Research On Real Estate Price Batch Evaluation Based On Deep Neural Network

General secretary Xi Jinping put forward the requirement of "no speculation in housing" at the 19 th national congress.under this background,the real estate industry is still developing rapidly and there are a large number of investment and consumption transactions.Therefore,for the healthy development of the real estate industry,fair trade cannot be separated from objective and fair prices,which need asset evaluation to guarantee.In the evaluation practice,the market method,the income method and the asset-based method all have some restrictions,which are not applicable in some cases.Moreover,many influential aspects are determined subjectively by appraisers.For the final evaluation,such subjectivity is crucial,so it is difficult to define whether the evaluation is accurate.At the same time,due to the large number of commercial real estate transactions,if one by one assessment,the workload is relatively large compared with that of appraisers,accompanied by lower assessment efficiency,in addition to which the assessment cost is relatively large.Therefore,in addition to the traditional evaluation methods,we need to explore more scientific and reasonable evaluation methods to evaluate the real estate value in batches..On the basis of relevant documents and materials,this paper uses theories and empirical methods,qualitative and quantitative methods to study and demonstrate machine learning in real estate batch evaluation.First of all,it introduces the important value and significance of the topic selection under the macro background of the continuous development and expansion of the real estate industry,Secondly,it introduces the concept and relevant theories of batch evaluation,as well as real estate evaluation procedures and methods,such as multiple regression,characteristic value method,in-depth learning method,etc.Then,the important role of in-depth neural network in in-depth learning,the establishment of in-depth neural network model and the optimization of the model are deeply discussed.Finally,using the linked transaction data,this paper uses the deep neural network model to make an empirical analysis of real estate batch evaluation.Compared with other methods for batch evaluation of real estate,this paper uses sufficient data and optimizes the depth neural network to obtain better real estate price prediction effect.It is found that the depth learning model has the advantages of high efficiency and accuracy in batch evaluation of large data.Based on the theoretical research results,this paper holds that: first,the upgrading of hardware facilities and the exploration of various in-depth learning models will make the batch evaluation of real estate prices more accurate and faster;Second,with the research of evaluation theory and further improvement of the influencing factor system of house price,real estate evaluation will be closer to the market price and the prediction effect will be better.Therefore,the construction of a complete evaluation index system and the establishment of a database will be the direction of future research on batch evaluation.The research in this paper provides a theoretical method for real estate batch evaluation under the background of big data in the new era of our country.However,in the evaluation practice,most of the data used in real estate enterprise pricing are from the sales department.How to combine the theoretical model characteristics with the pricing strategy in the market is worth further research.