The Impact of Environmental Pollution on Residents’ Income Caused by the Imbalance of Regional Economic Development Based on Artificial Intelligence

Abstract

Regional economy is a human economic activity in a geographical region, a prominent and sustainable economy with distinctive regional characteristics. Regional economy is characterized by its integrity, relativity, relative independence, and spatial difference. With the increasing development of science and technology and big data, it has become a normal trend to use artificial intelligence technology to solve current social problems. In this paper, the social problems caused by the imbalance of regional economy are analyzed based on artificial intelligence. Through the application of KNN-SVM algorithm optimization, it is found that AI has relatively little impact on the development of the income level of the population under the age of 25 in each region. Compared with previous studies, the quality is compared with the innovation of this document, which is the development of a conceptual framework approach, an environmental pollution analysis mechanism, and income inequality analysis. The empirical research results show that under the strategic background of improving people’s livelihood, accelerating the reform of ecological civilization and promoting the construction of the health system, the relationship between environmental pollution and population income caused by unbalanced regional economic development can be re-analyzed through the best KNN-SVM algorithm. The implementation of the healthy China strategy has important theoretical and practical significance.

1. Introduction

Artificial intelligence was born at the World Dartmouth Conference in 1956, but up to now, the industry has not yet reached a consensus on what it is. At present, many definitions are used to agree that AI is a technology that uses machines to simulate human perception, cognition, reasoning, decision making, and other processes. Perception is a way to understand the world. Compared with human perception systems such as listening, speaking, seeing, and touching, the corresponding AI technologies such as speech recognition, semantic understanding, speech synthesis, and machine vision belong to perceptual AI technologies. Perceptual technology is a relatively mature part of various AI technologies at present and has been successively applied in various fields. Cognition is a psychological concept, which refers to the process in which the human brain receives information input from the outside, processes it, converts it into internal psychological activities, and then uses it to control human behavior activities [1,2,3].

The issue of income distribution has long been the focus of society and its residents. At present, the main social contradictions have changed accordingly. “Development thinking” is the main line of income distribution reform in the new era. From the perspectives of raising people’s income level, sharing the fruits of economic development, and promoting social equity, the reform of income distribution has given a new connotation of the times and put forward new goals and requirements. With the continuous development and application of artificial intelligence, its impact on the economy and society continues to deepen [4]. Economists have also begun to pay attention to the important impact of artificial intelligence on the economy: one area is the productivity improvement and job creation brought about by technological progress, and the other is the concern about the widening of employment and income inequality caused by “machine replacement”. This is because artificial intelligence, as a technology, has two major properties: first, artificial intelligence is a general-purpose technology that can be applied in a wide range of fields, so its impact on the economy is multi-faceted and profound; second, artificial intelligence is an enhanced version of automation, which can not only replace manual labor on a large scale through the intelligentization of capital but also replace a part of mental labor, thereby affecting the structure of labor demand, thereby affecting employment and income inequality. Because the resulting environmental pollution and high income inequality cannot be ignored, environmental pollution and the income gap, as two factors affecting the health level, have been confirmed in many studies. In the face of the current prominent environmental pollution and income distribution problems in China, the question is raised regarding whether they will hinder the implementation of the healthy China strategy in China. If it is a positive and promoting relationship, there is no conflict between narrowing the income gap and reducing environmental pollution. If it is a negative relationship, the income distribution policy to improve fairness will be contrary to the environmental policy. The imbalance of regional economic development is a common problem in the development process of all countries in the world [5].

In this paper, we will analyze the social problems caused by regional economic imbalance through the application of KNN-SVM algorithm optimization based on artificial intelligence. The empirical research results show that under the strategic background of improving people’s livelihood, accelerating ecological civilization reform, and promoting health system construction, the optimal KNN-SVM algorithm can re-analyze the relationship between environmental pollution and population income caused by unbalanced regional economic development.

2. Related Works

In refs. [1,2], it is found that many theoretical achievements with academic value have been produced, and different schools and styles have been formed. Refs. [3,4] pointed out that since the reform and opening up, China’s regional economic research has become increasingly active, developing very rapidly in theory, and a large number of research results have emerged. Refs. [5,6] believed that the inverted-U model, double-inverted-U model, and regional convergence theory are all the results of empirical tests on regional economic development gaps. However, whether regional economies can finally achieve convergence or whether the gap is widening has not been finally confirmed. The formation mechanism of the environmental poverty trap was analyzed by using the theories of dual-structure, human capital, poverty trap, and income distribution. This will enrich the theoretical research of the environmental poverty trap by analyzing the relationship between environmental and socio-economic development. Refs. [7,8] support the belief that it is closely related to economic growth. In recent years, China’s economic development goal has changed from pursuing “high-speed growth” in quantity to a “high-quality” development mode in quality. This transformation provides an improved path and opportunity for residents’ income distribution. However, first of all, it needs to be clearer, as understanding the influencing factors of income inequality in our country can help alleviate and improve the degree of income inequality. Previous studies have shown that the influencing factors of income inequality can be explained from the following aspects: economic growth, education inequality, and other possible influencing factors. There are abundant domestic and foreign studies on economic growth and income inequality.

