so many data sources, there will still be gaps when mapping air quality. Remembering that air quality can differ dramatically between city blocks, we can’t give users potentially false readings on their local air quality.
So how do we fill in those gaps?
By using proprietary dispersion algorithm. After collecting data from monitoring stations and traffic, performing strict quality assurance, and adding input from models relying on weather & satellite data, we use an algorithm based on years of academic and internal research to find out pollution levels in those blank areas.
The algorithms take into account things like local traffic, weather, wind patterns, and more. They then calculate air quality based on 17 different pollutants across the globe, and displays the results over a two-dimensional grid. The distance between two calculated grid points is 500 meters or less, giving us accurate air quality readings at the street level.
Here’s a quick video of how it looks.
#3 - Quality Assurance (QA) & Accuracy Testing
All collected data go through quality assurance processes to eliminate faulty or frozen data that would otherwise jeopardize our accuracy.
Now how do we know our algorithm produces accurate results? Because we validate our accuracy using several statistical methods.
For our in-depth quarterly check at the end of 2016, we performed rigorous testing on our algorithm to make sure it was delivering accurate results. We chose 100 random times, and for each time picked 50 random locations – making a total of 5,000 unique tests worldwide. For every station, we calculated pollution levels at the station’s location using only our algorithm, as if the station didn’t exist - but still using raw data from other governmental sensors. We then compared the result with measurements taken from the actual stations.
The result? While the accuracy would vary in function of pollutant types, with higher accuracy for ozone, and slightly lower accuracy for particulate matter, the mean error was smaller than one category span of the Air Quality Index (AQI).
The quarterly accuracy check for the first three months of 2017 was focused on areas without any governmental monitoring stations. To assess accuracy, we performed our calculations in 38 countries where we do have monitoring stations data, but using none of them to perform the test. We then compared the results between our algorithm calculations and the recorded data at stations. One more time, the accuracy test confirmed a very low mean absolute error, smaller than one category span of the AQI.
That’s reliable data, constantly tested so customers can put their trust in you to deliver accurate data and help protect their health.
Most air quality providers rely solely on pollution monitoring‐stations for their readings, often applying vague blanket data to large areas or just faulty data with no QA. But today’s consumer has little patience for inaccuracy.
With several data layers, strict QA & accuracy testing, and cutting edge algorithms, BreezoMeter brings you the quality of data needed to offer professional air pollution data. Successful collaborations with global brands such as L’Oreal, Dyson & Veolia prove the value of our services.