|
In this study, using 60 years of data of the National Bureau of Statistics national cotton area, yield and total production, 16 years of the eight provinces in cotton output data, and survey data for Cotton Research Institute, Chinese Academy of Agricultural Sciences. Research methods: First, by using Excel on the country's 60 years of cotton acreage, yields and total production data normalization and the overall analysis. Second, the application of SAS V8.0 software of the survey data, do data processing and analysis, the establishment of income, with the output value of the model. Sequence analysis of the country's total cotton output, and the time of the eight main cotton-producing provinces. Fourth and are relatively simple regression model analysis, multiple regression model, ARIMA model and combination model advantages and accuracy of yield prediction, the country with eight cotton-producing provinces total cotton production forecast, and trends in production and the actual yield differences were analyzed to dig implicit memory inherent within each series of data to provide a scientific basis for decision-making for cotton. The key findings are as follows: 1, the the farmers output value of the model output value equation to the total cost of cotton production, seed cotton prices affect factor composition equation. SAS analysis results: Prob gt; F has a value of less than 0.0001, belongs to a very significant and fitting (R-Square) to 0.9896 can use the equation for multiple regression model, and the results to reach a significant level. Equation Y = -1895 0.0004763 * X1 5.12993 * x2 3700.9362 * X3. 2, the total national cotton production forecast model (1) ARIMA time series the residual series test of the model can be drawn are greater than 0.05, Chi-Square, that take full advantage of the sequence data and the residual series test was significant, using the model 2010-2014 cotton output forecast 2011-12 cotton output in 2015 were 6.49 million, 6.81 million, 6.68 million , about 6.7 million and 6.89 million tons. (2) a simple regression yields prediction model, SAS results show that the regression equation is very significant, the parameters Pr gt; F value is less than 0.0001, reached a significant level. Have model Tt = 21.87025 10.16873t, to t = 61, 62, 63, 64 are orthogonal to and 65 into National Cotton trend in the next few years, the total output of 6.42 million, 6.52 million, 6.63 million, 6.73 million and 6.83 million tons. (3) multiple regression model, the formula: = yield * area of ??the total production, the establishment of the regression equation: Y = aX1 * X2 c (X1 yields the the X2 area of ??a adjustment factor, c is a constant), the regression model Y = 1.01285 X1 * X2-3.36129, R-Square = 0.9973, Pr gt; F is less than 0.0001, highly significant. National Cotton trend in the next few years the total production of 6.77 million, 6.78 million, 6.8 million, 6.82 million and 6.84 million tons. (4) a combination of the model by weighting formula to calculate the ARIMA model forecasts, the weight of a simple regression model and a multiple linear regression model, substituting a combination of model formula, the combination forecasting model for on Y = 0.364y1 O.041y2 0.595y3 (y1 y2, y3 are The ARIMA models predict a simple regression model to predict the results of multiple regression model to predict results), the National Cotton trend in the next few years the total output of 665,678,674,677 and 686 million tons. (5) The main conclusion: the next five years, the country's total cotton production in Q of 642 to 689 tons, an average of 6.76 million tons, slightly reduced with a five-year average of 6.95 million tons. Compare four models, four models were cut over the last five years, the combination forecasting model, representing a decrease of 2.7%; ARIMA model, a decrease of 3.4%; multivariate regression model, a decrease of 2.2%; simple regression model, a decrease of 4.6% . 3, the application of the ARIMA model of the main cotton producing provinces and regions of cotton production forecast Xinjiang cotton production forecast equation: △ X3t =, 330.85595 0.60203Xt 1 (?) T .75124 (?) 0.2483 (? T-1) t-2, (?) t-WN (0, σ2 (?)) Shandong total cotton production prediction model: Xt = 38.24298 0.20002Xt-?-0.26736Xt-2 0.53262Xt-3 (?) T 0.50082 (?) T-1 ( ?) t to WN (0, σ2 (?)) Hebei total cotton production forecast model: ΔX2t = 53.26862 0.39796Xt-1 (?) t 0.28533 (?) t-1 (?) t ~ WN (0, σ2 ( ?)) total cotton production forecast model: Xt = 688.50018 0.04804Xt-0.17018Xt-2-0.11307Xt-2 (?) t ~ WN (0, σ2 (?)) Hubei total cotton production forecast model: Xt to = 4.20328 0.46822Xt-1 0.53178Xt (?) t-.09005 (?) t-1 (?) t ~ WN (0, σ2 (?)) total cotton production forecast model: Xt = 205.3145 (?) t-0.59361 (?) t-1-0.63806 (?) t-2 0.37404 (?) t-3, (?) t ~ WN (0, σ2 (?)) Anhui total cotton production forecast model: Xt = .6335- 0.755.1Xt-1-0.4004Xt at-2 (?) t ~ WN (0, σ2 (?)) Jiangsu total cotton production forecast model: Xt = 41.63724-0.83734 (?) t-0.24745 (?) t-1- 0.68303 .30283 (?) t-2 (?) t-3 (?) t ~ WN (0, σ2 (?)) simulation results pointed out that the next few years, the total production of cotton in Xinjiang, Shandong, Hebei, Hubei and Jiangsu an upward trend, in which the speed of Shandong rise faster, Xinjiang and Hebei rises slowly rising trend of total output in Hubei and Anhui slowest; Jiangsu fluctuations. Hunan total production trend remains unchanged; Henan Province, showed a downward trend.
|