County traffic safety evaluation method combining RSR method and ordered sample clustering method

Abstract

According to the county traffic safety evaluation method combining the RSR method and the ordered sample clustering method provided by the invention, the whole evaluation process does not involve artificial subjective scoring, objective data indexes are selected as all traffic safety evaluation indexes, objective evaluation is carried out on to-be-evaluated county rating based on the ordered sample clustering method, and the evaluation efficiency is improved. It is ensured that the dependence on the personal ability of the operator is reduced. And classifying the traffic safety comprehensive evaluation values WRSR of the county region to be evaluated based on an ordered sample clustering method, and obtaining a rating result based on a classification result.

Description

Technical Field

[0001] The invention relates to the technical field of traffic safety, and in particular to a county traffic safety evaluation method combining an RSR method and an ordered sample clustering method. Background Art

[0002] In recent years, with the rapid development of county economy, transportation elements such as people, vehicles and roads in counties have also shown a rapid growth trend. However, problems such as weak county transportation infrastructure, low safety awareness of traffic participants, and large number and wide range of potential accidents have gradually emerged. In the traffic management system, managers must first scientifically evaluate the traffic safety risks in counties under their jurisdiction, find areas where risks are concentrated, and then effectively make improvements. In many cases, managers rely on their personal perception of county traffic safety risks in their jurisdiction to evaluate.

[0003] In the prior art, there are also technical methods for evaluating regional traffic safety status. For example, application number CN106530706A provides a regional traffic safety evaluation method based on connectivity spatial weight matrix. Collect relevant data for regional traffic safety analysis, establish a Bayesian conditional autoregressive model with traffic analysis cells as basic units as a regional accident risk prediction model, identify the network connectivity by the number of paths contained between regions, construct a spatial weight matrix, identify the network connectivity by the number of paths contained between regions, and construct a spatial weight matrix. This patented method is suitable for urban built-up areas where traffic cells are easier to divide, but not for counties.

[0004] The assessment of county traffic covers a wide area, and there is no ready-made traffic community planning. Moreover, county traffic involves many indicators such as traffic accidents, traffic hazards, and traffic management. When selecting evaluation indicators, evaluation indicators in different fields may produce completely opposite evaluation results for the same evaluation object, so the actual significance of the evaluation results formed by the combination of different indicators is greatly weakened. These are the difficulties in county traffic safety evaluation. Summary of the invention

[0005] In order to solve the problem that the existing traffic safety evaluation method for urban traffic areas is not suitable for county traffic safety evaluation with more complicated situations, the present invention provides a county traffic safety evaluation method combining RSR method and ordered sample clustering method, which can reduce the dependence on the operator's personal ability and evaluate the county traffic safety level more scientifically.

[0006] The technical solution of the present invention is as follows: a county traffic safety evaluation method combining RSR method and ordered sample clustering method, characterized in that it includes the following steps:

[0007] S1: Construct traffic safety evaluation indicators for the county to be evaluated;

[0008] Assume: there are n counties to be evaluated, and each county to be evaluated corresponds to m traffic safety evaluation indicators;

[0009] The traffic safety evaluation index is an objective data index;

[0010] S2: Collect relevant data of the traffic safety evaluation index and establish an original data matrix X with n rows and m columns;

[0011]

[0012] Among them, x nm represents the value of the mth traffic safety evaluation index of the nth county to be evaluated;

[0013] S3: Use the RSR method to rank the original data matrix X and establish the rank matrix R;

[0014]

[0015] Among them, R ij Represents x ij In x 1j , x 2j ,···,x nj rank in;

[0016] S3: Expressing the information amount of the traffic safety evaluation index by the coefficient of variation V of the index to obtain the index information amount V;

[0017] The compatibility of the traffic safety evaluation index is expressed as the sum of the Spearman correlation coefficients C between the index and other indexes, and the index compatibility C is obtained;

[0018] The amount of information V corresponding to the jth indicator j for:

[0019]

[0020] in,

[0021] The index compatibility C corresponding to the jth index j for:

[0022]

[0023] Among them, m is the total number of traffic safety evaluation indicators, c ij is the Spearman correlation coefficient between the i-th traffic safety evaluation index and the j-th traffic safety evaluation index;

