1. bookAHEAD OF PRINT
Journal Details
License
Format
Journal
eISSN
2444-8656
First Published
01 Jan 2016
Publication timeframe
2 times per year
Languages
English
Open Access

Economic Research on Multiple Linear Regression in Fruit Market inspection and Management

Published Online: 15 Jul 2022
Volume & Issue: AHEAD OF PRINT
Page range: -
Received: 25 Apr 2022
Accepted: 29 Jun 2022
Journal Details
License
Format
Journal
eISSN
2444-8656
First Published
01 Jan 2016
Publication timeframe
2 times per year
Languages
English
Introduction

With the increase in people's emphasis on health and requirements for quality of life, the demand for fruits by residents of various countries has increased significantly. Because of its geographical location, Thailand has the advantage of producing high-quality tropical fruits [1, 2, 3]. On average, fruit exports can achieve economic benefits of more than 1.3 billion U.S. dollars. The fruit plantation industry has developed into an important economic support industry in Thailand at this stage [45]. At the same time, with the improvement of the level of economic and trade, Thailand has become a member of many trade groups or economic groups in the world. Against this background, Thailand's fruit export trade will also develop rapidly [6].

With the launch of the ”Internet +” action plan, all walks of life have entered an era of rapid development relying on the Internet and information technology [7, 8, 9]. As of December 2019, the number of listed Internet companies in China and abroad reached 102, with a total market value of 8.97 trillion RMB. Among them, the sum of market value of Tencent, Alibaba and Baidu companies accounted for 73.9% of the total market value, and online game, e-commerce, cultural media, online finance and software tool companies of listed companies accounted for 28.4%, 14.7%, and 10.8% of the total companies respectively [10]. 6.2%, 9.8%, 5.9%. The original offline business model of agricultural products companies has also begun to develop in the direction of relying on networks and information technology [11].

The online and offline integration business model has become the current trend of social development. The competition of individual companies has also become the supply chain and the supply chain. Competition among companies, the financial business online of the agricultural product supply chain has become an inevitable trend.

Timme and others earlier proposed the concept of supply chain finance, which was defined as a collaborative relationship under the environment of specific targets such as objects, processes, liquid assets, fixed assets, and people in the supply chain. Hofmann pointed out that supply chain finance is an activity in which multiple organizations in the supply chain participate together to plan, coordinate, and control the flow of funds in the supply chain through the assistance of the organizations. Aberdeen proposed that supply chain finance is a financing scheme based on electronic trading platforms that provides transaction information. Participants, including financial institutions, core enterprises, and upstream and downstream supporting companies, work together to optimize financing costs and settlement costs.

Michael believes that supply chain finance is to aggregate, integrate and package all relevant information in the supply chain, and coordinate with various financial strategies such as cost management to optimize the cost of capital acquisition. The viewpoint jointly proposed by Jinzheng Zhen and Lu Guangsheng (2011) is that the current Sino-Thai fruit transaction volume continues to grow, but there are still many major problems in the whole process, such as the uneven development of trade between the two countries and the different product structures. The existence of these problems has adversely affected the fruit trade between China and Thailand. Zhang Fang, Wu Shuangjiang, and Tian Fangtao (2016) described the overall situation of China-Thailand trade exchanges and pointed out that China's current trade deficit in such trade exchanges has widened. In addition, Zheng Shujuan (2016) also pointed out that during the development of China-Thailand fruit trade, Chinese merchants have adopted some control measures for the export of Thai fruits.

The current form of online financial business for fresh produce is in a considerable state. However, all parties involved in this business have risk factors that cannot be underestimated. For agricultural product suppliers, most of them are scattered farmers, and processing companies and distributors are usually small and mediumsized enterprises. The emerging e-commerce platforms have imperfections in terms of entry barriers and technologies. The laws and regulations in this regard are not yet complete. Therefore, factors such as credit, market, and technology affect the stability of the online supply chain of fresh produce, thereby affecting its risks. This article starts from the fresh agricultural products that are closely related to people's lives, and constructs an analysis model of the online supply chain financial risks of fresh agricultural products to provide a theoretical basis for the development of the agricultural and financial industry markets. This will be important for both theory and practice meaning.

