农业经济增长影响因子的偏最小二乘回归分析——以宁夏回族自治区为例

ANALYSIS OF THE RELATIONSHIP BETWEEN AGRICULTURE ECONOMYDEVELOPMENT AND INFLUENCING FACTORS BY PARTIAL LEAST-SQUARE REGRESSION (PLS)——A CASE STUDY OF NINGXIA HUI AUTONOMOUS REGION IN CHINA

关键词:  农业经济 影响因子 偏最小二乘回归法 宁夏

Key words:  agriculture economy  influencing factors  Partial Least-Square Regression (PLS)  Ningxia

摘要:

偏最小二乘回归分析方法具有主成分分析、典型相关分析和线性回归分析等方法的特点，因此其能够较好地解决变量间存在的多重共线性问题。宁夏是我国西北地区重要的农业省区之一，农业在其国民经济中占有重要地位。文章以宁夏回族自治区为研究区域，运用偏最小二乘回归分析方法，构建了宁夏农业经济增长影响因子的偏最小二乘回归模型。通过对所得模型检验，模型对X和Y的解释能力较强，分别为0977和0999，t1/u1平面图表明，所计算的t1、u1能够充分地解释变量集与因变量集的信息，自变量集与因变量集之间存在明显的线性关系，并且拟合优度R2=0983 1，回归模型精度较高，可靠性较强。研究结果表明，所选取的各项指标中，农业产值、化肥施用量、渔业产值、财政支农支出和牧业产值等是影响宁夏农业经济增长的主要因素。因此，宁夏应加快农业产业结构调整，推进农业产业化进程；加大农业科技投入力度；增加财政支农的力度，提高支农资金的使用效率；加大农业基础设施建设，加快农业剩余劳动力的转移。

Abstract:

Partial Least-squares (PLS) regression is a statistical tool and has characteristics of principal components analysis method, canonical correlation analysis and linear regression, so it can deal with multiple regression problems when the number of observations are limited and correlations between variables are high. Meanwhile, PLS regression reduces the dimensionality of the regression by using fewer PLS factors than the number of independent variables. Ningxia is one of the conventional agriculture provinces in Northwest China. Due to natural and historical reasons, the social economic development of Ningxia shows marked difference between the north and the south, the northern part of Ningxia is the plain area of the Yellow River and the south is a mountainous area. Taking Ningxia as the object, this paper constructed a Partial Least-Square Regression (PLS) model with agriculture economy development and influencing factors based on data of Ningxia from 2002 to 2011. The test results showed that the model could explain the X and Y, of 0977 and 0999, respectively; the t1/u1 plane graph showed that the calculated t1,u1 can adequately explain the set of variables and dependent variables of information, variable set and the dependent variables were obvious linear relationship and the goodness of fit R2=0983 1. PLS Regression had high precision and reliability through test. The result showed that agricultural product, consumption of chemical fertilizers, gross output of fishery, expenditure for supporting rural production and gross output of animal husbandry were factors influencing agricultural economic growth. In the future, Ningxia should actively adjust the planting structure, speed up the development of forestry, animal husbandry, fishery and downstream processing industry of agricultural products, increase scientific and technological content of agriculture products, increase the financial investment to agriculture, strengthen the farmer labor skills training, and accelerate the transfer of surplus agricultural labor force.