Urban Residential Land Price Appraisal via Quantifying Impact Factors Based on Deep Belief Networks
Hua Ai, Qiang Liu, Yuxin Jiang, Jing He
Abstract
The relationship between urban residential land prices and the explanatory variables is highly complex. This causes that it is difficult to quantify the impact factors for appraising urban residential land prices. This paper explores the use of Deep Belief Networks for quantifying impact factors of urban residential land prices. The proposed approach applies grid cells to express the samples finely 458 features are extracted as input from collected raw data of 37 important impact factors, residential land prices are divided into 9 levels as output, and both BBRBM and GBRBM are utilized to form the network. A deep belief network model is finally employed to appraise urban residential land prices via their impact factors, and the average accuracy can achieve about 90%.