Keywords: total energy consumption; Linear regression; Gradual regression; Residual analysis; * * * Linear diagnosis
First, the background of the topic
Energy is one of the material bases for the progress of human society. The emergence of two oil crises has aroused more and more attention to energy supply. In recent years, with the development of industrialization, countries around the world consume more and more energy, and the energy crisis is becoming more and more serious. Therefore, it is particularly important to analyze the factors that affect energy consumption. China is the largest developing country in the world, and it is also a big producer and consumer of energy. Energy output is second only to the United States and Russia, ranking third in the world; The consumption of basic energy accounts for110 of the world's total consumption, ranking second in the world after the United States. Since 1993, China has changed from a net exporter of energy to a net importer. The total energy consumption has exceeded the total supply, and the external dependence of energy demand has risen rapidly. Energy sources such as coal, electricity, oil and natural gas are in short supply in China. Therefore, effectively grasping the factors that affect the total energy consumption is beneficial for the government to formulate relevant economic development strategies and energy-saving policies, which is of great significance for realizing and perfecting Scientific Outlook on Development and is of great benefit to people's daily life.
Second, the introduction of data sets.
The data in this paper are all from the website of China National Bureau of Statistics, and the relevant data from 199 1 to 20 10 are selected for analysis (see appendix for details). According to the availability of relevant information and data, the indicators affecting the total energy consumption in China can be summarized as the following structure table.
Third, the establishment and test of regression equation
1, global significance test of regression equation
Firstly, the regression equation is established, and the overall significance test is carried out by R software. The results are as follows.
As can be seen from Table 2, the overall test P value of the regression equation is very small, and the regression effect is very significant. At the same time, the regression model's goodness-of-fit judgment coefficient r and the revised judgment coefficient are 0.9926 and 0.99, respectively, which shows that the fitting degree is high and the fitting effect is good. However, in the significance test of regression coefficient, when the significance level is α=0.05, the coefficient x 1 fails the test, which shows that if all variables are selected to construct the equation, the effect is not good, and it is necessary to select variables to establish the "optimal" regression equation.
2. Establish regression equation by stepwise regression, forward regression and backward regression.
Stepwise regression analysis was performed on the analysis data set with R software, and stepwise regression was performed with function step ().
The regression result is:
We can find that the equation obtained by stepwise regression method has high overall significance and good fitting degree. The variables finally determined are x2- consumption level of residents (100 yuan), x3- proportion of hydropower, nuclear power and wind power (%), x4- average passenger distance (km), and X5- purchase price index of raw materials, fuel and power. Under the condition of significance level α=0.05, the regression coefficients of all selected items have passed the test, which has high significance. Therefore, the regression equation is obtained:
The stepwise regression results made by R software The regression equations obtained by forward regression method and backward regression method are the same and will not be described in detail.
Fourth, the diagnosis of regression equation
Next, we will further study some characteristics of the regression model to test whether there are abnormal values that bring obvious instability to the regression model.
1, residual analysis
Use R software to carry out residual analysis on the obtained optimal regression equation, and draw a standardized residual diagram as follows.
Figure 1 standardized residual diagram of regression equation
The data in the graph are random scattered points, and there is no abnormal trend. It is a normal residual graph, and the regression equation is stable. Because in general, the observed values with the absolute value of standardized residuals greater than or equal to 2 are considered as suspicious points, and the observed values greater than or equal to 3 are considered as abnormal points. As can be seen from Figure 1, these 20 points are basically distributed among each other or slightly deviated from each other, but they are not obvious. Therefore, we can say that the optimal regression equation has no obvious abnormal value and the optimal regression equation is stable.
2, * * * linear diagnosis
If the independent variables in the regression equation have linear or approximate linear relationship, the significance of the variables will be hidden, the error of parameter estimation will increase, and a very unstable model will be produced. The optimal regression equation is diagnosed linearly by R software, and the variance expansion factors of independent variables are as follows.
