基于随机森林模型的功能区特定地物识别研究

Study on Recognition of Specific Features in Functional Areas Based on Stochastic Forest Model

针对现阶段基于POI数据的城市功能区划分受限于各类POI数量差别过大易出现空间耦合现象而导致部分地物无法识别的问题,本文提出一种基于人口热力数据的特定地物识别方法。以公园景点类地物为例,本文讨论人口热力数据对应公园景点类地物的时空关系特点,基于研究区域的规划图数据进行特征优选,通过随机森林模型的训练与学习来进行公园景点类地物的识别,经过精度检验后准确率达到86%,具有良好的可行性。

In order to solve the problem that the urban functional area division based on POI data is limited by the difference of different POI numbers,it is easy to appear spatial coupling phenomenon,which leads to some features can not be identified.This paper proposes a specific feature recognition method based on population thermal data.In this paper,the characteristics of the spatiotemporal relationship between the thermal data of population and the spatial-temporal relationship of the landscape features in the park are discussed.Based on the planning map data of the study area,the feature optimization is carried out.Through the training and learning of the random forest model,the recognition rate of the park scenic spots is 86%.the method is feasible.