Accident economic loss influence factor analysis method and system

Abstract

The invention relates to an accident economic loss influence factor analysis method and system, and the method part comprises the steps: firstly building a plurality of random parameter models based on a GB2 model, then carrying out the effect comparison through employing a LOOIC value, and selecting a proper random parameter model for analysis, obtaining positive influence or negative influence of various explanatory variables on accident economic loss; and finally, based on sensitivity analysis, mining the influence degree of the explanatory variable on the accident economic loss. Through the scheme of the invention, an analysis method for objectively, comprehensively and accurately mining accident economic loss influence factors is provided, the application range is wide, and the influence of related factors on the economic loss caused by the accident can be better represented.

Description

Technical Field

[0001] The present invention relates to the field of data science and technology, and in particular to a method and system for analyzing factors affecting economic losses caused by accidents. Background Art

[0002] Unlike the severity of the accident, the economic losses of fishing boat accidents are continuous, and the analysis results of the economic losses of fishing boat accidents provide detailed information on the loss of human life and the loss of ship property. Therefore, it is crucial to analyze the impact of influencing factors on the economic losses caused by fishing boat accidents through technical means.

[0003] The economic loss of a ship accident is a non-negative continuous variable. The distribution of such catastrophic economic losses obviously deviates from normality, often showing right skewness, excessive kurtosis and heavy-tail characteristics. In the prior art, the Tobit regression method, the generalized F distribution model, the Bayesian lognormal regression model, etc. are used to explain the non-negativity of property losses and total losses, but these technical solutions largely ignore the problem of heavy-tail distribution, thus showing that they are insufficient in accurately describing the complex distribution characteristics of economic losses and lack flexibility in heavy-tail distribution modeling.

[0004] In addition, the occurrence of fishing vessel accidents is extremely complex and involves many factors, and it is impossible to cover all safety factors. Unobserved safety factors may have different effects on variables related to these unobserved factors, that is, the problem of unobserved heterogeneity. However, most of the existing models constructed by the fishing vessel accident analysis methods assume that the distribution mean of each random parameter remains unchanged between observations, ignoring the bias brought by unobserved heterogeneity to the estimated parameters, which will cause model specification errors, lead to inaccurate inferences, and may underestimate or overestimate the direct marginal effects. Summary of the invention

[0005] The technical problem to be solved by the present invention is how to solve the problems existing in the analysis methods of the prior art, such as the difficulty in handling the analysis of influencing factors when the economic losses of ship accidents present a heavy-tailed distribution and the problem of ignoring unobserved heterogeneity causing deviations in estimated parameters. In order to overcome the above defects of the prior art, the present invention provides a method for analyzing the influencing factors of economic losses caused by accidents and a system for analyzing the influencing factors of economic losses caused by accidents.

[0006] The present invention provides a method for analyzing factors affecting economic losses caused by accidents, comprising the following steps:

[0007] S1: Constructing the Bayesian random parameter GB2 model based on the observation information of sample characteristics of captured fishing vessel accidents;

[0008] S2: constructing multiple random parameter models for analyzing the factors affecting the economic losses of fishing boat accidents based on the Bayesian random parameter GB2 model and the observation information of sample characteristics;

[0009] S3: Calculating the leave-one-out cross-validation information criterion value of each of the random parameter models by a calculation module;

[0010] S4: Select the random parameter model with a low value of the leave-one-out cross-validation information criterion to analyze the factors affecting the economic losses of fishing vessel accidents, and obtain the positive or negative impact of various explanatory variables on the economic losses of fishing vessel accidents;

[0011] S5: Based on the positive or negative impact of various explanatory variables on the economic losses caused by fishing boat accidents obtained in step S4, the degree of influence of the explanatory variables on the economic losses caused by accidents is explored.

