IREEN (Intelligent Real Estate Evaluation Network) is a software for residential property appraisal and real estate market analysis produced by AI.Automatica. We aim to beat current available software in terms of valuation accuracy by deploying modern AI aglorithms and developing new statistical tools tailored to the problem of property appraisal. One of the highlights of IREEN is the introduction of computer vision, i.e. we use images of the object and of its environment. A most accurate appraisal software will yield many highly valuable tools for market analysis and forecasting.
Our target customer segments are banks and real estate related firms like appraisal offices, realtors, insurance companies and investors.
for the in- and outside visual desirability of objects and their environment - we also compute a visual desirability index
elaborated structural modeling of geo-spatial relationships between prices in a deep learning setup
see how geo-information (e.g. nearby facilities, air-quality, noise, etc.) is influencing prices
e.g. request the average price per square meter, average income etc. in a usder defined neighbourhood
compute what an object in the neighbourhood would cost if it had the exact features of your target object
generate automatically an appraisal report with all relevant data, export to word/pdf and edit
Evaluation of real estate objects and markets is complex: properties are extremely heterogeneous and prices are driven by a myriad of different factors ranging from tangible features (e.g. age of the building) to intangible features (e.g. design trends). Automated valuation models for real estate (AVMs) cannot deal with this complexity and are often considered as insufficient. However, in recent years, the situation developed much in favour of computer valuation: For most of the transactions, the objects are advertised online and real estate experiences a digital revolution, providing a lot more data for software valuation. On the other hand progresses in computer vision and natural language processing make it possible to also include information contained in images and text. Geo-information systems like GoogleMaps are becoming more informative and accurate (e.g. public transport information in GoogleMaps). Hence the amount of sensor input data available for prediction (number of data points), and the dimension (pictures, text, geo-information etc.) have increased tremendously. This opens up whole new possibilities for AVMs. Our goal is to revolutionise computer valuation by the means of technological progress in machine learning and data availability.
Valuation of real estate objects is a supervised learning problem, i.e. we aim to predict a target variable P (here the price, which is also the „supervisor“) based on some input information X, e.g. age or size of the building.
The prediction accuracy can in general be improved through
1. more data points
we cooperate with online real estate platforms and data suppliers
2. more relevant input features (higher dimension)
we use images, textual information and geo-information
3. sophisticated math
IREEN is developed by scientists who work at the frontedge of technological progress in AI and statistical learning
4. prior domain knowledge explicitly modeled
we translate know-how about the real estate market to mathematics that a computer can exploit in valuation
Yes! Human appraisors should create a valuation report, if a lot of money is involved. Human appraisors are excellent in considering the peculiarities of an object. On the other hand they cannot process and take into account the patterns contained in millions of datapoints (global market) on real estate offers and transactions. We let the data talk. Both, unprofessionals and professionals on the real estate market benefit from our valuation software and market analysis tools in their decision making and bargains.
The idea of IREEN was born in february 2019. IREEN is an AI.Automatica product which has been funded and supported by the Ludwig Boltzmann Society. As of March 2020, the Project also is part of the INITS startup incubator program. Bernhard Kerres supports us in business related tasks.
PhD candidate in econometrics
PhD candidate in mathematics
PhD candidate in machine learning