[0024] S4: Calculate the comprehensive weights W of the m traffic safety evaluation indicators respectively according to the indicator information volume V and the indicator compatibility C;

[0025] The comprehensive weight W of the jth traffic safety evaluation index j The calculation method is:

[0026]

[0027] S5: Combine the rank matrix R and the comprehensive weight W to calculate the comprehensive traffic safety evaluation value WRSR of n counties;

[0028] The comprehensive traffic safety evaluation value WRSR corresponding to the i-th county to be evaluated i The calculation method is:

[0029]

[0030] Among them, m is the total number of traffic safety evaluation indicators;

[0031] S6: construct an ordered sequence based on the WRSR values ​​of the counties to be evaluated, and classify the ordered sequence using an ordered sample clustering method to obtain the optimal classification results of the traffic safety levels of the n counties to be evaluated.

[0032] It is further characterized by:

[0033] In step S6, the following steps are included:

[0034] a1: construct an ordered sample sequence;

[0035] Arrange the WRSR values ​​of n counties from high to low to form a sequence y of n ordered samples: y (1) ,y (2) , …, y (n) ;

[0036] a2: Calculate the classification diameter of all existing classification methods;

[0037]

[0038] Where D(i, j) represents the classification {y (i) ,y (i+1) ,…y (j)} diameter, Represents the mean of all samples in the classification;

[0039] a3: The optimal classification method for solving WRSR values;

[0040] Construct a loss function L[b(n, k)], where L[b(n, k)] represents the loss function for classifying n samples into k categories;

[0041]

[0042] Among them, \(i\) t represents the first element of the \(t\)-th category, \(i_1 = 1\), \(i\) k+1 - 1 = \(n\);

[0043] Solve for the said loss function to obtain the minimum loss function \(L[P(n,k)]\) and the optimal classification method \(P(n,k)\);

[0044] a4: Determine the optimal number of classifications:

[0045] For any number of classifications \(k\) satisfying \(2\leq k\leq n\), plot the curve of \(L[P(n,k)]\) changing with \(k\), and take the inflection point of the curve as the optimal classification \(k\) b ;

[0046]

[0047] Obtain the optimal grading results of the traffic safety levels of \(n\) counties to be evaluated;

[0048] Step a3 includes the following detailed steps:

[0049] b1: Calculate the minimum loss function \(L[P(n,k)]\);

[0050] When \(k = 2\), its calculation method is as follows:

[0051]

[0052] When \(2\lt k\leq n\), its calculation method is as follows:

[0053]

[0054] Among them, \(P(n,k)\) represents the classification method that makes the loss function reach the minimum;

[0055] b2: Find the dividing point \(j\) k , so that the recurrence formula reaches the minimum;

[0056] \(L[P(n,k)] = L[P(j\) k - 1,k - 1)]+D(j\) k ,n);

[0057] Obtain the \(k\)-th category \(G\) k =\(\{j\) k ,j\) k + 1,\(\cdots\),n\};

[0058] b3: Find \(j\) k-1 , so that it satisfies:

[0059] \(L[P(j\) k-1 ,k - 1)] = L[P(j\) k-1 - 1,k - 2)]+D(j\)k-1 , j k -1);

[0060] Get the k-1th class G k-1 ={j k-1 , j k-1 +1,…,j k -1}, we can get all classes G1, G2, ..., G k , which is the optimal solution with the minimum loss function:

[0061] P(n, k) = {G1, G2, ..., G k};

[0062] The traffic safety evaluation indicators include: P1 road traffic accident death rate per 100 kilometers, P2 traffic accident death rate per 100,000 people, P3 traffic accident death rate per million vehicles, P4 proportion of traffic accidents caused by major violations, P5 proportion of deaths and injuries among key groups, P6 accident rate per 100 vehicles of road transport enterprises, P7 number of accident-prone sections, P8 rate of unprocessed violations of more than 3 motor vehicles, P9 violation rate per 100 vehicles of road transport enterprises, P 10 Helmet use rate of motorcycle and electric bike riders, P 11 Traffic management personnel staffing rate per 100 kilometers of road, P 12 Front-end sensing equipment installation rate and P per 100 km highway 13 Intensification of investigation and punishment of key illegal activities;