Focusing on the shortcomings of the existing regularization Markov blanket (edge) discovery methods, two different solutions based on regularization models are proposed, and the permutation in the process of regularization model discovery of Markov edges is also studied. Testing method. The research content can be summarized into two parts. The first part is to propose two different solutions to the shortcomings of the existing MRLM based on the regularization model. The first is to solve the Markov edge discovery problem on existing collinear datasets that cannot be handled by existing algorithms by introducing appropriate super parameters on the basis of existing MRLM. The second solution is to bypass the method of covariance calculation to solve the shortcomings of MRLM. The specific method is to use the classic regularization method directly on the low-dimensional data set to explore the relationship between the discovery performances of Markov edges of different regularization models. The second part is to conduct further research on the replacement test method used in the regularization model, focusing on the specific expressions of the three different implementation methods of the replacement method in the regularization model and the performance of Markov blanket (edge) discovery.

Based on artificial neural network algorithms to mine information from numerous structural diagrams, give data sets from experiments and specific simulation calculations, use the obtained data to train machine learning models, analyze the structure of materials to predict unknown performance, or construct machine learning models from databases To accelerate understanding of structure-property relationships at the nanoscale and predict new materials; or to optimize the microstructure of a material to match target performance. Material databases created by several national key laboratories provide data-driven materials design tools. The relationship between material structure and performance is a key issue in the field of materials.

The microstructure of a material determines its properties, and the microstructure size of materials with different functions will also vary greatly. David et al. And Tobias et al. Used domain analysis and analytic hierarchy process to construct evaluation models for supply chain financial risks. Bai Shizhen and others built a supply chain financial risk assessment model based on BP neural network and using matlab's BP neural network tool. Hu Haiqing et al. built a credit risk assessment model for small and medium-sized enterprises from the perspective of supply chain finance based on a comparative study of support vector machines SVM and BP neural network. Through comparison of empirical results, it is shown that the credit evaluation model based on SVM in a small sample is more effective. He Juan and other methods based on Copula-CVa R-EVT discussed the optimization of the supply chain financial pledge portfolio, and they made important contributions to reducing the supply chain financial risks. Luo Qingping adopted a factor analysis method to determine the important factors affecting the financial risk of the supply chain. Xu Qifa studied the optimization of loan portfolios in supply chain finance based on Copula-quantile regression. Based on structural analysis, Huang Xiangyu and others constructed a linkage model of agricultural product warehouse receipts pledge risk, and used a system dynamics method to describe the dynamic process of factors such as the agricultural product enterprise's operating capacity subject to warehouse receipts pledge business and other factors. Xu Peng used analytic hierarchy process and fuzzy comprehensive evaluation method to study the external environment change risk, credit level risk and supply chain stability risk of supply chain finance, thereby constructing a three-tier agricultural product pledge financing business risk evaluation index system. Wang Jingjing designed a financial risk prevention system for the online agricultural product supply chain using factor analysis. Figure 1 is a schematic diagram of Markov edge discovery based on a regularization model and a sub-framework diagram of this study

Figure 1

Network Security Situation Awareness Model

Multiple linear regression model algorithm

Table 1 sets the variables used in the quantitative linkage analysis model of financing companies' interests in the online supply chain financial system of the fruit market. In the financing enterprise benefit system, the financing enterprise's income only comes from the payment received after completing the order, that is, sales income. Its expenditure includes production input costs, transaction costs such as packaging logistics paid during the transaction, and storage costs of fresh produce and loan interest paid to financial institutions, for dimensional consistency, the net interest rate of the financing company is used to represent the interests of the financing company, then: A1=[(A2A3A4A5L1)A2]*100 {A_1} = \left[ {{{\left({{A_2} - {A_3} - {A_4} - {A_5} - {L_1}} \right)} \over {{A_2}}}} \right]*100 A2=DELAY1(C1*price,13) {A_2} = DELAY\,1\left({{C_1}*price,13} \right) A3=0.7A6 {A_3} = 0.7\,{A_6} A6=0.2*A7*C1 {A_6} = 0.2*{A_7}*{C_1}

Variable set of financing system

name Code Description
Daily interest on loans L1 Unit: 10,000 yuan, which means the interest paid by the financing enterprise to the financial institution during the financing process
Daily interest rate of loans R1 The unit is dimensionless, indicating the change in daily interest rate of the loan per unit time
Financing companies' net interest margin A1 Unit: %, representing the income of the financing company
Sales revenue A2 Unit: 10,000 yuan, representing sales in come of financing enterprises
Production input cost A3 Unit: 10,000 yuan, indicating the cost of production input of the financing enterprise
Financing companies average transaction costs A4 Unit: 10,000 yuan, which means the transaction cost of packaging, transportation, etc, that the financing company pays to complete the order
Average storage fee A5 Unit: 10,000 yuan, which means the financing company keeps expenses for storage
Loan amount A6 Unit: 10,000 yuan, indicating the amount of financing obtained by the financing company
Financing rate A7 Units are dimensionless, indicating the credit ratio determined by financial institutions
Number of agricultural products A8 Unit: Ton, indicating the amount of agricultural products produced by the financing company
Order volume C1 Unit: Ton, indicating the quantity of agricultural products agreed to be sold in the order
Bargaining power of financing companies C2 The unit is dimensionless, indicating the level of bargaining power of financing enterprises per unit time
Base Daily Interest C3 Unit: %, indicating the loan interest rate determined by the financial institution according to market conditions per unit time