[0012] The method for analyzing factors affecting economic losses caused by accidents disclosed in the present invention aims at the above problems. First, a Bayesian random parameter GB2 model is constructed through observation information. The model can solve the problems of highly non-normal distribution and heavy-tailed distribution. Then, based on the model, multiple random parameter models are established to improve statistical fit and provide more insights. Then, the leave-one-out cross-validation information criterion value is used for effect comparison, and a suitable random parameter model is selected for analysis, and the positive or negative influence of various explanatory variables on the economic losses caused by accidents is obtained. Finally, based on sensitivity analysis, the influence of explanatory variables on the economic losses caused by accidents is excavated, thereby overcoming the problems existing in the analysis methods of the prior art, such as the difficulty in analyzing factors affecting economic losses caused by ship accidents with heavy-tailed distribution and the bias caused by ignoring unobserved heterogeneity in estimating parameters. The method is objective, comprehensive and accurate, can explore factors affecting economic losses caused by accidents, has a wide range of applications, and can better characterize the influence of relevant factors on the economic losses caused by accidents.

[0013] In a possible implementation, in the Bayesian random parameter GB2 model:

[0014] The probability density function of a random variable is as follows:

[0015]

[0016] In the formula,

[0017] a is the shape parameter, which determines the peak and tail thickness of the economic loss distribution;

[0018] p is the probability of an accident with smaller economic losses;

[0019] q is the probability of a major economic loss accident;

[0020] B(p,q) is the B function;

[0021] λ reflects the level of economic loss of the accident and is a scale parameter that controls the overall level and scope of the loss. Its calculation formula is as follows:

[0022] λ=exp(β′x)=exp(β0+β1x1+β2x2+...+β k x k ),

[0023] In the formula,

[0024] x is the vector of covariates that affect the economic losses of fishing boat accidents;

[0025] β is the coefficient vector corresponding to x;

[0026] k is the number of covariates;

[0027] The kth moment of a random variable is given by:

[0028]

[0029] This scheme can associate the distribution mean and variance of the random parameters with the explanatory variables, and there are differences between different observations, which further improves the model performance and effectively solves the problems of highly non-normal distribution and heavy-tailed distribution.

[0030] In a possible implementation, the multiple random parameter models in step S2 include a general random parameter model, a mean heterogeneity random parameter model, and a mean and variance heterogeneity random parameter model; which helps to explain the unobserved heterogeneity caused by unobserved safety factors, thereby improving statistical fit and providing more insights.

[0031] In one possible implementation, the general random parameter model is expressed as follows:

[0032]

[0033] In the formula,

[0034] β n is a random parameter vector;

[0035] b is the average parameter vector of all observations of the sample characteristics of fishing boat accidents;

[0036] is a randomly distributed error term used to capture possible heterogeneity.

[0037] In a possible implementation, the mean heterogeneity random parameter model is expressed as follows:

[0038]

[0039] In the formula,

[0040] Z n To influence β n vector of explanatory variables for the mean;

[0041] Θ is Z n The corresponding estimable parameter vector; if the elements in Θ are significantly different from zero, it means that the mean heterogeneity random parameter model has mean heterogeneity.

[0042] In one possible implementation, the mean and variance heterogeneity random parameter model is expressed as follows:

[0043]

[0044] In the formula,

[0045] W n is a vector of explanatory variables used to capture the heterogeneity of the standard deviation σ;

[0046] ω is W n The corresponding estimable parameter vector;

[0047] If no element in ω is significantly different from zero, then the random parameter model with heterogeneity in mean and variance only has heterogeneity in the mean;

[0048] If no element in Θ is significantly different from zero, it means that the mean and variance heterogeneity random parameter model degenerates to the general random parameter model.

[0049] In a possible implementation, in step S3, the calculation formula for obtaining the leave-one-out cross-validation information criterion value by the calculation module is as follows:

[0050] LOOIC=-2ELPD LOO ,

[0051]

[0052] In the formula,

[0053] p M (y i |θ) represents the M likelihood in the regression model;

[0054] p M (θ|y -i ) represents the posterior value of the parameter θ, where the observed value y i Being excluded;

[0055] This scheme provides a calculation formula for the leave-one-out cross-validation information criterion value to ensure effect comparison, thereby selecting a suitable random parameter model for analysis and obtaining the positive or negative impact of various explanatory variables on the economic losses of accidents.

