Batch evaluation of collective owned commercialised construction land using machine learning

Area of Science:

• Agricultural and Economic Sciences
• Land Economics
• Machine Learning Applications

Background:

• China's rural land system reform includes the market entry of collective owned commercialised construction land (CCCL).
• Traditional appraisal methods lack efficiency and accuracy for batch appraisals of CCCL market-entry prices.

Purpose of the Study:

• To develop a machine learning-based batch appraisal model for CCCL market-entry prices.
• To enhance the efficiency and precision of CCCL price prediction.
• To identify key factors influencing CCCL prices.

Main Methods:

• Implementation of three machine learning models: Random Forest (RF), Back Propagation Neural Network (BPNN), and Support Vector Machine (SVM).
• Development of a tailored indicator system for price prediction in Beiliu City, a reform pilot area.
• Comparative analysis of model performance based on prediction accuracy and mean absolute error.

Main Results:

• The Random Forest (RF) model demonstrated superior performance with a prediction accuracy of 94.77% and a mean absolute error of 17.50 yuan.
• RF outperformed BPNN (91.21% accuracy) and SVM (91.94% accuracy).
• CCCL prices are influenced by factors like township economic levels and market entry methods, exhibiting unique characteristics.

Conclusions:

• Machine learning models, particularly RF, are effective for batch appraisal of CCCL market-entry prices.
• The study provides a scientific basis for standardizing the land market and guiding policy formulation in rural land reform.
• Accurate CCCL price prediction is crucial for efficient land market operations and reform.