Among them, because the sales revenue is after the order transaction is completed, that is, the payment can be received after about 13 weeks, and there will be a certain delay, so the delay function is used to represent the sales revenue. Most of the financing amount obtained by financing enterprises is used for production input. According to the actual situation, the ratio of production input to financing amount is set to 0.7. The loan amount is affected by the order transaction volume. The larger the order transaction volume, the more funds are required, and the financing amount is also affected by the financing rate. A7=WITHLOOKUP([(0,0)(1,1)],(0,1),(0.2,1),(0.20185,0697368),(0.4,0.7),(0.4,504386),(0.5,0.5),(0.507645,0.302623),(0.7,0.3),(0.703364,0),(0.993884,0)) \matrix{{{A_7}} \hfill & {= WITH\,LOOKUP\,\left({\left[ {\left({0,0} \right) - \left({1,1} \right)} \right],\,\left({0,1} \right),\left({0.2,\,1} \right),\,\left({0.20185,\,0697368} \right),} \right.} \hfill \cr {} \hfill & {\left({0.4,\,0.7} \right),\,\left({0.4,\,504386} \right),\,\left({0.5,\,0.5} \right),\,\left({0.507645,\,0.302623} \right),\,\left({0.7,\,0.3} \right),} \hfill \cr {} \hfill & {\left. {\left({0.703364,\,0} \right),\,\left({0.993884,\,0} \right)} \right)} \hfill \cr}

The financing rate is determined by financial institutions based on the risk situation of the entire market and the assessment of the qualifications of financing enterprises. There is no linear relationship. Generally, when the financial risk of the fresh produce online supply chain is very low, when it is 0 – 0.2. The financing rate can be as high as 100%. When the supply chain risk is low, the financing rate is 70% to 100% when the supply chain risk is 0.2 to 0.4. When the supply chain risk is high, the financing rate is 30% to 50% at 0.5 – 0.7. When the supply chain risk is high, it will be difficult to obtain loans from financial institutions. The relationship between the financial risks of the online supply chain of fresh produce. A4=0.01ln(C1) {A_4} = 0.01{ln}\left({{C_1}} \right)

This relationship equation expresses the quantitative relationship between the average transaction cost of financing companies and the order transaction volume. A5=0.015A8 {A_5} = 0.015\,{A_8} A5=0.024A3+0.1 {A_5} = 0.024\,{A_3} + 0.1

After the fresh agricultural products are produced, some enter the market for sale, and some need to be stored on their own. They are in a state of sale. In the process, there will be costs such as inventory fees and damage costs for fresh agricultural products. L1=INLEG(R1) {L_1} = IN\,LEG\,\left({{R_1}} \right)

Loan interest is charged on a daily basis, and when financing companies calculate costs, loan interest should be the cumulative amount of time, so the point function is used to represent loan interest. R1=A6*C2*C3 {R_1} = {A_6}*{C_2}*{C_3}

The change in the daily interest rate of the loan is related to the benchmark daily interest rate and the financing company's bargaining power. The benchmark daily interest rate is generally the loan interest charged by the financial institution to the financing institution based on market conditions. However, when the financing enterprise has a higher quality, or has a better Operating capacity, or through the core company's credit diffusion function, it can negotiate with financial institutions to obtain lower daily interest rates on loans, thereby bringing a win-win situation for financial institutions and themselves.

In the e-commerce platform benefit system, the revenue of the e-commerce platform is derived from the membership fees paid by the parties involved in the supply chain to the e-commerce platform and the loan commissions obtained from the promotion of financing, and its expenditure is only the transaction fees paid to promote financing and transactions. For dimensional consistency, the net profit rate of the e-commerce platform is used to represent the benefits of the e-commerce platform, then: A9=[(C4+A10A11)C4+A10]*100 {A_9} = \left[ {{{\left({{C_4} + {A_1}0 - {A_1}1} \right)} \over {{C_4} + {A_1}0}}} \right]*100

Due to the popularity of the same e-commerce platform and the trust of customers, the number of merchants in the e-commerce platform is different, especially in the field of fresh produce, the number of merchants in the e-commerce platform is even smaller, and so the membership fee C4 should be based on Specific research objects to determine.

