Analysis of an Economic Coupling Relationship Model of the Coastal Ecological Fragile Zone Based on a Machine Learning Model

This study is based on a decision tree learning model and integrated system. In order to improve the classification accuracy of the base learner, the diversity of the results is analyzed. Accuracy and diversity are the focus of decision tree classifier research. The balance between decision tree classifiers is a key issue in creating a good ensemble classifier. By reducing the redundant features and training the model with the optimal feature subset, the accuracy of the model can be improved and the running time can be reduced. Normal distribution and generalized error distribution are better. The clustering algorithm based on the K-means partition is a number of compact and independent classification clusters based on the K -means partition algorithm. In order to objectively and accurately treat the machine learning model, an analysis model and economic coupling relationship model were used to evaluate the coastal ecological fragile zone. The error of the result of the classifier after training is smaller.


Introduction
The machine learning model is an integrated learning system that combines a variety of learning methods [1][2][3]. It is not like the previous data mining decision tree algorithm. Machine learning models gradually appear as decision treebased classifiers, and data mining-based classifiers are slowly playing an irreplaceable role in the public's field of vision. At present, machine learning has the research direction of multiple attribute sets. There are not only machine learning algorithms with a single attribute set such as decision trees and logistic regression algorithms but also machine learning algorithms integrated with subsets of attribute sets, such as random forests, attribute set, subset, BDT and so on. The integrated learning algorithm is a method in which the integrated system, the Bagging algorithm, several independent self-service data sets, the data set prediction, the classifica-tion results, and the decision tree classifier jointly complete the model work. When the errors of the learners of each integrated system are independent of each other, as the number of independent self-service data sets that make up the integrated system increases, the prediction and classification results of the data sets caused by the integrated system will drop sharply. It even tends to 0 [4][5][6]. Therefore, the integrated system has a more accurate model work effect than a single Bagging algorithm learner. Since each independent self-service data set-based learner deals with the same work problem at the same time, it cannot achieve completely independent data set predictions. Generally speaking, the higher the accuracy of the ensemble learner, the diversity of the classification results of the base learner will decrease. Accuracy and diversity are the focus of decision tree classifiers. How to balance the relationship between the decision tree classifier and the integrated system and create a good integrated classifier is a key issue. The statistical model learners have a strong dependent relationship; the integrated system represented by Bagging has good applicability and high matching degree and can be used to explain a variety of economic conditions. The statistical model of the base learner is parallelized, so there is no strong dependence between the statistical models of the base learner. When measuring the marine economy and resource environment, the statistical model uses a sampling method with replacement and selects objective and practical classifiers to train and test the data set. Using the balanced decision tree classifier and ensemble system, the error in the classification test set of the marine industry-based classifier in the large coastal marine ecosystem is calculated. The relationship between marine economy and marine resources and environment in various regions should be classified correctly when the model weight is updated. Finally, the Bagging algorithm is used to experimentally analyze the updated weights, which will contribute to the next prediction of the fragile economic coupling relationship of coastal ecology. Using quantitative methods such as index analysis, comprehensive evaluation, principal component analysis, BP neural network, general equilibrium model, or system dynamics, let the base classifier focus on training and classification. The most important thing about the principle of the Bagging algorithm is that various variables should be absolutely independent, and the sampling method with replacement should be used. The independent variables and dependent variables are strictly defined [7][8][9][10]. The statistical model of the base learner randomly selects n data from the original data set D with the number of N to form a traditional statistical model selfservice data set. Through repeated definitions and requirements, several independent self-service data sets with low accuracy can be obtained. China has vast sea areas and abundant marine resources. It is one of the world's maritime powers. It faces a situation of weak domestic economic growth, shortage of various resources, and serious environmental damage. Using statistical models to predict the country's marine resources to achieve the coordinated development of marine economy and marine resources and the environment is not accurate. The country's emphasis on marine economy has reached an unprecedented height. Decision tree algorithm modeling can help human beings develop marine resources. Machine learning models derived from exchange, distribution, and exchange tend to be overfitting. Through neural network algorithm modeling, we must insist on land and sea coordination, accelerate the construction of a machine learning model for a maritime power, and determine that China will highly match the local economic conditions of the land through the development of the marine economy in the new century. A boosting algorithm expounds the concept of marine economy and regards the learning algorithm as part of marine economic equivalence. The status of the marine economy has been greatly improved [11]. The BP neural network algorithm has greatly improved the strategic position of the marine economy in the decision-making of regional national policies. Compared with the traditional neural network algorithm, the prediction accuracy of the BP neural network algorithm is greatly improved, and there is no overfitting problem with a high probability; the calculation amount is relatively small, and the requirements for independent variables are not as strict as the traditional model. The BP neural network algorithm incorporates the marine economy into an important part of the national economy for application [12][13][14][15]. But so far, only a small number of researchers have applied this algorithm to the early warning assessment of marine resource development from the perspective of the social production process. And the concept of marine economy was elaborated. Based on the analysis of the economic coupling relationship model of the coastal ecological fragile zone based on the machine learning model, it was creatively set out to believe that the marine economy is the sum of the production, exchange, distribution, and exchange of the development of marine resources by humans.

