Housing Price Prediction

The objective of this paper is to evaluate the performance of the grey relational analysis in predicting the housing price for the real estate market of Taiwan. An instance-based approach which used k-nearest neighbor classifier was also applied for performance comparison. The grey relational analysis was modified to calculate the weighted synthesis of the top ten matching instances through various weighting strategies. The experimental results in this paper concluded that the grey relational analysis outperformed the instance-based approach in terms of the mean absolute error root mean square error and relative absolute error. In addition the synthesis strategy with descending weights performed better than the averaging weights during the integration process of matching instances. The result also suggested that the performance was slightly decreased if the top ten matching instances were reduced to five instances. The grey relational analysis integrated with the weighted synthesis model can assist both buyers and owners in identifying opportunities and estimating the potential risks in a worsening real estate market.