Loan commissions are due to the smooth implementation of financing by financial institutions due to the e-commerce platform, and the remuneration paid. Different loan amounts bring different benefits to financial institutions. Therefore, financial institutions will decide to pay to e-commerce according to a certain loan commission ratio. The platform's loan commission, but this compensation is usually paid after the financing company completes the order and repays on time, so there is a certain delay. The delay time will be one week later than the financing company's sales income, that is, about 14 weeks. Delay function to represent payment of goods.

The average transaction cost of an e-commerce platform mainly includes supervision costs and basic costs. In order to facilitate the successful completion of transactions and financing, e-commerce platforms will spend certain costs on technology and supervision. This cost is mainly related to the loan amount. The higher the loan amount, the higher the supervision cost.

At the same time, in order to maintain the foundation of the system to operate, the e-commerce platform also costs some basic costs.

For the case where the covariate matrix X is full rank, the least squares methodoften used in regression analysis to fit the sample data? The estimated value of the regression coefficient of the model can be solved using formulas, but if there is a correlation between the column vectors of x, or when the column vectors are close to the correlation, the determinant is equal to or close to zero, the inverse matrix does not exist, or the fitting results are greatly different. At this time, a regularization term (L2 normal form) needs to be added to the least squares loss function to modify this problem, and the regression parameters also change from unbiased estimation to biased estimation. The objective function of ridge regression is defined as: argminJ(β)=χβy22+λβ22 {argmin}J\left(\beta \right) = \left\| {\chi \beta - y} \right\|_2^2 + \lambda \left\| \beta \right\|_2^2

The estimated solution of ridge regression can be expressed as: β=(XTX+λI)1XTY {\beta^{'}} = {\left({{X^T}X + \lambda I} \right)^{- 1}}{X^T}Y

In real life, the sample dimension of a type of data set is much larger than the number of samples, and most of the values in the data set are missing or zero (sparse data). Studies have shown that the higher the dimensionality of the data set, the larger the number of samples required for traditional linear regression analysis to predict exponential growth. In the case of insufficient samples, regression analysis prediction will become unfeasible. Therefore, as far as possible without losing the characteristic information of the data, reducing the data dimension has become the only option. Based on the regularization model, LASSO adjusts the model parameters to reduce most of the regression coefficients of the model to zero and discards the variables corresponding to the model coefficients of zero to achieve the purpose of feature selection. Different from common feature extraction methods such as PCA, feature selection retains the original features, while feature extraction recombines the original features to form new features. At present, the LASO method has been widely used in high-dimensional data preprocessing.

It mainly focuses on the regularization model combining three different permutation test methods, and empirically studies the differences in the ability of the two regularization models to discover Markov edges. The regularization models used in this experiment are ridge regression model, LASO model and NVRMLM.

This experimental data set combines some of the data sets used in the experiments in Chapters 3 and 4. There are two levels of meaning here. One is to consider the impact of different permutation test methods on the final result. The other is to observe the collinearity and non-linearity. In-line dataset, the effect of different regularization models on experimental results. The specific attributes of the data set are summarized in Table 2.

Data attribute table

data set Alarm Alarm 10 Chilld Insurance
Number of nodes 36 360 20 28
MB 8/1 8/1 8/1 10/1
Results analysis

Extreme testing is an indispensable part of Vensim simulation. It is mainly used to check whether the equations in the model have stability. In extreme cases, they also conform to the laws of reality. Extreme tests generally select certain variables for their limit values, such as ”0” or ”infinite”.

According to the actual situation, this article selects the extreme transaction with the order transaction volume of ”0”. When the order transaction volume is ”0”, the financing company does not need to obtain funds from financial institutions at this time. Without a financing process, E-commerce platforms also cannot obtain loan commissions. At this time, the loan commission should also be ”0”. Because no financing is conducted, the risks of e-commerce platforms will be reduced accordingly, and the risks of financial institutions will be reduced accordingly. At the same time, the online supply chain financial risks will be reduced accordingly. The results shown in Figures 2 and 3 are obtained.

Figure 2

The loan commission change chart

Figure 3

Risk change chart of fresh agricultural products online supply chain

As can be seen from the figure, when the order transaction volume is set to 0, the loan commission is always 0 during the simulation period, and the financial risk of the online supply chain is also reduced, and the behavior of each variable is in line with the reality.

