The Local Dog Park May Be a Hotspot — but Not the Kind You Think

In the nascent stages of any technological advancement, the market often witnesses the emergence of products that fail to withstand rigorous scrutiny. The early years of commercially available geolocation data were no exception, with results that frequently bordered on the absurd.

I've encountered numerous destination leaders who express profound skepticism towards geolocation data. And for good reason. Their disillusionment stems from implausible findings, such as data suggesting that the most frequented location in their area is a local dog park or a municipal golf course. Such glaring inconsistencies understandably erode trust in the entire dataset.

In the initial phases of Zartico’s work, we too encountered these anomalies. While recognizing their dubious nature, we still believed that beneath these inconsistencies lay valuable, actionable insights waiting to be unearthed.

And we were right. Through the development of our patent-pending methodologies designed to address inherent data inconsistencies, Zartico now offers unparalleled accuracy in destination insights.

The perils of unrefined data

Zartico licenses the largest commercially available feed of raw geolocation data, processing an astounding 2 billion data points daily, sourced from a comparable number of mobile devices worldwide.

This vast dataset allows us to discern patterns that might elude those who do not use the entire U.S. raw feed. For instance, we identified inexplicable spikes in activity at neighborhood parks, exemplified by the anomalous hotspots overshadowing Loring Park, a 34-acre urban oasis in Minneapolis.

Here’s what we learned:

1. Hotspots occur when the geographic coordinates recorded with a mobile device observation are approximate, rather than precise. 
2. Approximate observations can cluster in centralized locations, often in such high concentrations that they eclipse the natural distribution of precise observations, thereby skewing insights focused on these areas.
3. These clustering data points constitute approximately 15% of all observations within our clients' defined points of interest.
4. In cases where a geographic centroid—the mathematical center of a city, county, or state—falls within one of these locations, this percentage can escalate to as high as 38%.

All geolocation data streams inherently contain a mix of approximate and precise observations. Without proper mitigation, these hotspots can lead to gross misrepresentations, such as erroneously identifying a neighborhood dog park as a premier attraction, and thereby casting doubt on the story your geolocation data is telling.

Developing a sophisticated filtering mechanism

We recognized the critical importance of accurate geolocation insights for destination leaders - who rely on this data to inform your marketing strategies, engage with local officials, and make decisions about funding and product development. That’s why we invested significant resources in creating a solution to the hotspot problem.

Our data science team designed a series of experiments to understand the behavior of hotspots, then they developed a dynamic filter to proactively predict and neutralize them. This approach was necessitated by the constantly evolving nature of hotspots, which change in both location and characteristics over time. Moreover, an indiscriminate elimination of all potential hotspots risked discarding valuable, accurate data.  It’s a complex problem that requires a sophisticated solution. One that only Zartico has solved.

Zartico's probability-driven hotspot filter selectively removes problematic observations while preserving 95% of precise geolocation data. This allows us to isolate and amplify the signal amidst the noise, revealing true patterns of human movement and place engagement.

In this before-and-after image, the white spikes on the right denote hotspots in downtown Salt Lake City. The filtered image on the left shows the remaining geolocation signal, revealing the true patterns of people and places.

This groundbreaking work formed the basis of Zartico's inaugural patent application, detailed further in our comprehensive hotspot whitepaper. 

Pioneering data science innovation

Our efforts to solve data science anomalies extend beyond hotspot mitigation. Zartico's geolocation normalization process addresses data imbalances that can skew insights, enhancing representativity while retaining every legitimate data point for maximum visibility. This methodology, which has led to another patent application, has been extended to balance origin market data within our spending insights.  Read more about the vital process of normalization.

Zartico's position at the forefront of data science innovation is bolstered by our continuous assessment of the full stream of raw data for the entire United States and beyond. This panoramic view enables us to account for inconsistencies across multiple dimensions, translating into a suite of insights that inspire confidence and can withstand rigorous scrutiny and auditing.

With half a decade of experience and two patent applications to our credit, we are redoubling our efforts, investing in proprietary datasets and developing tools essential for destinations to maintain competitiveness in the contemporary tourism landscape. Zartico is dedicated to developing tools that will lead this industry into the future. 

While your local dog park may indeed be a cherished community asset, it is assuredly not your top attraction. With our refined methodologies, destinations no longer need to contend with unfiltered data that suggests otherwise, but can instead rely on Zartico’s sophisticated, accurate insights to drive strategic decision-making.