Real Estate Market Analysis For Investment and Buying Cycles Prediction

Maria Pereira Leitão

Abstract:The traditional methods used to estimate the Real Estate prices are sometimes too subjective and lack accuracy. The most common approaches to calculate a property’s price are the Market Approach, the Income Approach, and the Cost Approach. Artificial Intelligence is applied to house price prediction, and Machine Learning models are developed and tested to research the best algorithm to achieve better accuracy results to overcome the subjectivity carried by these methods. This project provides some background on how the Real Estate market functions and how some State-of-the-Art solutions address the industry’s requirement of Artificial Intelligence. It also describes a few experiments on several algorithms to understand how adequate they are in the scope of the problem, either by trying to achieve a precise duplicate detection model, or by aiming to develop a model capable of computing Real Estate values. In those experiments it was found that it is of value to separate the dataset by location, creating subsets of data. Also, from the several algorithms tested, the majority of subsets achieved better results with Random Forest and Gradient Boosting.

Computer Science,Business,Economics