[0056] Another technical solution of the present invention is to provide a system for analyzing factors affecting economic losses caused by accidents, including:

[0057] The model module is configured to construct a Bayesian random parameter GB2 model based on the observed information of sample characteristics of captured fishing vessel accidents;

[0058] A parameter model module is configured to construct a plurality of random parameter models for analyzing factors affecting economic losses of fishing boat accidents based on the Bayesian random parameter GB2 model and observation information of sample characteristics;

[0059] A calculation module is configured to calculate the LOOIC value of each of the random parameter models respectively;

[0060] A selection module is configured to select the random parameter model with a lower leave-one-out cross-validation information criterion value to analyze the factors affecting economic losses caused by fishing vessel accidents, and obtain the positive or negative effects of various explanatory variables on the economic losses caused by fishing vessel accidents;

[0061] The analysis module is set to the positive or negative impact of various explanatory variables on the economic losses of fishing boat accidents, and the influence of the explanatory variables on the economic losses of accidents is excavated.

[0062] The accident economic loss influencing factor analysis system disclosed in the present invention aims at the above problems. By setting a model module, the observation information is used to construct a Bayesian random parameter GB2 model, which can solve the problems of highly non-normal distribution and heavy-tailed distribution. On this basis, a parameter model module is set, and multiple random parameter models can be established based on the model, which can improve the statistical fit and provide more insights. By setting a calculation module and a selection module, in cooperation with the two, the leave-one-out cross-validation information criterion value can be used to compare the effects, select a suitable random parameter model for analysis, and obtain the positive or negative effects of various explanatory variables on the accident economic loss. Finally, the analysis module is used to dig out the influence of the explanatory variables on the accident economic loss based on sensitivity analysis, thereby overcoming the problems existing in the analysis method of the prior art that it is difficult to handle the analysis of the influencing factors when the economic loss of ship accidents is in a heavy-tailed distribution and ignoring the unobserved heterogeneity to bring deviations to the estimated parameters. The system is objective, comprehensive and accurate, can dig out the influencing factors of the accident economic loss, has a wide range of applications, and better characterizes the influence of relevant factors on the economic losses caused by accidents. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Figure 1 This is a flow chart of a method for analyzing factors affecting economic losses caused by accidents disclosed in an embodiment of the present application;

[0064] Figure 2 This is a schematic diagram of the structure of a system for analyzing factors affecting economic losses caused by accidents disclosed in an embodiment of the present application. DETAILED DESCRIPTION

[0065] First, those skilled in the art should understand that these implementations are only used to explain the technical principles of the embodiments of the present application, and are not intended to limit the protection scope of the embodiments of the present application. Those skilled in the art can make adjustments to them as needed to adapt to specific application scenarios.

[0066] In the description of the embodiments of the present application, it should be noted that, unless otherwise clearly specified and limited, the terms "electrical connection" and "electrical connection relationship" should be understood in a broad sense, that is, referring to a connection method with an electrical relationship, for example, it can be a circuit connection through a wire, or it can be an electrical connection through a radio signal channel (channel), or a combination of the two. In addition, "electrical connection" and "electrical connection relationship" can be based on mechanical connection (such as a wire set in a connection key); it can be a direct connection or an indirect connection through an intermediate medium. For ordinary technicians in this field, the specific meanings of the above terms in the embodiments of the present application can be understood according to specific circumstances.

[0067] Two embodiments will be used below, and the present application will be further described in detail in conjunction with the accompanying drawings and specific embodiments.

[0068] Embodiment 1:

[0069] See also Figure 1 As shown, the embodiment of the present application discloses a method for analyzing factors affecting economic losses caused by accidents, and the method comprises the following steps:

[0070] S1: Construct a Bayesian random parameter GB2 model based on the observational information of sample characteristics of captured fishing vessel accidents.

[0071] In this example, the economic loss of fishing boat accidents is taken as the response variable, and the distribution parameter is specified as a function of the covariate, thereby mining the relationship between the accident economic loss data and the explanatory variables. On this basis, it is assumed that the scale parameter λ can change with the covariate, and then the Bayesian random parameter GB2 model is obtained: the probability density function of the random variable is as follows:

[0072]

[0073] In the formula,

[0074] a is the shape parameter, which determines the peak and tail thickness of the economic loss distribution;

[0075] p is the probability of an accident with smaller economic losses;

[0076] q is the probability of a major economic loss accident;

[0077] B(p,q) is the B function;

[0078] λ reflects the level of economic loss of the accident and is a scale parameter that controls the overall level and scope of the loss. Its calculation formula is as follows:

[0079] λ=exp(β′x)=exp(β0+β1x1+β2x2+...+β k x k ),

[0080] In the formula,

[0081] x is the vector of covariates that affect the economic losses of fishing boat accidents;

[0082] β is the coefficient vector corresponding to x;

[0083] k is the number of covariates.