In the process of constructing the model, only the main influencing factors were analyzed, and many unknown parameters were ignored. The sensitivity test is to perform a sensitivity analysis on the selected parameters to detect whether they are the main influencing factors of other variables in the system, whether it is sensitive, the initial value of the parameter can be adjusted according to the test results to meet the requirements of the simulation. In this paper, a sensitivity simulation test is conducted with the quality of financing companies as a variable to observe theimpact of changes on the system.

With different initial values of the quality of financing companies, the risks of financing companies and the changes in the financial risks of online supply chains of fresh produce are shown in Figures 4 and Figures 5. From the figure, it is found that when the quality of financing companys decreases, the risks of financing companies continue to increase. The financial risk of the online supply chain of fresh produce is also on the rise, which is in line with reality.

Figure 4

Risk change chart of financing enterprises under different quality of financing enterprises

Figure 5

Financial risk change chart of fresh agricultural product supply chain under different quality of financing enterprises

In the case-based simulation, as time goes by, under the performance of financing companies, the financial data change trend of the online supply chain of fresh agricultural products of financing companies, e-commerce platforms, and financial institutions is shown in Figure 6.

Figure 6

Risk change chart in online supply chain finance of fresh agricultural products

Observed from the figure above, it is found that the risks of financing enterprises and the financial risks of the online supply chain of fresh agricultural products are on the rise, and the risks of e-commerce platforms and financial institutions are on the decline. In the early stage of financing, e-commerce platforms and financial institutions need to pay certain costs for undertaking the review and credit work. Therefore, they face higher risks. In order to maintain their status, e-commerce platforms will spend more transaction costs. Platform risks are higher than financial institutions. At the end of the financing period, as the repayment time approaches, the financing company will default. At this time, the financing company's risk is getting higher and higher, which makes the online supply chain financial risk of fresh farmers' products at a higher level. It can be concluded that the risk of financing enterprises is the main factor affecting the financial risk of the online supply chain of fresh produce.

The risks of financing enterprises, the risks of e-commerce platforms, and the financial risks of online supply chains of fresh produce and agricultural products all show an upward trend. Among them, the risks of financing enterprises and the risks of financial institutions have slowly increased over time. Defaults are prone to default during the repayment period, that is, the financing company and the e-commerce platform conspire to obtain benefits together, so the risks of the financing company and the e-commerce platform are at a high level in the later stage, which affects the financial risks of the online supply chain of fresh produce. So that it is also at a high level.

Investigate the size of the F-Score values of the LASSA-P model, RRM-P, and MRRM-P under the three permutation test methods. On the whole, the original data method and the full model method of the three methods are better than the simplified model method corresponding to the three methods. Perhaps the three data sets mentioned above do not meet the applicable conditions of the simplified model. The model's raw data method is also slightly better than the full model method. As far as the model is concerned, the overall performance of LASS0-P is slightly better on the above three types of discrete multi-valued data sets, followed by RRM-P, and NVRMLM-P is the worst. This conclusion is somewhat different from the conclusion of the binary data set in Chapter 3. Different. From the perspective of the replacement test method, the best method should be the original data method, followed by the full model method, and again the simplified model method.

In addition, from the implementation process of the simplified model method, itcan be seen that the calculation cost of this algorithm is higher than the original data method and the full model method, and the constraints are harsh. Therefore, it is not applicable to Markov edge discovery or P-value calculation. For the data set Almml0, s1000vl, the data dimension is as high as 370, the number of samples is 1,000, and the calculation cost is very high. The author uses a relatively highend desktop computer to run the three replacement inspection methods for nearly 24 hours, but no results. At the same time, it also shows that the locality of the replacement test method is not suitable for high-dimensional multi-sample data sets.

Figures 7, Figures 8, and Figures 9 show that on binary discrete data sets, in addition to the difference in computing time.

Figure 7

Relationship between MRLM (-P) and RRLM (-P) F-Scroe

Figure 8

The relationship between the accuracy rates of MRLM (-P) and RRLM (-P)

Figure 9

The relationship between the running time of MRLM (-P) and RRLM (-P)

In terms of accuracy and overall performance (F-Score), the discovery efficiency of MRRM (or MRLM-P) and RRLM (or RRLM-P) Markov edges (subsets) is basically similar; on continuous data sets, the discovery efficiency of MRRM Markov edges (subsets) is much higher than that of ridge regression models. It can also be seen in the figure that after using the replacement test for the two regression models, their respective discovery efficiency has been significantly improved. For example, on a continuous data set, the F-Score value of the MRLM using the T-distribution method is below 0.5, and the F-Score value of the MRLM-P using the permutation test method is rapidly increased to above 0.9, indicating that the permutation test is effective for the model. Finding efficiency plays an important role. From the calculation time, the calculation time of MRLM and RRLM are both very small and very similar. After combining the permutation test, the calculation time of MRLM-P and RRLM-P increases significantly, and the calculation time gradually increases with the increase of the number of samples.