Refs. [9,10,11,12] pointed out that before we take action, we need to better analyze the impact of AI on future inequality. Most economists analyze it theoretically and empirically and believe that it will promote economic growth. The technological progress after the information technology revolution is not only capital-oriented but also skill-oriented. Refs. [13,14,15] also introduced the adaptability of skilled labor and unskilled labor to new technologies to build an endogenous model and further explained the impact of technological progress on the wage gap between skilled workers and unskilled workers and believed that skilled workers had shorter time to adjust and adapt to new technologies, lower costs, and faster speed to adapt to new technologies.

Refs. [16,17] pointed out that the relevant research has involved many fields and has achieved extremely rich research results. Corresponding to the research theme of this paper, the following results only comb the research of domestic and foreign scholars on China from the measurement and decomposition of regional economic development imbalance. The studies in refs. [18,19] mentioned that appropriate imbalance is conducive to the development of interregional economy. Since the reform and opening-up, China’s economy has made great progress and made tremendous achievements. Compared with 1978, by 2017, China’s GDP had increased 201 times, with an average annual growth of about 15%. The proportion of the total economy in the world economy rose from 1.8% in 1978 to 14.84% in 2017. China’s economy has experienced 40 years of high-speed development, and its economy, culture, and politics have all developed in an all-round way. Its comprehensive strength has been continuously increased.

Ref. [20] believed that it is of great importance to China in the transition period to formulate economic and environmental policies for sustainable development and achieve a balance between economic benefits and environmental protection; this provides necessary theoretical support for the government to formulate relevant policies, reform the income distribution system, and break through and avoid pollution traps. From the perspective of income distribution, we can understand the causes of environmental pollution, provide a new perspective of environmental governance, explore how to narrow the income gap and meet the needs of pollution control, and open up the channel of “golden mountains, silver mountains and green rivers and mountains” on the basis of clarifying the differential mechanism of the two so as to provide reference for the construction of appropriate policy tools and control strategies. With regard to the changes of the inter-provincial regional gap, ref. [21] held that the changes of the inter-provincial gap took 1978 as the turning point and generally obeyed the law of the inverted-U-shaped curve; that is, after the reform and opening up, the inter provincial regional development gap gradually narrowed. Refs. [22,23] pointed out that after the reform and opening up, the inter-provincial gap was narrowed from 1978 to 1985, and the gap continued to expand after 1985. In the process of research, Chinese scholars have further revised and improved the “inverted-U” theory. Ref. [24] believed that Williamson’s ”inverted U-shape” theory is correct, but due to the imperfection of the method, it needs to be further revised. He and his collaborators introduced a double-S-curve model, reflecting that the two regions developed one after the other and formed a certain development law at the speed of “slow, fast, and slow”. They also introduced the regional growth function through empirical analysis; clearly and comprehensively reflected the main reasons, mechanisms, and change trends of the gap between the two regions; and accurately guided the coordinated change between regions. The “inverted-U” theory was thus developed and improved to a new height. Ref. [25] examined China’s regional economic gap from a quantitative perspective and found that China’s regional economic development gap presents a complex inverted-S shape.

In summary, analyzing residents’ income from the perspective of environmental pollution provides new ideas for reducing income inequality. Accelerating the reform of the ecological civilization system and building a healthy China requires researchers to clarify the relationship between environmental pollution and residents’ income caused by unbalanced regional economic development.

3. KNN-SVM Algorithm Optimization

The basis of the idea of KNN-SVM algorithm is to redefine the SVM classifier as a 1NN classifier in which the support vector of each category only takes one representative point. As shown in Figure 1, x is a sample to be classified, and f and x are support vectors. Whenever a new sample to be classified is obtained, we first calculate the distance difference between it and the support vector point. If the difference is larger than the threshold value of x that we set far from the separation hyperplane, then the SVM classification algorithm can be used for correct classification; if the difference is small, i.e., less than the threshold f that we set—that is, x is located near the classification hyperplane if the SVM classification algorithm is used—then only the distance between the sample point to be classified and the nearest support vector is calculated, in which it is easy to make mistakes.

4. Methods

Then, in order to improve the classification accuracy in this case, we use the KNN algorithm, take the support vector near the sample point to be classified as the representative point of the KNN algorithm, calculate the distance from the sample point to be classified to the nearby support vector, then determine the K value, and check which category the K support vectors belong to; thus, the mode of the category is selected as the category of the sample to be classified.