[0063] When calculating the ranking, the ranking of benefit-type indicators is obtained by ranking the indicator data from low to high, and the ranking of cost-type indicators is obtained by ranking the indicator data from high to low. The same indicator data is ranked average;

[0064] The Spearman correlation coefficient c ij The calculation method is:

[0065]

[0066] The present application provides a county traffic safety evaluation method that combines the RSR method and the ordered sample clustering method. The entire evaluation process does not involve manual subjective scoring. All traffic safety evaluation indicators select objective data indicators. The rating of the county to be evaluated is objectively evaluated based on the ordered sample clustering method to ensure that the reliance on the operator's personal ability is reduced. The comprehensive traffic safety evaluation value WRSR of the county to be evaluated is classified based on the ordered sample clustering method, and the rating result is obtained based on the classification result. It can minimize the WRSR difference between all counties to be evaluated in the same level (classification), that is, counties with the same traffic safety level are classified into the same category; at the same time, the WRSR difference between counties to be evaluated at different levels (classifications) is large, ensuring that the classification result is scientific and practical in accordance with the actual situation. At the same time, after solving L[P(n, k)], a curve of L[P(n, k)] changing with the number of classifications k is drawn, and the inflection point of the curve is used as the optimal classification k. b , ensuring that the classification results can be separated from the dependence on the operator's personal ability and experience, and effectively distinguishing the differences in traffic safety levels between different counties based on actual data effects. This method measures the information content V of each indicator based on the coefficient of variation of the traffic safety evaluation index among all indicators; when calculating the comprehensive weight W, a lower weight is given to indicators with low information content. When the evaluation results of the indicators for each evaluation object are too convergent, the impact of the indicators on the final evaluation results is weakened, making the results of this method more practical. This method constructs the indicator compatibility C. By introducing the Spearman correlation coefficient into the indicator compatibility C, the compatibility between an evaluation indicator and other evaluation indicators in forming the evaluation results is measured. When calculating the comprehensive weight W, a lower weight is given to indicators whose evaluation results are incompatible with other indicators. Even if there are special indicators incompatible with other indicators among the m traffic safety evaluation indicators, their impact on the overall evaluation results can be reduced, which improves the overall stability of the evaluation method. When calculating the comprehensive traffic safety evaluation value WRSR corresponding to the county to be evaluated, the comprehensive weight W is constructed through the indicator information volume V and the indicator compatibility C to ensure the availability and objectivity of the comprehensive traffic safety evaluation value WRSR, thereby ensuring the accuracy of the final rating result. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] Figure 1 It is a flowchart for the comprehensive evaluation of county traffic safety;

[0068] Figure 2 It is a flow chart of the steps of the RSR method;

[0069] Figure 3 It is a flowchart of the steps of the ordered sample clustering method;

[0070] Figure 4 This is an example of the curve of the minimum classification loss function changing with the number of classifications. DETAILED DESCRIPTION

[0071] like Figure 1 As shown, the present application includes a county traffic safety evaluation method combining an RSR method and an ordered sample clustering method, characterized in that it includes the following steps:

[0072] S1: Construct traffic safety evaluation indicators for the counties to be evaluated;

[0073] Assume: there are n counties to be evaluated, and each county to be evaluated corresponds to m traffic safety evaluation indicators;

[0074] Traffic safety evaluation indicators are objective data indicators. To ensure the objectivity of the calculation results, the traffic safety evaluation indicators set in this application do not include any subjective scoring items. Each indicator must be statistically quantifiable, and the basic data can be accurately obtained within a certain time frame. Specific traffic safety evaluation indicators can be adaptively set according to the actual needs of each region.

[0075] In this embodiment, the traffic safety evaluation indicators include: P1 road traffic accident death rate per 100 kilometers, P2 traffic accident death rate per 100,000 people, P3 traffic accident death rate per million vehicles, P4 proportion of traffic accidents caused by major violations, P5 proportion of deaths and injuries among key groups, P6 accident rate per 100 vehicles of road transport enterprises, P7 number of accident-prone sections, P8 rate of unprocessed violations of more than 3 motor vehicles, P9 road transport enterprise violation rate per 100 vehicles, P 10 Helmet use rate of motorcycle and electric bike riders, P 11 Traffic management personnel staffing rate per 100 kilometers of road, P 12 Front-end sensing equipment installation rate and P per 100 km highway 13 Strengthen the investigation and punishment of key illegal activities.