Machine Learning and Coastal Ecology
2.1. Random Forest. The research of marine economy is more of an adjunct to the advanced complex random forest developed by the decision tree. It mainly focuses on basic issues such as the definition, characteristics, and subject orientation when selecting the attributes of the marine economy. The theory of marine economy will automatically select the best attributes. The base classifier of the random forest algorithm is a decision tree. And it is worth noting that it is not only the decision tree algorithm used but also random attribute selection during model training, as shown in Figure 1 [16][17][18].
2.2. Extreme Random Tree. Extra tree does not use the best points when selecting branch points but uses the updated random weights to pick out a feature value to divide the tree classifier [19][20]. Let the base classifier focus on training classification, so the whole algorithm is very random, as shown in Figure 2. 2.3. Measurement of the Development of Marine Economy and Resources and Environment. When the marine economy and resource environment are measured, the tuple data is sampled with replacement, and the training set of measurement results is selected to a certain extent. Use the objective and practical base classifier q to train the test data set A, and calculate the error of the marine industry-based classifier in the large coastal marine ecosystem. Correspondingly, the relationship between marine economy and marine resources and environment in each region is classified correctly during model training, and the development theory and research methods of the marine eco-economic system should be expanded in the next weight update to reduce the weight of this tuple. Finally, using selective experimental analysis, nonlinear utility functions, establishing statistical models, and updated weights, exponential analysis, comprehensive evaluation, principal component analysis, BP neural network, and general equilibrium models or quantitative methods such as system dynamics allow the base classifier to focus on training classification.

Application of the Machine Learning
Model in the Economic Coupling Relationship of the Coastal Ecological Fragile Zone (a) AIC algorithm model [21][22][23] dS t = μsS t dt + σS t dW t + H t dN t ,

Wireless Communications and Mobile Computing
Machine learning model: Coastal ecological forecast: Marine resources are produced, exchanged, distributed, and exchanged: Economic coupling relationship model: Economic coupling relationship of the coastal ecological fragile zone: (c) HQ algorithm model [27][28][29][30] dS t S t = μdt + σdBt, Prediction of the relationship between marine economy and resources: Result test training: The classifier is training Compare the result error after training       Table 1.
From Table 1, we can conclude that the GARCH model under the generalized error distribution with fixed parameters is the smallest under the three information criteria, and all the coefficients are significant. Normal distribution and generalized error distribution feature screening results are better. In the normal distribution, AIC = 5:169, SC = 5:212, and HQ = 5:186; in the generalized error distribution, AIC = 4:9, SC = 4:943, and HQ = 4:917, as shown in Figure 3.

K-Means
Cluster Analysis. The K-means algorithm is such a clustering algorithm based on centroid partition, using distance as the criterion, indicating the higher degree of similarity, as shown in Table 2 and   Table 3. Both autocorrelation and heteroscedasticity are 5better resolved as shown in Figure 5.
The correlation test is based on an improved algorithm based on GBDT. Use the Taylor expansion to expand the LjungBox function to the first order. XGBoost uses Taylor's formula to expand the objective function to a second-order column with fixed parameters. XGBoost stores a lot of objective function information, which can make the economic coupling relationship model of the coastal ecological fragile zone of the machine learning model have relatively less loss and lower variance, as shown in Figure 5.

Error Comparison.
In order to objectively and accurately evaluate the economic coupling relationship model analysis model of the coastal ecological fragile zone of the machine learning model, compare the result error of the classifier after training and change the weight of the training tuple according to the classifier error. Each tuple of the machine learning model has a probability of being picked out, and the error training several times biases the model to being "error-prone." At the beginning, the same "error-prone" weight of the biased model is set, so that the probability of training after any one of the biased models is selected as "error-prone" is set by the weight. Regarding the issue of iterative update weights of economic coupling relations in the coastal ecological fragile zone, the classification of the model is "error-prone" as much as possible, as shown in Table 4.
As shown in Figure Table 5. The SC algorithm model is the optimal model. In

Conclusion
Based on the machine learning model, this study analyzes and predicts the economic coupling relationship of the coastal ecologically fragile zone and builds a machine learning model for human production, exchange, distribution, exchange, and development of marine resources through the decision tree algorithm. By reducing the redundant features and using the optimal feature subset to train the model, the accuracy of the model can be improved and the running time can be reduced. Normal distribution and generalized error distribution are better. The clustering algorithm based on the K-means partition is a number of compact and independent classification clusters based on the K-means partition algorithm. By reducing the redundant features, the optimal number and the optimal statistics are obtained. In order to objectively and accurately treat the machine learning model, an analysis model and economic coupling relationship model were used to evaluate the coastal ecological fragile zone. The future work will focus on the optimization and analysis of the experimental application effect and performance of the model, focusing on the optimization of data sets.

Data Availability
The experimental data used to support the findings of this study are available from the corresponding author upon request.