Research On Land Use Planning Decision Support Model Based On Rough Set Theory And Its Application

Nowadays the common Land Use Planning Decision Support System is Spatial DecisionSupport System (SDSS) based on the combination of GIS technology and DSS technology. SDSS,which focuses on Computer Assisted Cartography (CASC) and Spatial Data Management, is notso powerful and flexible in system analysis with limited analytical functions and cannot extracthidden pattern and its rules; the logic structure and hierarchy of intelligence in SDSS cannot meetthe needs of various decisions in solving more complex land use planning, especially forunstructured problems. Therefore, new technological means is needed to solve those problems;that is, the ability of information system in analyzing, processing and planning data should beenhanced. The best choice to solve the above problems is Data Mining with the function ofknowledge discovery.The mining approach in this thesis is based on the attributes of land use planning data. LandUse Planning Information System covers more complex elements than those in other areas. Greatspatial-temporal discrepancy and variability, imbalanced economic development, locationdifferences, influence of policy factors, etc, they all decide that the rules in the process ofimplementing land use planning can both be deterministic and indeterministic and data use canboth be integrated and disintegrated, which just provides ample scope for rough set theory todisplay its merits which are more suitable to discover the information of indeterministic data.Compared with Fuzzy Method and Neural Network Method, rough set theory requires no priorknowledge of data in the process of obtaining decision rule and reasoning; the outcomes acquiredare more convenient to evaluate and interpret.Rough set theory, originally developed by Polish mathematician Z. Pawlak in 1982, is adecision analysis tool for intelligent data and is more suitable for the mining of data with anindeterministic attribute. Up to now, it has been proved that rough set theory is quite useful inpractice. Many learning systems or application systems based on rough set theory have beendeveloped internationally with a good yielding.This thesis centers on the study and discussion of how to realize its information discoveryaccording to the attribute of data in land use planning, and then proposes Land Use PlanningDecision Support Model Based on Rough set theory.Firstly, a preliminary study is given to rough set theory. A related theoretical system has beenintroduced, which is comprehensive and with clear illustrations. The overview of rough settheory's main features is quoted from Z. Pawlak's works. Also, a summary of its theoreticalfeatures has been given and the application of rough set theory in disintegrated information system has been specially probed according to its own attributes in research.Secondly, the attribute of data in land use planning have been analyzed and summarized inview of rough set theory's features. Data in land use planning is more complex; therefore, data inthis field should be classified into different types and the data types appropriate to the analysis inland use planning should be summarized according to rough set theory.The key to data mining lies in the pre-processing of data and attribute reductions in the earlierstages. Consequently, this thesis highlights the study of data pre-processing, discusses the issues ofmissing data processing and continuous attributes discretization according to the attribute of datain land use planning. Different approaches to process missing data have been suggested withcorresponding examples. Not only traditional theory and approaches have been discussed in detailsbut also optimal algorithm for discretization has been proposed.Thirdly, under its theoretical system study, rough set theory's application advantages in landuse planning have been analyzed according to rough set theory's features. Based on the generalland use scheme project in Autonomous County of Changyang Tujia Nationality, specific decisionproblems have been discussed through material and basic data collecting in the planning. Throughexplanation, the rules extracted generally reveal the interaction among the basic data, and meet therequirements in decision-making.Nowadays, knowledge discovery under rough set theory has mainly combined qualitativeexplanation and quantitative analysis in rule extraction. In order to further verify the value of newknowledge, further discussions in rule application have been done in this thesis and the appliedmethods of visualization have been brought forward precipitously.Land use planning information systems are mostly developed on the platform of GIS. Besidedaily management, their main functions are data inquiry, data storage and graph data processing.Thus, this thesis also probes into the integration of data mining system and GIS technologyaccording to land use planning. Mining rules or knowledge with rough set theory and back feedingthat knowledge into GIS system to process, which can manifest new knowledge's differencebetween before application and after application in visualized forms, and then give decisionmakers a direct impression. The above two methods have been illustrated with case studies. Whathas to be highlighted is that in the process of GIS diagramming study, the statistics analysis ofdecision rule in grain output influencing factors has effectively demonstrated the rationality ofdesigning related parameters for land natural quality attribute weight and point rules in theregulations of Farmland Grading.Lastly, two modeling plans have been put forward according to the attribute of data in landuse planning, the specific mining tasks and the study in previous two stages: one is to developdependent decision support system; the other is to develop mining module of land use planninginformation system. After a comprehensive study of development in present information systemand data mining system, together with a comparative study of the previous two plans, the firstmodeling plan has been chosen as the basis of model study in this thesis. Domain features in landuse planning have been highlighted in model design. After the plan has been determined, thestructure and function of every functional module in system has been discussed in details. Since it is just a tentative study to bring rough set theory into land use planning, therealization of system model still needs scientific demonstrations. However, this thesis has not doneany empirical research. Knowledge acquisition under rough set theory can give decision makersknowledge support, which is theoretically and practically valuable in developing decision supportsystem. Some key issues concerning rough set theory have been discussed. With a case analysis,the theory and approach presented in this thesis have proved its application value of rough settheory in land use planning. It is predictable that the application competence of rough set theorywill be greatly improved after a further research on rough set theory.