In the distribution-free approach, the linear quantile model for hierarchical data is as follows:

$$Q_{y_{ij}}(\tau )=X_{ij}^T{\beta }_{\tau }+Z_{ij}^T{u}_{\tau ,i}$$

(1)

The solution parameters can be obtained by the following formulas:

$$\underset{{}_{\beta ,u}}{\min }{\displaystyle {\sum }_{i=1}^M{\displaystyle {\sum }_{j=1}^{n_i}{\rho }_{\tau }}}\left(y_{ij}-X_{ij}^T\beta -Z_{ij}^T{u}_i\right)+{\displaystyle {\sum }_{i=1}^{M}P}\left(u_i\right)$$

(2)

The linear quantile mixed model (LQMM) joint densities for the τ quantiles of the M group are as follows:

$$p(y,u)={\left\{\frac{\tau (1-\tau )}{{\sigma }_{\tau }}\right\}}^N{\prod }_{i=1}^M\exp \left\{-\frac{1}{{\sigma }_{\tau }}{\displaystyle {\sum }_{j=1}^{n_i}{\rho }_{\tau }}\left(y_{ij}-{\mu }_{\tau ,ij}\right)\right\}P\left(u_i\right)$$

(3)

A method for lqmm estimation combines Gaussian quadrature and non-smooth optimization. The corresponding log-likelihood function is as follows:

$$l\left({\beta }_{\tau },{\displaystyle {\sum }_{\tau },}{\sigma }_{\tau }\mid y\right)={\displaystyle {\sum }_{i=1}^M\log }\left\{{\displaystyle {\sum }_{k_{1=1}}^K\dots }{\displaystyle {\sum }_{k_{q=1}}^{K}p}\left(y_i\mid v_{k_1,\dots ,k_q}\right){\prod }_{l=1}^q{w}_{kl}\right\}$$

(4)

According to the Sigmoid function, we obtain the prediction function of logistic regression as follows:

$$h_{\theta }(x)=g\left({\theta }^{T}x\right)=\frac{1}{1+e^{-{\theta }^{T}x}}$$

(5)

Finally, we obtain the final form of the parameter update:

$${\theta }_j={\theta }_j-\alpha \frac{1}{m}{\displaystyle \sum _{i=1}^m\left(h_{\theta }\left(x_i\right)-y_i\right)}{x}_i^j$$

(6)

The abscissa and ordinate of the Lorentz curve are, respectively, set as the cumulative population proportion and the cumulative income proportion. As shown in Figure 2, the absolute average line refers to the state when the cumulative income value and the cumulative population value are equal.

Finally, grouping calculation method, the idea of the group calculation method is the same as that of the geometric definition calculation integral method, as shown in Figure 3.

Now, we select all support vectors as the representative points of KNN algorithm and record them as ksvm1; because we study the binary classification problem, we first divide the support vector into positive and negative categories, calculate the expectation of each category of attributes, and finally obtain two representative points, which are called KSVM2. Combining the two classical algorithms SVM and KNN, an improved SVM classification algorithm is formed. Using this algorithm to classify data effectively improves the classification accuracy. In particular, the KSVM2 algorithm calculates the mean value of each dimension of support vector as the sample point of the KNN algorithm, and the classification effect is therefore better [26,27,28].

If the relationship between the two cannot be clearly defined, a single strategy of narrowing income distribution and reducing environmental pollution may not achieve the desired effect. At present, environmental pollution and income distribution in China are relatively prominent. It is urgent to formulate effective environmental policies and improve the fairness of income distribution. For a country, environmental pollution and destruction will undoubtedly bring serious obstacles to economic growth. Resource depletion, environmental degradation, and ecological damage are all major problems faced by the country’s sustainable development, especially for developing countries with relatively slow economic growth. At the same time, countries with rapid economic growth are also have a large amount of resource input. For a long time, my country has driven economic growth with an extensive economic growth model with high input, high consumption, and high emissions. The unprecedented development of economic growth is known as the growth miracle. However, the problem of environmental pollution has become particularly serious, and it is not advisable to sacrifice environmental pollution in exchange for economic growth.

A more general expression is as follows:

$$Y=X\gamma +Zu+\epsilon$$

(7)

The corresponding flow chart of the KNN-SVM algorithm optimization is shown in Figure 4.

This paper undertakes a descriptive statistical analysis of the micro-individual data and macro-industry data used to understand the individual factors of income at all levels and the overall characteristics of the industry. According to the descriptive statistical analysis of variables shown in

Table 1. Descriptive statistical analysis of variables.

Variable	Significance	Number of Cases	Average Value	Standard Deviation	Minimum Value	Maximum
One layer of microvariables
lnY	Natural logarithm of income	8686	10.40	0.83	7.17	13.64
Y	Personal income	8686	43,948.53	38,560.96	1300.00	840,000.00
Gender	Gender	8686	0.58	0.49	0.00	1.00
Edu	Years of education	8686	10.88	4.03	0.00	22.00
Exp	Hands-on background	8686	19.31	11.76	0.00	45.00
$Ex{p}^2$	Square term of work	8686	511.31	500.90	0.00	2025.00
Urban	Experience urban and rural	8686	0.66	0.473	0	1
Marriage	Marriage	8686	0.73	0.446	1	1
Secondary microvariable
Mon	Monopoly degree	19	39.80	34.50	2.68	99.11
Hum	Human capital	19	18.62	15.65	0.20	47.47
Fdi	Foreign investment	19	5.26	7.01	0.02	25.57

, it can be seen that the natural logarithmic mean of personal income is 10.40, and the corresponding average personal income is CNY 32,859.63 per year. From this, we obtain the amount of CNY 43,948.53 because taking the logarithm narrows the individual gap to a certain extent. The income after taking the logarithm shows the shape of the tail after the peak, and the KNN-SVM algorithm optimization in this paper is more suitable for this situation.

