Predicting housing price in China based on long short-term memory incorporating modified genetic algorithm

Predicting the future trend and fluctuation of housing price is an important research problem of housing market. The machine learning approach is rarely used in existing studies, while the traditional prediction models have strict requirements on input variables and are weak in solving nonlinear problem. To overcome the problems of traditional models, a long short-term memory (LSTM) approach is proposed to predict the housing price of a city by using historical data. The proposed LSTM incorporates a modified genetic algorithm with multi-level probability crossover to select appropriate features and the optimal hyper-parameters. The data of housing price and related features of Shenzhen, China, from year 2010 to 2017 have been used to test the performance of the model. The results indicate that the proposed method has good performance in modeling housing price and is obviously outperforms other algorithms including back propagation neural network, support vector regression and different evolution LTSM. Therefore, this proposed model can be used efficiently for predicting housing price and thus can be a good tool for policy makers and investors to monitor the housing market.