局部搜索算法在动态建立基准地价模型中的应用

传统基准地价评估中将地价样本检验剔除与回归模型建立分离 ,不能获取最优的基准地价评估模型。应用求解组合优化问题的局部搜索算法 ,将地价样本检验与模型建立有机结合 ,通过邻域搜索来实现地价样本组合和地价模型的动态优化。对算法的关键步骤进行了设计 ,通过合理的邻域函数和约束初始解设置 ,克服了算法在搜索过程中陷入局部最小的不足。试验结果表明 ,在样本置信水平相近的情况下 ,应用局部搜索算法计算的基准地价评估模型优于传统方法的建模结果。

Application of Neighbor Search Algorithm in Modeling Land Benchmark Price

An optimized Land Benchmark Price model can't be got by traditional method because sample's testing and regression analyzing are separated.Thus,a new method based on Neighbor Search Algorithm is brought forward to combine the processes of sample' testing with Modeling dynamically,througe the sample's combination will be optimized and the model will be globally optimal.Moreover, a serie key processes,such as the target function,initialization,neighborhood function and state subrogation are designed.And the deficiency of algorithm is overcome that part-optimal result will disappear during the period of search .A better result than traditional method was got by an instance in the end ,which has been applied in several projects and proved reasonable.