However, for residents, environmental pollution has significantly affected their production and life, leading them to fall into the trap of environmental poverty. To a certain extent, residents’ environmental poverty can be understood as the occurrence of poverty caused by environmental pollution. At this stage, the contradiction in our society is that residents’ need for a better life is becoming more urgent, and unbalanced development also exists. A series of problems caused by unbalanced development have become prominent, and environmental pollution has only recently been paid attention. Figure 5 is the natural logarithmic distribution of the personal income of men and women. In addition to observing that men’s income is generally higher than women’s, it can also be seen that men’s representative line in the graph is taller and thinner than that representing women, which indicates that men’s income is more concentrated, and the income gap is smaller.

Among them, r represents the range. When the distance between two points is less than this value, it means that there is an autocorrelation between them. On the contrary, if it is greater than this value, the autocorrelation will disappear, indicating that there is no effect. The relationship between them is shown in Figure 6.

In order to distinguish the upper and lower layers, the block diagram of the algorithm structure is shown in Figure 7.

According to the comprehensive index of environmental pollution calculated above, this paper draws a density function diagram of the performance of the comprehensive index of environmental pollution in different years, and its change trend is shown in Figure 8 below. Figure 8 represents the density function distribution of the comprehensive index of environmental pollution in different provinces in 1998, 2005, 2012, and 2017. It can be found that the comprehensive index of environmental pollution presents a right deviation distribution in different years. Generally speaking, the curve of environmental pollution density function changes slightly with the passage of time, which indicates that the environmental pollution index of China has not been improved very well in recent years.

In summary, with the deepening of reform and opening up, the general trend of per capita income of Chinese residents continues to increase, and the trend of the increase of household business income lags behind the increase of wage income; therefore, the gap is becoming more and more obvious, the proportion of residents’ wage income in the household income has grown rapidly, and wage income has become the most important source of promoting residents’ income.

5. Case Study

It can be concluded that AI will expand the income inequality of residents both in theoretical mechanism and empirical analysis. In the future, due to its strong penetration, artificial intelligence will gradually penetrate into all corners of the economy and society, and its impact on income inequality will undoubtedly continue to deepen. This chapter will put forward some policy suggestions on how to make artificial intelligence play a positive role and at least reduce the adverse effects in the development process. It mainly puts forward relevant policy recommendations from five perspectives, including government planning and guidance, industrial development and employment creation, higher education and education training, social security and reemployment training, urban and rural resource distribution, and regional industrial development in hoping to give full play to the positive role of artificial intelligence, reduce the adverse impact of artificial intelligence on employment and income inequality, and effectively deal with structural shocks. It can be clearly seen from Figure 9 that the absolute gap of the domestic economy is gradually expanding.

It can be seen from Figure 10 that the relative difference in China’s economic development fluctuates significantly. The fluctuation can be roughly divided into three parts. The first part is from 2000 to 2004, and the relative difference shows an expanding trend; the second part is from 2004 to 2007, the relative difference first narrowed and then expanded, but the overall change is not very obvious; the third part is from 2007 to 2018, and the relative difference is shrinking. In 2016, it fluctuated slightly, but on the whole, it shows a downward trend.

The Theil index decomposition method formula is as follows:

$$T_P=T_B+T_W$$

(8)

$T_P$ refers to the Theil index, $T_B$ refers to the total difference between regions, and $T_W$ refers to the difference within regions.

$$\begin{array}{l}{T}_B={\displaystyle \sum _i\left(\frac{GD{P}_i}{GDP}\right)}\ln \left(\frac{GD{P}_i/N_i}{GDP/N}\right)\\ T_W={\displaystyle \sum _i\left(\frac{GD{P}_i}{GDP}\right)}{T}_{pi},wherein{T}_{pi}={\displaystyle \sum _i{\displaystyle \sum _j\left(\frac{GD{P}_{ij}}{GD{P}_i}\right)}}\ln \left(\frac{GD{P}_{ij}/N_{ij}}{GD{P}_i/N_i}\right)\end{array}$$

(9)

The results are shown in Figure 11 and

Table 2. Difference contribution rate between regions.