[0084] At the same time, in the Bayesian random parameter GB2 model: the kth moment of the random variable is as follows:

[0085]

[0086] S2: Based on the Bayesian random parameter GB2 model and the observation information of sample characteristics, multiple random parameter models are constructed to analyze the factors affecting the economic losses of fishing boat accidents.

[0087] In this embodiment, the multiple random parameter models in step S2 include a general random parameter model, a mean heterogeneity random parameter model, and a mean and variance heterogeneity random parameter model, wherein the general random parameter model (hereinafter referred to as RPM) is expressed as follows:

[0088]

[0089] In the formula,

[0090] β n is a random parameter vector;

[0091] b is the average parameter vector of all observations of the sample characteristics of fishing boat accidents;

[0092] is a randomly distributed error term used to capture possible heterogeneity.

[0093] The random parameter model with heterogeneity in mean (hereinafter referred to as RPMHM) is described as follows:

[0094]

[0095] In the formula,

[0096] Z n To influence β n vector of explanatory variables for the mean;

[0097] Θ is Z n The corresponding estimable parameter vector; if the elements in Θ are significantly different from zero, it means that the mean heterogeneity random parameter model has mean heterogeneity.

[0098] The random parameter model with heterogeneity of mean and variance (hereinafter referred to as RPMHMV) is expressed as follows:

[0099]

[0100] In the formula,

[0101] W n is a vector of explanatory variables used to capture the heterogeneity of the standard deviation σ;

[0102] ω is W n The corresponding estimable parameter vector; if no element in ω is significantly different from zero, it means that the mean and variance heterogeneity random parameter model only has mean heterogeneity; if no element in Θ is significantly different from zero, it means that the mean and variance heterogeneity random parameter model degenerates to the general random parameter model.

[0103] S3: Calculate the leave-one-out cross-validation information criterion value of each of the random parameter models through a calculation module.

[0104] The leave-out cross-validation information criterion (LOOIC) value is twice the expected logarithmic prediction density of the leave-out cross-validation. Specifically, in step S3, the calculation module obtains the LOOIC value by the following formula:

[0105] LOOIC=-2ELPD LOO ,

[0106]

[0107] In the formula,

[0108] p M (y i |θ) represents the M likelihood in the regression model;

[0109] p M (θ|y -i ) represents the posterior value of the parameter θ, where the observed value y i are excluded.

[0110] S4: Select a random parameter model with a lower LOOIC value to analyze the influencing factors of economic losses from fishing vessel accidents, and obtain the positive or negative impact of various explanatory variables on the economic losses from fishing vessel accidents.

[0111] The following table shows the observation information of the sample characteristics of the fishing boat accidents used in this embodiment. The comparative analysis results using this information are shown in the following table:

[0112]

[0113]

[0114]

[0115] The estimated value is expressed as mean (standard deviation); the Bayesian confidence interval is [2.50%, 97.50%], excluding 0; ** indicates that the significance level of 5% has been reached. As can be seen from the table, the LOOIC value of the RPMHM model is the smallest, and then the random parameter model is selected to analyze the factors affecting the economic losses of fishing boat accidents, and obtain the positive or negative effects of various explanatory variables on the economic losses of fishing boat accidents.

[0116] S5: Based on the positive or negative impact of various explanatory variables on the economic losses of fishing boat accidents obtained in step S4, the degree of influence of the explanatory variables on the economic losses of accidents is explored.