Generally speaking, if the variance enlarges the factor, it shows that the model has a strong * * * linear problem. From the results in Table 4, it can be seen that the variance expansion factor of the independent variables in the optimal regression equation does not exceed 10, so it can be explained that all the independent variables in the regression equation do not have serious * * * linearity problems, and the regression equation has good effect and stability.
Interpretation and analysis of the best regression equation of verb (verb abbreviation)
To sum up, after establishing regression equation and making regression diagnosis, the final optimal regression equation is:
From this model, it can be seen that the consumption level of residents, the proportion of hydropower and nuclear energy, the average distance of passenger transport and the purchase price index of raw materials are the best factors affecting the total energy consumption in China.
First, the improvement of residents' consumption level will promote the growth of energy consumption. The total energy consumption will increase by 32.05 million tons of standard coal for every increase in residents' consumption level of 100 yuan. Because people's living standards have improved, they can increase their expenditure on energy consumption, such as the use of vehicles such as cars and the use of electrical appliances such as air conditioners and freezers. It can be said that economic growth and energy consumption are inseparable. As far as the development trend and economic growth of China are concerned, it will remain in a state of high energy consumption for a long time to come.
The proportion coefficient of hydropower, nuclear power and wind power is negative, which shows that with the increase of the proportion of hydropower and nuclear power, energy consumption will decrease to a great extent, that is, for every percentage point increase in the usage of hydropower, nuclear power and wind power, the total energy consumption will decrease by177.4 million tons of standard coal. It can be seen that increasing the research and development and use of clean energy plays a vital role in saving the consumption of non-renewable energy in China. At the same time, we can't forget that the use of new clean energy will also bring great relief to our increasingly serious environmental pollution problem.
In recent years, with the increase of population and the improvement of people's living standards, people's travel frequency has gradually increased. However, the increase of average passenger distance will increase the energy consumption of vehicles. The energy consumption of 949,000 tons of standard coal will be increased per kilometer of passenger distance. The energy consumption of vehicles is a factor that cannot be ignored.
The increase in the purchase price index of raw materials, fuel and power represents the progress of industrialization. However, the progress of industrial development will also bring pressure on energy supply to society. Every unit increase in the price index will increase the energy consumption of 0.55 1 10,000 tons of standard coal. Therefore, how to balance the relationship between industrial development and energy crisis is particularly important.
Conclusion and suggestion of intransitive verbs
According to the established optimal regression model, we find that economic development level, energy structure, transportation consumption and industrial production are the main factors affecting the total energy consumption. In order to realize the concept of sustainable development, this paper gives the following suggestions:
1. The government should intensify publicity and education, raise people's awareness of energy conservation, enhance their sense of hardship and responsibility, and make the whole society clearly realize that energy conservation and consumption reduction is an important way to achieve sustainable development and a guarantee for people's better survival.
2. It is suggested that the government should increase its support for scientific and technological innovation, speed up the reform of energy structure and improve the utilization rate of new clean energy (such as wind energy and hydropower), which is an important way and guarantee to realize sustainable development.
Although the vigorous development of industrialization will bring pressure on energy supply to society, they are not contradictory, and energy supply should not be a factor hindering industrial development. Enterprises should strive to improve energy efficiency and avoid waste. At the same time, encourage enterprises to use new clean energy and set an example for society. Although it may increase the production cost at first, it is of great help to the long-term development of the country and cannot be limited to immediate interests.
[References]
[1] Wang songgui, Chen Min, Chen Liping, linear statistical model, higher education press, 20 1 1 year1/month.
[2] Tang Yincai "R Language and Statistical Analysis", Higher Education Press, 20 12 March.
[3], Wei, He, Xavier Lee, Analysis and Empirical Research on Influencing Factors of Energy Consumption in Liaoning Province,/21.12/kloc-0+0230+5438.100000000605
(Author: 1. School of Science, Central University for Nationalities, Beijing100081; 2. School of Mathematics, Sichuan University, Chengdu 6 10065)
appendix