In the case-based simulation, as time goes by, under the state of default of the financing company, the financial data change trend of the online supply chain of fresh agricultural products of financing companies, e-commerce platforms, and financial institutions is shown in Figure 10.

Figure 10

Risk change chart in online supply chain finance of fresh agricultural products

Observed from the above chart, it is found that the risks of financing companies, the risks of e-commerce platforms, and the financial risks of online supply chains of fresh produce and agricultural products all show an upward trend. Among them, the risks of financing companies and financial institutions have slowly increasedover time. Since the financing company is prone to default in the final repayment period, that is, the financing company and the e-commerce platform conspire to obtain benefits together, so the risks of the financing company and the e-commerce platform are at a relatively high level in the later stage, and the fresh produce online The impact of supply chain financial risks has put it at a high level.

In the simulation operation of the default of the financing company, the financing company does not pay the loan interest to the financial institution, and even does not repay at the end. Therefore, the return of the financing company has always been higher than that during the performance period, but it has also shown a trend of segmental growth, and at the end of the financing period, the highest net interest margin of the financing company was obtained.

In the early stage of financing, the e-commerce platform has increased its net interest income through the income of membership fees. However, due to the default of the financing company in the late financing period, the e-commerce platform cannot obtain a loan commission, and its net profit rate has been in a downward trend. And at the end showed a straight downward trend.

When a financing company defaults, the financial institution's loan amount cannot be recovered and there is no income from the daily interest rate of the loan.

However, in order to ensure their own interests, the financial institution has set a minimum net interest rate of return. Therefore, during the financing period, the financial institution's income has been at Constant and lowest state.

As the financing rate of financing companies receiving loans is affected by the financial risks of the online supply chain of fresh agricultural products, with the rise of the financial risks of the online supply chain of fresh agricultural products, the net interest rate of return of the financing companies is in a state of segmental change, but With the completion of the order volume in the financing process, the financing company has been profitable, so it has been showing a growing trend. At the end of the financing period, the financing company completes all orders, and the net interest rate of return of the financing company reaches the maximum.

The risks faced by financial institutions are in a relatively stable state. At the same time, the returns of financial institutions are also in a relatively stable and slowly increasing trend. Since the main income of financial institutions is not derived from interest income on loans, the net interest margin of financial institutions is at a relatively low level compared to financing companies and e-commerce platforms. As the financing rate is affected by the financial risk of online supply of fresh agricultural products, as the financial risk of the online supply chain of fresh agricultural products increases, the financing rate slowly declines, and the amount of loans obtained by financing companies decreases. It is in a state of segmental change and occasional decline, but with the completion of order volume in the financing process, the e-commerce platform has always been profitable, and at the end of the financing period, financing companies complete all orders, and the net profit margin of the e-commerce platform presents.

The literature reviews the operating mode, influencing factors, and risk control measures of the online supply chain of fresh agricultural products. Based on the financial supply chain theory and system dynamics theory of the online supply chain of fresh agricultural products, it builds a healthy product through the financing process. A system dynamics model of the financial risk of the online supply chain of fresh agricultural products. The benefits and risks faced by all parties in the online supply chain of fresh agricultural products are studied.

Through a qualitative linkage analysis of the online supply chain's financial risk system for fresh produce, it was found that changes in risk were mainly affected by the linkage of interest systems, including the benefit system of financing enterprises, the benefit system of e-commerce platforms, and the benefit system of financial institutions. Each benefit system has an impact on the overall risk through the risks of financing companies, e-commerce platform risks, financial institution risks and other risks.

Through a quantitative linkage analysis of the financial risk system for the online supply chain of fresh produce, it was found that the risks of financing enterprises, e-commerce platform risks, financial institution risks and other risks have an impact on the overall risk, which accounted for 53%, 24%, 12% and 11%.

Financing enterprise risk is an important factor affecting the financial risk of the online supply chain of fresh agricultural products. At the same time, e-commerce platforms also play an important role in the risk system. Therefore, financing companies should determine the appropriate input cost based on the results of the risk analysis; e-commerce platforms The information of financing companies should be carefully reviewed, and they should assume regulatory responsibilities during the transaction process; while financial institutions should carefully select cooperative e-commerce platforms and determine appropriate financing rates based on risk prediction results, thereby effectively reducing financial risks in online supply chains.