[0076] S2: Collect relevant data of traffic safety evaluation indicators from the historical data of the designated evaluation period in the county to be evaluated, and establish an original data matrix X with n rows and m columns;

[0077]

[0078] Among them, x nm represents the value of the mth traffic safety evaluation index of the nth county to be evaluated;

[0079] S3: Use the RSR method (Rank-sum ratio) to rank the original data matrix X and establish the rank matrix R;

[0080]

[0081] Among them, R ij Represents x ij In x1j , x 2j ,···,x nj The rank in .

[0082] When calculating the ranking, the ranking of benefit-type indicators is obtained by ranking the indicator data from low to high, and the ranking of cost-type indicators is obtained by ranking the indicator data from high to low. The same indicator data is ranked average. The average ranking calculation method is based on the existing technology.

[0083] S3: The information amount of the traffic safety evaluation index is expressed by the coefficient of variation V of the index to obtain the index information amount V;

[0084] The compatibility of traffic safety evaluation indicators is expressed as the sum of the Spearman correlation coefficients C between the indicator and other indicators, and the indicator compatibility C is obtained;

[0085] The amount of information V corresponding to the jth indicator j for:

[0086]

[0087] in,

[0088] The index compatibility C corresponding to the jth index j for:

[0089]

[0090] Among them, m is the total number of traffic safety evaluation indicators, c ij is the Spearman correlation coefficient between the i-th traffic safety evaluation index and the j-th traffic safety evaluation index; in this method, in order to eliminate the possible negative correlation effect of the indicators, the number of indicators m is added as positive compensation.

[0091] Spearman correlation coefficient c ij The calculation method is:

[0092]

[0093] S4: Calculate the comprehensive weights W of m traffic safety evaluation indicators according to the indicator information volume V and the indicator compatibility C;

[0094] The comprehensive weight W of the jth traffic safety evaluation index j The calculation method is:

[0095]

[0096] The evaluation of county traffic safety level covers a wide range of fields, including traffic accidents, traffic hazards, traffic management and other aspects. When selecting evaluation indicators, evaluation indicators in different fields may produce completely opposite evaluation results for the same evaluation object, so the actual significance of the evaluation results formed by different indicators is greatly weakened. To solve this problem, this method designs the concept of indicator compatibility based on the Spearman correlation coefficient. As a component of the indicator weight, it gives a lower weight to the indicators whose evaluation results are incompatible with other indicators, ensuring the consistency of the indicators for the evaluation of county traffic safety level, effectively reducing the impact of selecting special indicators on the overall evaluation results, and improving the overall stability of the evaluation method.

[0097] S5: Combine the rank matrix R and the comprehensive weight W to calculate the comprehensive traffic safety evaluation value WRSR of n counties;

[0098] The comprehensive traffic safety evaluation value WRSR corresponding to the i-th county to be evaluated i The calculation method is:

[0099]

[0100] Among them, m is the total number of traffic safety evaluation indicators; through 1 / n, the final value falls in the interval [0, 1].

[0101] S6: Construct an ordered sequence based on the WRSR values ​​of the counties to be evaluated, and use the ordered sample clustering method to classify the ordered sequence to obtain the optimal classification results of the traffic safety levels of the n counties to be evaluated.

[0102] like Figure 3 As shown, step S6 includes the following steps.

[0103] a1: construct an ordered sample sequence;

[0104] Arrange the WRSR values ​​of n counties from high to low to form a sequence y of n ordered samples: y (1) ,y (2) , …, y (n) .

[0105] a2: Calculate the classification diameter D(i, j) of all existing classification methods;

[0106]

[0107] Where D(i, j) represents the classification {y (i) ,y (i+1) ,…y (j)} diameter, It represents the mean of all samples (WRSR values) in the classification.

[0108] a3: The optimal classification method for solving the WRSR value;

[0109] Construct a loss function L[b(n, k)], where L[b(n, k)] represents the loss function for classifying n samples into k categories;

[0110] For the solution of the loss function, the classification method that minimizes the loss function is the optimal classification method

[0111]

[0112] Among them, i t represents the first element of the t-th category, i1 = 1, i k+1 -1 = n;

[0113] For the solution of the said loss function, the minimum loss function is obtained as L[P(n, k)] and the optimal classification method P(n, k).

