PREDICTION OF REAL ESTATE PRICES WITH DATA MINING ALGORITHMS

Abstract

In real world, the amount of data has increased in many markets such as real estate, sport, technology etc. In this sense, it is difficult to manage and analyze the data manually. Data mining is a field that enables to understand large amounts of data and make improvements on it. Estimation and evaluation of the data sets have been studied by various data mining algorithms. This study aims at implementing and comparison of data mining algorithms on real estate price prediction. The data is obtained from the University of California Irvine (UCI). This model takes into account the transaction date, house age, distance to the nearest MRT station, number of convenience stores in the living circle on foot, and geographic coordinate information. Different data mining algorithms including random forest, gradient boosting and linear regressor have been trained on real estate data for pricing house. These prediction models have been built by taking different size of test set. Root Mean Square Error (RMSE), Mean Square Error (MSE) and Mean Absolute Error (MAE) metrics are used in the measurement techniques. The best result was obtained by gradient boosting regression when the amount of test set is 20%, and the mean absolute error for this method was 3.92.