[0117] The method for analyzing factors affecting economic losses caused by accidents disclosed in this embodiment first constructs a Bayesian random parameter GB2 model through observation information, and the model can solve the problems of highly non-normal distribution and heavy-tailed distribution; then, based on the model, multiple random parameter models (RPM, RPMHM and RPMHMV) are established, which can improve statistical fit and provide more insights; then, the LOOIC value is used to compare the effects, and a suitable random parameter model is selected for analysis, and the positive or negative influence of various explanatory variables on the economic losses caused by accidents is obtained; finally, based on sensitivity analysis, the influence of the explanatory variables on the economic losses caused by accidents is excavated, thereby overcoming the problems existing in the analysis methods of the prior art, such as the difficulty in analyzing factors affecting economic losses caused by ship accidents with heavy-tailed distribution and the neglect of unobserved heterogeneity to bring deviations to estimated parameters. The method is objective, comprehensive and accurate, can explore factors affecting economic losses caused by accidents, has a wide range of applications, and can better characterize the influence of relevant factors on the economic losses caused by accidents.

[0118] Embodiment 2:

[0119] See also Figure 2 As shown, this embodiment further discloses a system for analyzing factors affecting economic losses due to accidents on the basis of the first embodiment, the system comprising a model module, a parameter model module, a calculation module, a selection module and an analysis module, wherein the parameter model module is electrically connected to the model module, the calculation module is electrically connected to the parameter model module, the selection module is electrically connected to the calculation module, and the analysis module is electrically connected to the selection module.

[0120] In the system, the model module is configured to construct a Bayesian random parameter GB2 model based on the observation information of sample characteristics of obtained fishing boat accidents; the parameter model module is configured to construct a plurality of random parameter models for analyzing the influencing factors of economic losses of fishing boat accidents based on the Bayesian random parameter GB2 model and the observation information of sample characteristics; the calculation module is configured to calculate the LOOIC value of each of the random parameter models respectively; the selection module is configured to select the random parameter model with a lower LOOIC value to analyze the influencing factors of economic losses of fishing boat accidents, and obtain the positive or negative influence of various explanatory variables on the economic losses of fishing boat accidents; the analysis module is configured to determine the positive or negative influence of various explanatory variables on the economic losses of fishing boat accidents, and to dig out the degree of influence of the explanatory variables on the economic losses of accidents.

[0121] The accident economic loss influencing factor analysis system disclosed in this embodiment constructs a Bayesian random parameter GB2 model with observed information by setting a model module, and the model can solve the problems of highly non-normal distribution and heavy-tailed distribution; on this basis, a parameter model module is set, and multiple random parameter models can be established based on the model, which can improve statistical fit and provide more insights; by setting a calculation module and a selection module, in cooperation with the two, the leave-one-out cross-validation information criterion value can be used to compare the effects, select a suitable random parameter model for analysis, and obtain the positive or negative effects of various explanatory variables on the accident economic loss; finally, through the analysis module based on sensitivity analysis, the influence of the explanatory variables on the accident economic loss is excavated, thereby overcoming the problems existing in the analysis method of the prior art that it is difficult to handle the analysis of the influencing factors when the economic loss of ship accidents is in a heavy-tailed distribution and ignoring the unobserved heterogeneity to bring deviations to the estimated parameters. The system is objective, comprehensive and accurate, can explore the influencing factors of accident economic losses, has a wide range of applications, and better characterizes the impact of related factors on the economic losses caused by accidents.

[0122] In the description of the embodiments of the present application, it should be noted that in the description of the present application, terms such as "inside" and "outside" indicating directions or positional relationships are based on the directions or positional relationships shown in the drawings. This is only for the convenience of description, and does not indicate or imply that the device or component must have a specific orientation, be constructed and operated in a specific orientation. Therefore, it cannot be understood as a limitation on the present application.

[0123] In the description of the present application, the description with reference to the terms "one embodiment", "some embodiments", "in the present embodiment", "specific example", or "some examples" etc. means that the specific features, mechanisms, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, mechanisms, materials or characteristics described may be combined in a suitable manner in any one or more embodiments or examples. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.

[0124] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed in the present application should be included in the protection scope of the present application. Therefore, the protection scope of the present application shall be based on the protection scope of the claims.

