Real estate appraisal plays a vital role in people’s daily life. People rely on
the estimation of decisions on buying houses. It is well recognized that the criminal
activities around the house have significant impacts on house prices. House buyers can
make more reasonable decisions if they are aware of the criminal activities around the
house. Therefore, a machine learning-based method is proposed by combining house
attributes and criminal activities. Specifically, the method firstly infers the intensity of
criminal activities from historical crime records.Then, a novel machine learning
algorithm, extremely randomized trees (ExtraTrees), is implemented to estimate the
house price based on the extracted comprehensive crime intensity and real-world house
attributes.The experimental results show that the proposed method outperforms
contemporary real estate appraisal methods by achieving higher accuracy and
robustness.
Keywords: Ensemble learning, Extremely randomized trees, Machine learning,
Real estate appraisal.