Years	2000	2001	2002	2003	2004	2005	2006
Difference contribution rate between regions, Db	0.7132	0.7136	0.7158	0.7205	0.7219	0.7247	0.7288
Years	2007	2008	2009	2010	2011	2012	2013
Difference contribution rate between regions, Db	0.7263	0.7223	0.7164	0.7122	0.7037	0.6956	0.6898
Years	2014	2015	2016	2017	2018
Difference contribution rate between regions, Db	0.6849	0.6818	0.6806	0.6834	0.6874

.

These tables indicate that differences between regions have a more profound impact on total regional differences. The variation trend of the differences among the eight regions is very similar to that of the national overall differences, while the changes within the regional differences are not significant, indicating that the changes of the inter-regional differences drive the changes of the national total differences. From

Table 3. Moran’s I.

Variables	I	E(I)	sd(I)	z	p-Value
Ave	0.401	−0.033	0.112	3.984	0.000

, the global Moran index of the average value of China’s per capita real GDP from 2000 to 2018 is 0.401, and at a significant level, it indicates that the spatial correlation of economic development between provinces (cities) in China is strong.

After the implementation of the Western development strategy in various regions in the West, and the outflow of labor from other regions also tends to decrease. In order to more intuitively analyze the changing trends of the net labor outflow and inflow in the eight regions, we selected the following six years of labor net outflow and net inflow data to draw a line in Figure 12 as follows:

The results are shown in

Table 4. Estimated results for three periods.

Variable	2000–2006		2007–2012		2013–2018
	Estimated Value	p-Value	Estimated Value	p-Value	Estimated Value	p-Value
LF	0.7854	0.768	−19.2796	0.002	20.5684	0.271
KF	−2.6898	0.521	30.8120	0.031	39.5171	0.108
TF	0.4365	0.068	0.1234	0.857	2.4101	0.039
LNINV	7.2449	0.000	5.4011	0.013	−1.2421	0.234
LNGOV	−4.5271	0.004	0.1414	0.928	2.7468	0.067
LNIND	7.1417	0.010	−11.5610	0.026	−4.5986	0.338
LNINF	−1.4258	0.064	−0.2745	0.753	1.6706	0.002
-cons	−48.9301	0.001	29.4853	0.181	17.2126	0.451
R-sq	0.2577		0.5694		0.4912
Log-likelihood	−613.0554		−493.1359		−483.7068

.

In

Table 5. Effects of Environmental Pollution on Income Inequality: Human Capital Channels.

	Influence of Environmental Pollution on Human Capital
	Medical and Health Expenditure	Life Years	Years of Education
Environmental pollution	0.0444	−0.0176	−0.0666
	(0.0204)	(0.0415)	(0.0222)
Environmental governance	0.3225	0.5321	0.5236
	(0.0534)	(0.0536)	(0.0525)
Pollution Environmental treatment	−0.0534	−0.5326	−0.5234
	(0.0526)	(0.0634)	(0.06325)
Control variable constant term	Yes	Yes	Yes
	−3.136	6.688	−0.115
Adjust $R^2$	(0.144)	(0.295)	(0.256)
	0.946	0.055	0.834

, the environmental pollution significantly reduces life-years and years of education. Thus, environmental pollution widens income inequality by increasing health care spending and reducing years of life and education.

This study uses the actual per capita GDP of different regions (30 provinces and municipalities and 4 major regions). The results are shown in Figure 13.

Taking two perspectives as research units, Figure 13 shows that the overall regional economic development across the country shows a significant imbalance, but the degree of economic development difference among the 30 provinces and municipalities is greater than that among the 4 major regions. Taking 30 provinces and municipalities as the research unit, it is concluded that the coefficient of variation of the overall per capita GDP tends to shrink, with a decline rate of 24.88%, indicating that the degree of regional development imbalance is shrinking. The author investigated the rankings of 30 provinces and municipalities at four time points in 1998, 2003, 2007, and 2011, respectively, as shown in

Table 6. Regional Comparison of Basic Public Service Level and Economic Development Level.

Particular Year	Rank	Basic Public Service Level (From High to Low)	Economic Development Level (From High to Low)
1998	Top five	Beijing, Shanghai, Tianjin, Heilongjiang, Jilin	Shanghai, Beijing, Tianjin, Zhejiang, Jiangsu
	Last five	Hunan, Ningxia, Anhui, Guangxi, Guizhou	Yunnan, Ningxia, Shaanxi, Gansu, Guizhou
2003	Top five	Beijing, Shanghai, Tianjin, Xinjiang, Heilongjiang	Shanghai, Tianjin, Beijing, Zhejiang, Jiangsu
	Last five	Ningxia, Chongqing, Anhui, Guangxi, Guizhou	Sichuan, Ningxia, Yunnan, Gansu, Guizhou
2007	Top five	Beijing, Shanghai, Tianjin, Heilongjiang, Liaoning	Shanghai, Tianjin, Beijing, Zhejiang, Jiangsu
	Last five	Sichuan, Anhui, Guangxi, Ningxia, Guizhou	Sichuan, Ningxia, Yunnan, Gansu, Guizhou
2011	Top five	Beijing, Shanghai, Tianjin, Liaoning, Guangdong	Shanghai, Tianjin, Jiangsu, Zhejiang, Beijing
	Last five	Yunnan, Anhui, Guangxi, Ningxia, Guizhou	Qinghai, Ningxia, Yunnan, Gansu, Guizhou

, and came to the following conclusions: there are areas with higher economic development level, and in the areas with low economic development level, the basic public service level is also low, such as Ningxia, Guizhou, and Sichuan. At the same time, it is also found that the top five places for basic public services are mostly in the eastern region and the northeast region, and the top five places for economic development are all in the eastern region; the bottom five in the horizontal ranking are all western provinces [29,30].

