| 1. | They are agricultural productive materials price growth rate , sown area of grain crops growth rate , grain yield per area growth rate - , natural disaster covered grain areas growth rate , net grain import change rate , grain reserve change rate , population growth rate , per income growth rate , city and town population growth rate , food industry production value growth rate , year - end pig number growth rate , medical & pharmaceutical and textile industry production value growth rate , grain marketization degree , inflation rate using the previous year as base year ( preceding year = 100 ) , public grain purchases price growth rate , investment in agricultural science and technology growth rate , investment in agricultural infrastructure growth rate , growth rate of graduates number from agriculture , forestry , science & technology universities and colleges and specialized secondary schools , government expenditure for agriculture and agricultural credit growth rate , international grain price growth rate , rmb exchange rate growth rate , last grain price growth rate , economic crop price growth rate , , meanwhile , a new method is attempted to be used in this paper and the grain price early - warning problem is transformed into machine learning problem by introducing statistic learning theory and svm method which are gaining popularity in machine learning field at present in the world
在此基础上，筛选出23个警兆指标：农用生产资料价格增长率、粮食播种面积增长率、粮食单产增长率、粮食受灾面积增长率、粮食净进口量变化率、粮食储备变动率、人口增长率、人均收入增长率、城镇人口增长率、食品工业产值增长率、猪年末头数增长率、医药纺织工业产值增长率、粮食市场化程度、以上年为基年的通货膨胀率、国家粮食定购价格增长率、农业科技投入增长率、农业基础设施投入增长率、农、林、科技高校大、中专毕业生人数增长率、财政支农资金比重及农业信贷增长率、国际粮食市场价格增长率、人民币汇率增长率、上期粮食价格增长率、经济作物价格增长率。同时论文在预警方法上作了新的尝试，把粮食价格预警问题转换成一个机器学习问题，引进当前国际上机器学习领域中比较热门的统计学习理论和支持向量机方法，用顺序回归算法对历史数据进行学习建立了粮食价格预警模型。 |
