An agricultural land benchmark land price evaluation method based on depth learning algorithm

Abstract

The invention provides an agricultural land benchmark land price evaluation method based on depth learning algorithm. The method comprises the following steps: collecting market transaction data and constructing a sample data set of land price; using z-score method normalizes the feature data of samples and dividing the original sample data set into training sample set and test sample set randomly; constructing a benchmark land price evaluation model of agricultural land was constructed by using the depth confidence network algorithm, and the parameters of the deeper network structure with thehighest fitting precision were preserved by training and learning samples based on the deeper network structure; inputting eigenvalues of the agricultural land evaluation unit into the trained deep network structure to calculate the land price of the evaluation unit; using total score frequency method to demarcate the evaluation unit level, and selecting an area weighting method to calculate thebenchmark land price. The invention can establish the mapping relationship between land price and land price influencing factors with high fitting precision, and the obtained datum land price and agricultural land quality maintain good consistency in spatial distribution law.

Description

technical field [0001] The invention relates to the technical field of benchmark land price evaluation of agricultural land, in particular to a method for evaluating benchmark land price of agricultural land based on a deep learning algorithm. Background technique [0002] The agricultural land appraisal (ALA) work is an important work carried out by China to promote the deepening reform of the rural land use system. local standard price system. The evaluation of the benchmark land price of agricultural land is of great significance for the smooth development of rural land management work such as the transfer of land contractual management rights, land acquisition compensation, land consolidation, and rational allocation of land assets. [0003] The methods commonly used in China’s land price evaluation work include arithmetic mean model and regression model: the arithmetic mean model takes the average value of sample land prices in a homogeneous area to determine the bench...

AI Technical Summary

Problems solved by technology

[0005] Aiming at the technical problems existing in the existing models, such as strong subjectivity and the inability of the shallow network to represent the complex functional relationship between the influencing factors and the benchmark land price of agricultural land, the present invention proposes a method for evaluating the benchmark land price of agricultural land based on a deep learning algorithm. Using the classic algorithm in the deep learning method - deep belief network (deep belief network, DBN) to construct the agricultural land reference land price evaluation model, using the deep network structure of the DBN algorithm to more accurately represent the influencing factors and the agricultural land reference land price The complex functional relationship between them improves the rationality of the benchmark land price of agricultural land

Images