The specific results are listed in

Table 7. Statistical analysis table of real GDP per capita in 31 provinces in my country.

Particular Year	Full Range	Average Difference (CNY)	Standard Deviation (CNY)	Relative Range	Unbalance Difference	Coefficient of Variation
1998	22,891.15	3306.74	4674.59	3.2063	0.9086	0.6548
1999	26,312.49	3618.72	5345.68	3.6309	0.9197	0.7377
2000	27,296.45	365.08	5583.97	3.7108	0.9229	0.7591
2001	27,251.26	3810.88	5659.64	3.6740	0.9226	0.7630
2002	27,147.80	3876.53	5704.42	3.6331	0.9224	0.7634
2003	28,371.33	4057.70	5964.14	6.6333	0.9229	0.7639
2004	30,533.22	4318.02	6364.83	6.3289	0.9238	0.7565
2005	24,904.33	4206.12	5873.17	2.8647	0.9019	0.6756
2006	24,705.08	4263.06	5842.74	2.7816	0.8997	0.6578
2007	24,770.40	4308.54	5827.64	2.7056	0.8958	0.6.65
2008	24,431.60	4410.64	5835.57	2.4941	0.8793	0.5957
2009	23,853.64	4491.61	5871.49	2.3900	0.8695	0.5883
2010	19,763.21	4270.60	5434.69	1.8833	0.8275	0.5179
2011	19,717.74	4267.57	5387.06	1.7443	0.8074	0.4766
2012	19,520.55	4166.89	5245.17	1.6932	0.7885	0.4550
2013	18,976.08	4131.17	5172.16	1.6236	0.7687	0.4425
2014	18,109.24	4064.25	5074.52	1.5529	0.7488	0.4351
2015	17,584.63	4010.44	5010.96	1.5409	0.7576	0.4394
2016	18,247.27	4055.78	5182.95	1.5952	0.7661	0.4531
2017	20,411.01	4308.90	5600.15	1.6514	0.7791	0.4531

,

Table 8. Statistical analysis of absolute differences in real GDP per capita in five regions of China.

Particular Year	East			Central			Southwest			Northwest			Northeast
	Full Range	Mean Difference	Standard Deviation	Full Range	Mean Difference	Standard Deviation	Full Range	Mean Difference	Standard Deviation	Full Range	Mean Difference	Standard Deviation	Full Range	Mean Difference	Standard Deviation
1998	19,363.67	3973.23	5651.60	1851.94	674.65	451.71	2369.31	840.85	605.16	2939.17	1050.38	783.69	3446.06	1724.13	1172.37
1999	22,682.80	4917.81	6807.19	1721.04	660.44	438.49	2183.57	770.84	503.87	2602.46	927.52	695.14	3478.30	1764.30	1273.67
2000	23,673.47	5196.54	7131.54	2000.68	748.56	509.37	2135.95	747.88	492.25	3109.32	1082.33	796.26	3748.82	1889.03	1339.88
2001	23,686.72	5400.27	7245.84	2048.17	772.92	518.63	2180.13	779.35	525.04	2963.21	1,024,006	745.41	3477.62	1753.21	1245.51
2002	23,780.85	5435.90	7272.81	1811.64	675.92	445.71	2312.70	832.02	576.24	2815.93	990.17	751.71	3368.40	1695.92	1199.39
2003	25,100.88	5789.77	4600.71	1681.95	591.10	391.20	2944.06	978.16	672.52	3078.30	1131.32	868.61	3237.55	1620.27	1105.94
2004	27,404.72	6033.62	8128.93	1632.68	577.95	419.86	3222.92	1071.14	732.67	3188.26	1253.84	982.41	3206.19	1603.85	1106.25
2005	21,714.43	5454.30	7041.66	2091.19	767.04	608.72	3181.31	1052.92	725.44	4749.97	1658.34	1264.77	3023.05	1606.06	1214.11
2006	21,436.80	5321.82	6883.99	1936.12	759.97	629.95	3174.52	1064.99	751.72	5376.22	1842.68	1353.89	2888.00	1603.85	1232.32
2007	21,587.23	5230.73	6831.70	2041.56	841.40	693.92	3226.92	1106.72	801.21	6269.26	2017.97	1431.64	3021.10	1606.11	1258.92
2008	21,258.53	5165.33	6706.21	2246.72	996.42	827.31	3496.04	1175.04	829.40	7638.77	2548.50	1628.65	3621.80	1646.21	1458.80
2009	20,746.90	5221.86	6760.81	2177.44	848.64	684.55	4380.00	1461.79	993.38	9519.92	3204.84	2039.07	4442.86	1925.21	1654.46
2010	16,400.44	4746.44	5949.44	2203.13	865.92	665.24	4544.71	1587.52	1193.29	9805.16	3266.11	2048.89	4796.48	2266.71	1816.24
2011	16,139.60	4659.58	5752.65	2446.95	925.43	714.77	5183.64	1862.03	1442.72	10,999.23	3667.54	2295.29	5141.80	2463.86	1925.98
2012	16,154.68	4521.35	5589.49	2598.74	976.56	734.67	5102.88	1849.84	1400.92	11,135.77	3702.14	2348.69	5563.63	2629.27	2017.81
2013	15,890.74	4474.87	5526.55	2686.84	1002.66	723.66	4949.55	1792.55	1306.19	10,676.60	3529.52	2283.04	5991.89	2813.97	2129.82
2014	15,328.57	4383.65	5125.87	2923.29	1135.25	857.76	4921.10	1823.86	1274.55	10,252.89	3381.13	2230.87	5969.54	3015.66	2094.72
2015	14,555.50	4368.09	5391.68	3382.78	1169.79	956.75	5055.35	1857.01	1250.49	9660.53	3164.04	2058.97	5566.37	2977.18	1918.61
2016	15,140.26	4569.14	5630.59	4056.90	1412.35	1030.50	5523.04	1998.39	1331.27	8951.04	2951.22	1968.85	2007.42	1418.43	1065.51
2017	16,980.54	4946.40	6110.60	3684.02	1374.99	991.90	5934.76	2151.35	1503.85	7162.80	2482.65	1838.41	2624.45	1444.49	1107.24

