A New Method Based on Association Rules Mining and Geo-filter for Mining Spatial Association Knowledge

Association rule mining methods, as a set of important data mining tools, could be used for mining spatial association rules of spatial data. However, applications of these methods are limited for mining results containing large number of redundant rules. In this paper, a new method named Geo-Filtered Association Rules Mining(GFARM) is proposed to effectively eliminate the redundant rules. An application of GFARM is performed as a case study in which association rules are discovered between building land distribution and potential driving factors in Wuhan, China from 1995 to 2015. Ten sets of regular sampling grids with different sizes are used for detecting the influence of multi-scales on GFARM. Results show that the proposed method can filter 50%–70% of redundant rules. GFARM is also successful in discovering spatial association pattern between building land distribution and driving factors.

Rules of Pesticide Use for Food Crop Diseases and Pests: A Concise Diagram Based on Correspondence Analysis

In order to ensure food security, the selection of pesticide must take into account the controlling effect and economy in the chemical control of food crop diseases and pests. Although China has registered a large number of pesticides and fungicides on food crops, traditional statistical methods cannot treat the complex relationships between crops, pests and diseases, as well as pesticides. To this end, the correspondence analysis was used to mine the data of pesticides being registered after January 1, 2010 for the prevention and control of pests and diseases on wheat, corn, and rice in China, in order to reveal the general rules of pesticide use for the prevention and control of grain crop pests and diseases and to provide scientifically recommended pesticide regimens.(1) Recommended pesticides for wheat diseases and pests: For underground pest control, phoxim, diazinon or clothianidin is recommended;imidacloprid is recommended for aphid control;avermectin is recommended for mite control;chlorpyrifos is recommended for midge control;the pesticides for rust control could be selected according to the order of epoxiconazole>flutriafol>kresoxim-methyl>cyproconazole≈hexaconazole;tebuconazole is recommended for the prevention and control of smut;difenoconazole is recommended for the simultaneous occurrence of take-all, smut and sheath blight;it is recommended to use carbendazim or thiophanate-methyl for the control of scab;and triazolone or pyraclostrobin is recommended for the control of powdery mildew.(2) Recommended pesticides for corn diseases and pests: For the control of underground pests, fipronil, carbosulfan or chlorpyrifos are recommended;thiamethoxam is recommended for control of planthoppers and corn aphids;Bacillus thuringiensis or phoxim is recommended for control of corn borers;pyraclostrobin is recommended for the control of northern leaf blight;tebuconazole is recommended for the control of head smut;and fludioxonil+metalaxyl-M is recommended for the control of corn stalk rot.(3) Recommended pesticides for rice diseases and pests: It is recommended to use triazophos, avermectin analogs or methoxyfenozide for the control of borers;for rice planthoppers, the pesticides could be selected in order of pymetrozine>imidacloprid>thiamethoxam>thiazide>nitenpyram;avermectin analogs are recommended for the control of rice leaf rollers;tebuconazole is recommended for the control of rice false smut;prochloraz and its manganese salts are recommended for the control of bakanae disease;hymexazol is recommended for the control of wilt disease;thifluzamide or hexaconazole is recommended for the control of sheath blight;and the pesticides for the control of rice blast are in order of tricyclazole>kasugamycin>fenoxanil>isoprothiolane>Bacillus subtilis. In addition, potential applications of correspondence analysis in other fields of plant protection are discussed.