Conclusion

By analyzing the influencing factors of the main fruit export market trade, and obtaining the influence degree and changing trend of each factor, the structural effect is the most important factor affecting Thailand's exports to the main fruit export market in the entire period of 2003–2019, which promotes the increase in Thailand's fruit export value indicates that the demand for fruit imports in major export markets has expanded. In addition, the huge disadvantage in competitiveness is the main factor hindering the growth of Thai fruit exports. The discovery of Markov edges (subsets) based on the method of regularization model is different from the selection of feature variables of traditional regularization models. The significance of the work is to ensure that the selected feature variable is Marco in theory. Husband side (subset). Improvements in the lack of MRLM and the empirical conclusions of regularization models have contributed new knowledge in this field, providing new theoretical support points for subsequent discoveries based on Markov edges of regularization models.

Figure 1

Network Security Situation Awareness Model
Network Security Situation Awareness Model

Figure 2

The loan commission change chart
The loan commission change chart

Figure 3

Risk change chart of fresh agricultural products online supply chain
Risk change chart of fresh agricultural products online supply chain

Figure 4

Risk change chart of financing enterprises under different quality of financing enterprises
Risk change chart of financing enterprises under different quality of financing enterprises

Figure 5

Financial risk change chart of fresh agricultural product supply chain under different quality of financing enterprises
Financial risk change chart of fresh agricultural product supply chain under different quality of financing enterprises

Figure 6

Risk change chart in online supply chain finance of fresh agricultural products
Risk change chart in online supply chain finance of fresh agricultural products

Figure 7

Relationship between MRLM (-P) and RRLM (-P) F-Scroe
Relationship between MRLM (-P) and RRLM (-P) F-Scroe

Figure 8

The relationship between the accuracy rates of MRLM (-P) and RRLM (-P)
The relationship between the accuracy rates of MRLM (-P) and RRLM (-P)

Figure 9

The relationship between the running time of MRLM (-P) and RRLM (-P)
The relationship between the running time of MRLM (-P) and RRLM (-P)

Figure 10

Risk change chart in online supply chain finance of fresh agricultural products
Risk change chart in online supply chain finance of fresh agricultural products

Data attribute table

data set Alarm Alarm 10 Chilld Insurance
Number of nodes 36 360 20 28
MB 8/1 8/1 8/1 10/1

Variable set of financing system

name Code Description
Daily interest on loans L1 Unit: 10,000 yuan, which means the interest paid by the financing enterprise to the financial institution during the financing process
Daily interest rate of loans R1 The unit is dimensionless, indicating the change in daily interest rate of the loan per unit time
Financing companies' net interest margin A1 Unit: %, representing the income of the financing company
Sales revenue A2 Unit: 10,000 yuan, representing sales in come of financing enterprises
Production input cost A3 Unit: 10,000 yuan, indicating the cost of production input of the financing enterprise
Financing companies average transaction costs A4 Unit: 10,000 yuan, which means the transaction cost of packaging, transportation, etc, that the financing company pays to complete the order
Average storage fee A5 Unit: 10,000 yuan, which means the financing company keeps expenses for storage
Loan amount A6 Unit: 10,000 yuan, indicating the amount of financing obtained by the financing company
Financing rate A7 Units are dimensionless, indicating the credit ratio determined by financial institutions
Number of agricultural products A8 Unit: Ton, indicating the amount of agricultural products produced by the financing company
Order volume C1 Unit: Ton, indicating the quantity of agricultural products agreed to be sold in the order
Bargaining power of financing companies C2 The unit is dimensionless, indicating the level of bargaining power of financing enterprises per unit time
Base Daily Interest C3 Unit: %, indicating the loan interest rate determined by the financial institution according to market conditions per unit time

Ahmad R, Farooqi A, Zhang J, et al. Steady flow of a power law fluid through a tapered non-symmetric stenotic tube[J]. Applied Mathematics and Nonlinear Sciences, 2019, 4(1):249–260. AhmadR FarooqiA ZhangJ Steady flow of a power law fluid through a tapered non-symmetric stenotic tube[J] Applied Mathematics and Nonlinear Sciences 2019 4 1 249 260 10.2478/AMNS.2019.1.00022 Search in Google Scholar

Harisha, Ranjini P S, Lokesha V, et al. Degree Sequence of Graph Operator for some Standard Graphs[J]. Applied Mathematics and Nonlinear Sciences, 2020, 5(2):99–108 Harisha RanjiniP S LokeshaV Degree Sequence of Graph Operator for some Standard Graphs[J] Applied Mathematics and Nonlinear Sciences 2020 5 2 99 108 10.2478/amns.2020.2.00018 Search in Google Scholar

