基于C5.0决策树算法的元胞自动机土地利用变化模拟模型

C5.0决策树算法可以揭示数据中的结构化信息，建立直观、易于理解的树形结构，可以用于元胞自动机转换规则的获取。提出了基于C5.0决策树算法的元胞自动机模型，分析了模型构建的理论和原理以及实现方法。在此基础上，以杭州市城市用地演变为例，利用C5.0决策树算法从已有的城市用地及其影响因子数据中挖掘出城市用地的演变规则，并将获得的转换规则应用到元胞自动机模型中进行城市用地演变的动态模拟与预测。结果表明：（1）利用C5.0决策树算法获取转换规则可以在较低采样率的前提下保证具有较高的精度；（2）与其它获取元胞自动机转换规则的方法相比，C5.0决策树算法生成的规则清晰明确，很适合用来获取元胞转换规则；（3）基于C5.0决策树算法的元胞自动机模型在模拟城市用地演变方面具有较高的模拟精度，模拟结果可靠.

LAND USE CHANGE CA MODEL BASED ON C5.0 DECISION TREE

C5.0 Decision Tree is capable of discovering structural information in data,and building tree intuitively and intelligibly,thus it can be used to obtain transformation rules of cellular automata(CA).In this paper,we gave out a CA Model based on C5.0 Decision Tree,analyzed theories and principles,and build up method of this kind of models.And then,taking city land use change of Hangzhou City as a case,we got land use change rules from historical land use data and its factors by using C5.0 Decision Tree.Combining this rules and CA model,we simulated and predicted land use changes in Hangzhou City.Results show:(1)by C5.0 Decision Tree,we could get more accurate rules under lower sampling rate;(2)comparing with other rules getting methods,rules are clear and explicit in C5.0 Decision Tree,so it’s suitable to get CA rules;(3)we could get higher accuracy and more reliable result by using land use change CA Model based on C5.0 to simulate land use change.