Forecasting house price index of China using dendritic neuron model

The result of Chinese housing market continues to prosper or not is related to the development of China, and further it also has an impact on the world finance. Thus forecasting the house price index is very important and challenging. In this paper we propose an unsupervised learnable neuron model (DNM) by including the nonlinear interactions between excitation and inhibition on dendrites. We use DNM to fit the House Price Index (HPI) data and then forecast the trends of Chinese housing market. To verify the effectiveness of the DNM, we use a traditional statistical model (i.e., the exponential smoothing (ES) model) to make a performance comparison. Three quantitative statistical metrics including normalized mean square error, absolute percentage of error, and correlation coefficient are used to evaluate the forecasting performance of the two models. Experimental results demonstrate that the proposed DNM is better than ES in all of the three quantitative statistical metrics.