Claims

1. A method for analyzing factors affecting economic losses caused by accidents, characterized in that: The steps include: S1: Constructing the Bayesian random parameter GB2 model based on the observation information of sample characteristics of captured fishing vessel accidents; S2: constructing multiple random parameter models for analyzing the factors affecting the economic losses of fishing boat accidents based on the Bayesian random parameter GB2 model and the observation information of sample characteristics; S3: Calculating the leave-one-out cross-validation information criterion value of each of the random parameter models by a calculation module; S4: Select the random parameter model with a low value of the leave-one-out cross-validation information criterion to analyze the factors affecting the economic losses of fishing vessel accidents, and obtain the positive or negative impact of various explanatory variables on the economic losses of fishing vessel accidents; S5: Based on the positive or negative impact of various explanatory variables on the economic losses caused by fishing boat accidents obtained in step S4, the degree of influence of the explanatory variables on the economic losses caused by accidents is explored.

2. The method for analyzing factors affecting accident economic losses according to claim 1, characterized in that: In the Bayesian random parameter GB2 model: The probability density function of a random variable is as follows: In the formula, a is the shape parameter, which determines the peak and tail thickness of the economic loss distribution; p is the probability of an accident with smaller economic losses; q is the probability of a major economic loss accident; B(p,q) is the B function; λ reflects the level of economic loss of the accident and is a scale parameter that controls the overall level and scope of the loss. Its calculation formula is as follows: λ=exp(β′x)=exp(β0+β1x1+β2x2+...+β k x k ), In the formula, x is the vector of covariates that affect the economic losses of fishing boat accidents; β is the coefficient vector corresponding to x; k is the number of covariates; The kth moment of a random variable is given by:

3. The method for analyzing factors affecting accident economic losses according to claim 2 is characterized in that: The multiple random parameter models in step S2 include a general random parameter model, a mean heterogeneity random parameter model and a mean and variance heterogeneity random parameter model.

4. The method for analyzing factors affecting accident economic losses according to claim 3 is characterized in that: The general random parameter model is expressed as follows: In the formula, β n is a random parameter vector; b is the average parameter vector of all observations of the sample characteristics of fishing boat accidents; is a randomly distributed error term used to capture possible heterogeneity.

5. The method for analyzing factors affecting accident economic losses according to claim 4 is characterized in that: The mean heterogeneity random parameter model is expressed as follows: In the formula, Z n To influence β n vector of explanatory variables for the mean; Θ is Z n The corresponding estimable parameter vector; if the elements in Θ are significantly different from zero, it means that the mean heterogeneity random parameter model has mean heterogeneity.

6. The method for analyzing factors affecting accident economic losses according to claim 5 is characterized in that: The random parameter model with heterogeneity in mean and variance is expressed as follows: In the formula, W n is a vector of explanatory variables used to capture the heterogeneity of the standard deviation σ; ω is W n The corresponding estimable parameter vector; If no element in ω is significantly different from zero, then the random parameter model with heterogeneity in mean and variance only has heterogeneity in the mean; If no element in Θ is significantly different from zero, it means that the mean and variance heterogeneity random parameter model degenerates to the general random parameter model.

7. The method for analyzing factors affecting accident economic losses according to claim 6 is characterized in that: In step S3, the calculation module obtains the calculation formula of the leave-one-out cross-validation information criterion value as follows: LOOIC=-2ELPD LOO , In the formula, p M (y i |θ) represents the M likelihood in the regression model; p M (θ|y -i ) represents the posterior value of the parameter θ, where the observed value y i are excluded.

8. A system for analyzing factors affecting economic losses caused by accidents, characterized in that: The method for analyzing factors affecting economic losses caused by accidents according to any one of claims 1 to 7 comprises establishing a circuit connection relationship: The model module is configured to construct a Bayesian random parameter GB2 model based on the observed information of sample characteristics of captured fishing vessel accidents; A parameter model module is configured to construct a plurality of random parameter models for analyzing factors affecting economic losses of fishing boat accidents based on the Bayesian random parameter GB2 model and observation information of sample characteristics; A calculation module is configured to calculate the LOOIC value of each of the random parameter models respectively; The selection module is set to select a random parameter model with a low value of the leave-one-out cross-validation information criterion to analyze the factors affecting the economic losses of fishing vessel accidents, and obtain the positive or negative effects of various explanatory variables on the economic losses of fishing vessel accidents; The analysis module is set to the positive or negative impact of various explanatory variables on the economic losses of fishing boat accidents, and the influence of the explanatory variables on the economic losses of accidents is excavated.

Images