and

Table 9. Statistical analysis table of relative differences in real GDP per capita in five regions of China.

Particular Year		East			Central			Southwest			Northwest			Northeast
	Relative Range	Unbalance Difference	Coefficient of Variation	Relative Range	Unbalance Difference	Coefficient of Variation	Relative Range	Unbalance Difference	Coefficient of Variation	Relative Range	Unbalance Difference	Coefficient of Variation	Relative Range	Unbalance Difference	Coefficient of Variation
1998	1.6395	0.7686	5651.60	1851.94	674.65	451.71	2369.31	840.85	605.16	2939.17	1050.38	783.69	3446.06	1724.13	1172.37
1999	1.8247	0.7928	6807.19	1721.04	660.44	438.49	2183.57	770.84	503.87	2602.46	927.52	695.14	3478.30	1764.30	1273.67
2000	1.8580	0.8004	7131.54	2000.68	748.56	509.37	2135.95	747.88	492.25	3109.32	1082.23	796.26	3748.82	1889.03	1339.88
2001	1.8395	0.8019	7245.84	2048.17	772.92	518.63	2180.13	779.35	525.04	2963.21	1024.06	745.41	3477.62	1753.21	1245.51
2002	1.8267	0.8080	7272.81	1811.64	675.92	445.71	2312.70	832.05	576.24	2815.93	990.17	751.71	3368.40	1695.92	1199.39
2003	1.8409	0.8165	7600.71	1681.95	591.10	391.20	2944.06	978.16	672.52	3078.30	1131.32	868.61	3237.55	1620.27	1105.94
2004	1.8734	0.8291	8128.93	1632.68	577.95	419.86	3222.92	1071.14	732.67	3188.26	1253.84	92.41	3206.05	1606.06	1106.25
2005	1.4776	0.7863	7041.66	2091.19	767.04	608.72	3181.31	1052.92	725.44	4749.97	1658.34	1264.77	3206.05	1603.85	1214.11
2006	1.4380	0.7807	6883.99	1936.12	759.97	628.95	3174.52	1064.99	751.72	5376.22	1842.68	1353.89	2888.00	1603.11	1233.32
2007	1.4260	0.7807	6831.70	2041.56	841.40	693.92	3226.92	1106.72	801.21	6269.26	2107.97	1431.64	3021.10	1646.21	1258.92
2008	1.3492	0.7651	6706.21	2246.72	996.42	827.31	3496.04	1175.04	829.40	7638.77	2548.50	1628.65	3621.80	1925.21	1458.80
2009	1.3096	0.7562	6760.81	2177.44	848.64	684.55	4380.00	1461.79	993.38	9519.92	3204.84	2039.07	4442.86	2266.71	1654.46
2010	1.0286	0.6867	5949.44	2203.13	865.92	665.24	4511.71	1587.52	1193.29	9805.16	3266.11	2048.89	4796.48	2463.86	1816.24
2011	0.9720	0.6609	5752.65	2446.95	925.43	714.77	5183.64	1862.03	1442.72	10,999.23	3667.54	2295.29	5141.80	2629.27	1925.98
2012	0.9722	0.6525	5589.49	2598.74	976.56	734.67	5102.88	1849.84	1400.92	11,135.77	3702.14	2348.69	5563.63	2813.97	2017.81
2013	0.9500	0.6437	5526.55	2686.84	1002.66	723.66	4949.55	1792.55	1306.19	10,676.60	3529.52	2283.04	5991.89	3015.66	2129.82
2014	0.9156	0.6301	5425.84	2923.29	1135.25	857.76	4921.10	1823.86	1274.55	10,252.89	3381.13	2230.84	5969.54	2997.18	2094.72
2015	0.8855	0.6271	5391.68	3382.78	1269.79	946.75	5055.35	1857.01	1250.49	9660.53	3164.04	2058.97	5566.37	2788.02	1918.61
2016	0.8982	0.6357	5630.59	4056.90	1412.35	1030.50	5523.04	1998.39	1331.27	8951.04	2951.22	1965.85	2707.42	1418.43	1065.51
2017	0.9248	0.6481	6110.60	3684.02	1374.99	991.90	5934.76	2151.35	1503.85	7162.80	2482.65	1838.41	2624.45	144.49	1107.24

