An Artificial Intelligence Model for Predicting Real Estate Contract Cancelation based on Naive Bayesian Classification: A Case Study of Apartment Sales in Seoul Metropolitan Area

Real estate sellers aim to sell properties at higher prices than they did previously. For this reason, some real estate agents and their associates have been found to inflate asking prices and publish forged contracts. As a result, as of early January 2021, the Korean government has begun to retain the details of actual real estate transactions for cancelled sales contracts in the existing price disclosure system to prevent further increase in fake and false real estate offerings. In this paper, we propose an artificial intelligence model based on Naive Bayesian classification to predict whether real estate transactions will be canceled. The proposed approach was trained with transaction data from 35,649 apartment sales in the Seoul metropolitan area from January to August 2021. Experimental result show that the average precision of the proposed research model was 0.9177, and the those of the other validation indices were at least 0.5674. We expect this work to be useful in developing services and methodologies that can filter out fake real estate contracts in advance.