Research And Application Of Farmland Management Zones Generating Based On Boundary Sensitive C4.5Algorithm

China is a large agricultural country, how to increase food production has been amajor concern. The concern directly affects the level of fertility of arable crop growth,yield and quality. Fertile land is an important material basis for ensuring sustainabledevelopment of agriculture. For traditional farming, people divide the land for acertain area, and the mean soil nutrient of the whole field represents the status of thesoil of whole farm nutrient. Then land on the entire amount of fertilizer to use a singleuniform fertilization. However, the field soil and the characteristic of the crops are notuniform, thus with time the spatial variation and the configuration using a singlefertilizer demand reach both fertilization requirement is more likely to result in wasteof resources and environmental pollution. Consequence the method of precisefertilization emerged. Precision fertilization refers to the land based on soil nutrientchanges prefer to be divided into different management zones, and each managementarea is divided for the use of its specific fertilization program. The goal of the methodis to reduce consumption by the premise of protecting the environment and access tothe best harvest. Traditional variable fertilization obtains information primarilythrough the method of grid sample. However, the high cost of this method is notpractical in recent years. Many scholars began to stress on the issue according tovariability of soil nutrients and space divided into different plots in the same relativelyhomogeneous area management, soil nutrient management zones. Scientific andrational management of soil nutrient partitioning implementation of precisionagriculture technology is an efficient means of variable rate fertilizer and precisionagriculture at home and abroad to become a research hot spot.With the rapid development of science and technology in today’s world, thereare more and more new technologies and new methods could be used in themanagement area divided in domestic and international management zoning. Themainly empirical methods are classification methods, GIS software and statisticalmethods. Some scholars also exploit data mining technology into the division offarmland management area, using a variety of data mining algorithms to solveclassification problems management areas, such as k-means clustering algorithm,fuzzy c-means clustering algorithm, the weighted fuzzy clustering algorithm, particle swarm optimization improved ant colony algorithm and clustering algorithm, anddecision tree classification method. This paper elaborates the decision tree algorithmfor division of farmland management area. The main work and the results of thisstudy are listed:(1)Respectively, using a decision tree algorithm ID3algorithm and C4.5algorithms NongAnXian fertility data management area divided by the experimentalresults to prove that nomatter ID3algorithm or C4.5algorithm are both able to playNongAnXian management zoning guidance. And C4.5algorithm soil nutrient data ascontinuous data processing can be more accurate analysis of the main characteristicsof each district soil management and can achieve the target of precision division ofmanagement.(2)Based on the research of ID3and C4.5algorithms comparison of the proposedboundary, sensitive C4.5algorithm is raised. C4.5algorithm is sensitive borderimproved by algorithm C4.5, C4.5algorithms rely solely on traditional informationgain rate selection decision point, classification based on a single, easily influenced bythe number of categories of samples. C4.5algorithms boundary sensitive informationwill be introducing the sample into the spatial structure of the classification process tosolve the C4.5classification algorithm based on a single, vulnerable to the impact ofthe number of samples shortcomings category.(3)The boundaries of C4.5algorithms are sensitive test data and the imbalance inthe balance of the data. C4.5algorithm proved sensitive border this classificationmechanism effectively reduces the classification accuracy of data loss due to theimbalance caused by the introduction of spatial structure information and to have theability to handle imbalance data.(4)The boundary sensitive C4.5algorithm NongAnXian fertility datamanagement zoning, and the experimental results were obtained with the results ofthe expert division, the experimental results of C4.5algorithms compare theexperimental results prove that the sensitive border C4.5algorithm is able to managezoning NongAnXian play a guiding role, and can further improve the classificationaccuracy of C4.5.