Editor’s note: Today’s blog post comes from Erika Yamasaki, Group Product Manager, Google Maps Platform and Yolee Escuadra, Technical Sales Engineer at Navagis. In this blog they share how Navagis recently conducted an analysis to quantify how proximity to local amenities, such as subways stations and supermarkets, impact land prices.
How do you determine property value? For real estate developers, investors, and urban planners, this is a fundamental question that goes beyond looking at a simple address. The value of a property is deeply connected to the characteristics of its location, including amenities, services, and infrastructure that surrounds it. Being able to quantify how key location characteristics directly impact land value can help businesses better forecast pricing for new properties and provide real estate agents and developers key insights into valuation trends that can give them a competitive edge. This is where Places Insights becomes a differentiating factor for businesses by enhancing your analysis with rich, real-world context.
Our Google Maps Platform partner Navagis recently conducted an analysis to quantify how proximity to local amenities, such as subways stations and supermarkets, impact land prices. They examined two different Japanese cities, Tokyo and Yokkaichi, to understand these relationships. Their analysis provides a powerful blueprint for using geospatial data to unlock deeper, more rich insights into the real estate market.
The challenge of quantifying a neighborhood
Understanding property valuation can seem like a black box, especially with so many factors that could impact pricing. Furthermore, these variables can highly vary from neighborhood to neighborhood. Without quantifiable methods to understand property valuation, pricing new properties can rely on expert judgment and knowledge of the local market which makes it hard to scale. However, what if you could move beyond anecdotal evidence, for example the common belief that properties near transit are more valuable, and towards a more quantifiable data-driven model that can be tested and applied broadly.
Identify geospatial characteristics that impact property value using Places Insights and BigQuery ML
The solution is to use a modern, streamlined workflow that combines the power and scale of Google Cloud and Google Maps Platform. Places Insights is a point-of-interest (POI) dataset enabling statistical analysis of Google Maps’ comprehensive coverage and fresh Places data that is available in Analytics Hub. This analysis-ready dataset gives you access to the depth of POI attributes such as over 300 place types, and information about wheelchair accessibility, parking, payment options, and more.
We start with the base third party dataset of property locations and their prices, such as parcels_row table used in the Navagis case study. Next, you combine this data with information about the density of key POIs around each property from Places Insights. You can use the Places Count functions available in Places Insights to count the number of POIs based on key attributes like primary_type of “Subway Station” and within a defined radius of each property’s geocoded coordinates. For this case study, we looked for amenities within a 500m radius of each land parcel. The output is a new, enriched features table called parcel_features and includes the parcel ID, price, and new columns for amenity counts such as count_of_subway_stations, count_of_supermarkets, count_of_restaurants and more. You can tailor your analysis based on whatever types of POIs and POI attributes you believe may have an impact on property values.
Chart comparing aspects of Tokyo and Yokkaichi.
Finally, you can predict property values with BigQuery ML. With the enriched parcel_features table, customers can train a machine learning model on their own price data using a simple create model statement. By fine-tuning models using customer-provided pricing data, the customer can determine the specific statistical relationship, or weights, that amenities have on property values, revealing which places correlate to positive or negative price changes. The benefits of this approach is the model training and prediction happen directly within the data warehouse, simplifying the machine learning operations pipeline significantly.
Surprising insights from Tokyo and Yokkaichi
This analysis generated unexpected findings about the drivers of property value. The analysis for Tokyo revealed a significant correlation between subway station proximity and land value. A property within a 500 meter radius to a subway station was associated with an increase of ¥893,798 in land price, while proximity to a train station is correlated to a decrease of ¥411,128 in property value. We speculate this finding may be attributed to factors like noise for above ground trains, relative to subways which generally operate below ground. It is important to note that this model shows relationships (correlation) and does not provide anything about causation.
Graph and chart showing the price impact impact on real estate prices based on amenity type in Tokyo.
In the city of Yokkaichi, the analysis identified different key drivers for property values. Proximity to universities was the strongest positive driver correlating with a ¥9,135 price increase. This demonstrates that the model can capture unique local dynamics, showing that the amenities considered valuable can vary significantly by city.
Graph and chart showing the price impact impact on real estate prices based on amenity type in Yokkaichi.
The models had significant explanatory power. In Tokyo, the amenity data explained approximately 43% of the variation in land prices. In Yokkaichi, that number was even higher, with amenities explaining about 53% of the price variation. This level of insight, derived almost entirely from out of the box using insights from Google Maps Platform Places data, shows the potential of this approach.
Pie charts showing the % of amenity data that explained land prices.
Practical applications and business impact
The findings from this type of analysis can be turned into actionable advice for several industries. Real estate investors can quickly screen potential investments by analyzing the amenity profile of their neighborhoods and run "what if" scenarios to estimate the impact of future development. Realtors can leverage these key trends and insights to better market and position their listings relative to what’s most valued in the local market. Urban planners can make data informed decisions about where to build new parks, transit stops, or public services to maximize economic and social benefit. Retailers can use the same data for site selection to identify locations with the optimal mix of amenities that attract their target demographics.
Build smarter with machine learning
By combining your own business data with the rich, contextual data from Google Maps Platform directly within BigQuery, you can move from simple mapping to predictive modeling. The Navagis case study proves how a few simple SQL queries combined with BigQuery ML can unlock powerful, scalable, and surprising insights into the factors that drive real world value.
This analysis is just the beginning. By incorporating more advanced modeling or including data on market trends and local zoning, the predictive power can be enhanced even further, turning location intelligence into a true strategic advantage. Ready to start enriching your own data? Sign-up to express interest in testing Google Maps Platform’s Geospatial Analytics products.
