Zoning, land use planning, construction of buildings and infrastructure, energy consumption, and mobility produce an ever-growing amount of data. The use of this data for artificial intelligence (AI) solutions is still in its infancy, but growing. AI enables better forecasts and analysis to support zoning and investment decisions.
Cityfier by A-Insinöörit – forecasts for zoners and real estate developers
A-Insinöörit is a real estate and construction design and consulting company with 800 employees. Cityfier is a software product developed by A-Insinöörit that predicts and analyzes whether it is worthwhile to build, buy, sell, hold, repair, or dismantle in a given area. It's intended for investors, cities, and developers.
Cityfier as a business solution
The original development project was launched because the solution had a clear use value and, based on demand, commercial potential. The solution enables A-Insinöörit to better serve its existing customers.
When investors, cities, and developers have a tool like Cityfier to support their decision making, new demand arises for A-Insinöörit’s construction consulting and design services. As a result, Citifyer has a direct impact on the company’s core business.
Cityfier was already widely used before its developers decided to start enriching the models with general AI technology, especially natural language methods (NLP) and machine learning.
Cityfier uses data-based analytics, which brings together data from several sources and models the impact of traffic and service investments on the future value of real estate.
The application's development team studied the possibilities for using image recognition in cooperation with VTT, Technical Research Centre of Finland. Cityfier serves as a good example of how a functional service becomes more versatile and competitive when complemented by AI technologies.
Image recognition from satellite data
In recent years, various machine vision-based solutions have also become more common in built environment applications. Cityfier, together with VTT, aims to use image recognition from the European Space Agency (ESA) satellite data.
The goal is to identify green areas such as parks, play areas, roads, and coastlines to automatically create a picture of the factors that influence a given area’s price development. Satellite data is particularly useful in new land areas, where it is helpful for understanding the development potential and value.
The accuracy of the model (in this case image recognition) must always be tested before the solution can be implemented. This could not be done in new, partially unknown areas because accurate information on shorelines and green areas is not always available. That’s why the developers decided to test the model in areas where accurate data is available, such as the cities of Helsinki, Espoo, and Vantaa, in order to be able to compare the results of the image recognition with existing data.
Mapping and testing
When thinking about AI solutions, map the process you are creating the solution for and consider how you can test the model's performance to ensure the solution works as desired.
Next, Cityfier intends to further develop housing price forecasting models through machine learning and to find out which regional factors have the greatest impact on housing prices. Machine learning is suitable for tasks that require number crunching, such as analyzing the effect of zoning on property prices per square meter, which helps to optimize zoning and investment decisions. In the past, public sales data has been used for this purpose but this doesn’t provide a particularly good indication of the impact of zoning.
In Finland, educational institutions and universities are very active and there are many different cooperation projects. You can learn a lot from these kinds of projects while keeping costs reasonable.
Cityfier is still in the early stages of machine learning. In addition to A-Insinöörit’s own people, local service providers participate in the development work. In general, it makes sense to start development and testing with external technology experts and consider new recruitment only when the solution begins to generate value.