Ohsawa Y, Hayashi T. Tangled string for sequence visualization as fruit of ideas in Innovators Marketplace on Data Jackets[J]. Intelligent decision technologies, 2016, 10(3):235–247 OhsawaY HayashiT Tangled string for sequence visualization as fruit of ideas in Innovators Marketplace on Data Jackets[J] Intelligent decision technologies 2016 10 3 235 247 10.3233/IDT-150251 Search in Google Scholar

Iriarte A, P Yáez, Villalobos P, et al. Carbon Footprint of Southern Hemisphere Fruit Exported to Europe: The Case of Chilean Apple to the UK[J]. Journal of Cleaner Production, 2021, 293(6):126118 IriarteA YáezP VillalobosP Carbon Footprint of Southern Hemisphere Fruit Exported to Europe: The Case of Chilean Apple to the UK[J] Journal of Cleaner Production 2021 293 6 126118 10.1016/j.jclepro.2021.126118 Search in Google Scholar

Cicero N, Albergamo A, Salvo A, et al. Chemical characterization of a variety of cold-pressed gourmet oils available on the Brazilian market[J]. Food Research International, 2018, 109(jul.):517–525 CiceroN AlbergamoA SalvoA Chemical characterization of a variety of cold-pressed gourmet oils available on the Brazilian market[J] Food Research International 2018 109 jul. 517 525 10.1016/j.foodres.2018.04.06429803478 Search in Google Scholar

Della L J. Tailoring interventions for increasing fruit and vegetable consumption among urban and rural African American audiences: Applying market decision-analysis to the intervention design processes[J]. Fire Safety Journal, 2015, 23(4):413–438 DellaL J Tailoring interventions for increasing fruit and vegetable consumption among urban and rural African American audiences: Applying market decision-analysis to the intervention design processes[J] Fire Safety Journal 2015 23 4 413 438 Search in Google Scholar

Bo Y L, Shamsudin R, Baharudin B, et al. Performance Study of a Jatropha Curcas L. Fruit Shelling Machine for Kernel Recovery in Biodiesel Production[J]. Applied Engineering in Agriculture, 2015, 31:755–765 BoY L ShamsudinR BaharudinB Performance Study of a Jatropha Curcas L. Fruit Shelling Machine for Kernel Recovery in Biodiesel Production[J] Applied Engineering in Agriculture 2015 31 755 765 10.13031/aea.31.11211 Search in Google Scholar

Simmons D. Tracing the Impact of Baby-Boom Generation on the Economic Market Is No Trivial Pursuit[J]. Journal of Contaminant Hydrology, 2015, 182(3):287–96 SimmonsD Tracing the Impact of Baby-Boom Generation on the Economic Market Is No Trivial Pursuit[J] Journal of Contaminant Hydrology 2015 182 3 287 96 Search in Google Scholar

Gao Z, Pang G, Li L, et al. Design of hand-operated piggyback jaw gripper type simplified picker for pear[J]. Nongye Gongcheng Xuebao/Transactions of the Chinese Society of Agricultural Engineering, 2019, 35(1):39–45 GaoZ PangG LiL Design of hand-operated piggyback jaw gripper type simplified picker for pear[J] Nongye Gongcheng Xuebao/Transactions of the Chinese Society of Agricultural Engineering 2019 35 1 39 45 Search in Google Scholar

Sanderson V, Bamber N, Pelletier D N. Cradle-to-market life cycle assessment of Okanagan (Canada) cherries: Helicopters, seasonal migrant labour and flying fruit - ScienceDirect[J]. Journal of Cleaner Production, 2019, 229:1283–1293 SandersonV BamberN PelletierD N Cradle-to-market life cycle assessment of Okanagan (Canada) cherries: Helicopters, seasonal migrant labour and flying fruit - ScienceDirect[J] Journal of Cleaner Production 2019 229 1283 1293 10.1016/j.jclepro.2019.04.398 Search in Google Scholar

RA Sánchez-Ramírez, Charles V, MC González-Araya, et al. Measuring the performance of a dehydration plant of apples[J]. Applied Mathematical Modelling, 2016, 40(21–22):9118–9130 Sánchez-RamírezRA CharlesV González-ArayaMC Measuring the performance of a dehydration plant of apples[J] Applied Mathematical Modelling 2016 40 21–22 9118 9130 10.7835/ccwp-2015-10-0025 Search in Google Scholar

Recommended articles from Trend MD

Plan your remote conference with Sciendo