.

Through analysis, it can be concluded that basic public services affect regional economic growth and then regional economic development through human factors, as shown in Figure 14 for the impact mechanism.

According to the actual operation of China’s economy in the research period (as shown in Figure 15) and some studies on the economic cycle, this paper divides the research period into three periods: high-speed growth period (2000–2007), policy digestion period (2008–2013), and deep adjustment period (2014–2019).

Whether it is all years or sub-years, the TFP imbalance effect is always higher than the capital deepening imbalance effect, which is consistent with the research conclusions of many scholars. The imbalance effect is nearly 10 percentage points higher than the 2008–2013 model, which is in line with the major in-depth adjustment strategies such as the innovation-driven strategy and supply-side structural reform launched by the central government after China entered the new economic normal, as shown in Figure 16 and Figure 17.

The per capita scientific expenses of each province are used as an indicator to measure the progress of science and technology in a region. This is because the scientific expenses are the scientific expenses of the national budget allocation for the centralized management of the science and technology commissions at all levels and the scientific expenses of the Chinese Academy of Sciences system, as shown in the Figure 18.

Nationwide, the impact of income gap on pollution scale showed an inverse “n” curve. At first, income distribution was relatively equal, income gap was not easy to detect, and the positive feedback effect of economic growth was dominant, which was conducive to the improvement of environmental quality; then, the income gap widened, the social stratification was obvious, more and more low- and middle-income groups emerged, the incentive role of income gap was dominant, and backward production and unreasonable economic competition to improve income aggravated pollution. When the income gap further widens, high-income groups and elites will emerge, and the social demand for a high-quality environment and the initiative to improve the environment will be strengthened. In addition, the implementation of policies and environmental regulations will ultimately improve the environmental quality. At present, the scale of air pollution and soil pollution in China is in the stage of increasing with the expansion of income gap. The income gap is too large, which has deviated from the social optimal choice, while water pollution is the opposite.

In summary, the income gap in the central region has the most complex impact on environmental pollution, and all three paths exist; water pollution in the eastern region is not significantly affected by industrial effects, and technological effects are not significantly affected by wastewater pollution and soil pollution. The income gap in the western region does not have a significant industrial effect on air pollution, and the industrial effect of solid pollution and the technical effect path of fertilizer pollution are also not significant.

6. Conclusions

With the growing development of science and technology and big data, it has become a normal ongoing state to solve current social problems by using artificial intelligence technology. This paper discusses the impact of income gap on environmental pollution in China, analyzes the main impact paths among regions, and draws the main conclusions of the study with a coordinated strategy for social equity and environmental protection. Through the application of KNN-SVM algorithm optimization, this paper undertakes an empirical study on the imbalance of regional economic development in recent years, analyzes the influencing factors of the imbalance of regional economic development, and points out the reasons for the imbalance of regional economic development in China. Finally, aiming at the current situation of determining how to alleviate the imbalance of regional economic development in different functional areas, this paper makes a targeted discussion and puts forward a solution with practical guiding significance. The empirical research results show that under the strategic background of improving people’s livelihood, accelerating ecological civilization reform, and promoting health system construction, the optimal KNN-SVM algorithm can re-analyze the relationship between environmental pollution and population income caused by unbalanced regional economic development. At the same time, the study also points out that hard work and compliance with the law must be encouraged in order to increase wealth. We will adjust the structure and system of income distribution, raise the income level of residents, narrow the income distribution gap, and achieve common prosperity for the people. However, after overcoming endogeneity, people will find that environmental governance can effectively reduce environmental pollution, thereby reducing inequality.

In the future, we will continue to optimize the KNN-SVM algorithm so that it can obtain more accurate data in the AI analysis of the impact of unbalanced regional economic development on residents’ income.