[0114] a4: Determine the optimal number of classifications:

[0115] For any number of classifications k satisfying 2 ≤ k ≤ n, after solving L[P(n, k)] respectively, draw a curve of L[P(n, k)] changing with k, and take the inflection point of the curve as the optimal classification k b ;

[0116]

[0117] Obtain the optimal classification result of the traffic safety level of n counties to be evaluated.

[0118] Step a3 includes the following detailed steps.

[0119] b1: Calculate the minimum loss function L[P(n, k)];

[0120] When k = 2, its calculation method is as follows:

[0121]

[0122] When 2 < k ≤ n, its calculation method is as follows:

[0123]

[0124] Among them, P(n, k) represents the classification method that minimizes the loss function.

[0125] The following is to solve the optimal classification method P(n, k).

[0126] b2: Find the dividing point j k , so that the recurrence formula reaches the minimum;

[0127] L[P(n, k)] = L[P(jk -1, k-1)]+D(j k , n);

[0128] Get the kth class G k ={j k , j k +1,…,n}.

[0129] b3: Find j k-1 , so that it satisfies:

[0130] L[P(j k-1 , k-1)] = L[P(j k-1 -1,k-2)]+D(j k-1 , j k -1).

[0131] Get the k-1th class G k-1 ={j k-1 , j k-1 +1,…,j k -1}, we can get all classes G1, G2, ..., G k , which is the optimal solution with the minimum loss function:

[0132] P(n, k) = {G1, G2, ..., G k};

[0133] In the process of grading the evaluation object by the traditional RSR method, it is necessary to determine the WRSR distribution and query the comparison table of percentage and probability unit, then calculate the corresponding linear regression equation, and finally sort and grade. The process is relatively cumbersome. And when applied to the county traffic safety evaluation, the grading of the county traffic safety level mainly depends on the operator's experience and lacks scientific basis. To solve this problem, the ordered sample clustering method is used in this method to grade the county traffic safety level, which can minimize the difference between the county traffic safety level evaluation values ​​WRSR within each level, and the difference between each level is large. And by comparing the difference in the minimum loss function under the condition of each classification number, the preferred number of classifications can be determined. It effectively solves the defects of the RSR method in grading and can effectively distinguish the differences in traffic safety levels between different counties.

[0134] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a description of the specific implementation method, structure, characteristics and effects of the county traffic safety evaluation method proposed by the present method in combination with specific embodiments. In this embodiment, the traffic safety levels of 20 counties in a province are evaluated and graded.

[0135] Construct traffic safety evaluation indicators for the counties to be evaluated.

[0136] By screening the factors affecting road traffic safety in the county, traffic safety risk assessment indicators are selected from the aspects of traffic accidents, traffic hazards, traffic management, etc. Preferably, the evaluation indicators include: P1 road traffic accident death rate per 100 kilometers, P2 traffic accident death rate per 100,000 people, P3 traffic accident death rate per million vehicles, P4 proportion of traffic accidents caused by major violations, P5 proportion of deaths and injuries among key groups, P6 accident rate per 100 vehicles of road transport enterprises, P7 number of accident-prone sections, P8 rate of unprocessed violations of more than 3 motor vehicles, P9 road transport enterprise violation rate per 100 vehicles, P10 10 Helmet use rate of motorcycle and electric bike riders, P 11 Traffic management personnel staffing rate per 100 kilometers of road, P 12 Front-end sensing equipment installation rate per 100 kilometers of highway, P 13 Strengthen the investigation and punishment of key illegal activities.

[0137] Collect relevant data of traffic safety evaluation indicators and establish an original data matrix X with n rows and m columns.

[0138] In this embodiment, n=20, m=13. After determining the evaluation index, collect relevant data from the provincial statistics department, transportation department, and public security traffic management department to calculate the value of each index in each county. After the basic data is standardized, establish the index data matrix X as shown in Table 1.

[0139] Table 1: Traffic safety evaluation index data matrix X for each county in a province

[0140] <![CDATA[P1]]> <![CDATA[P2]]> <![CDATA[P3]]> <![CDATA[P4]]> <![CDATA[P5]]> <![CDATA[P6]]> <![CDATA[P7]]> <![CDATA[P8]]> <![CDATA[P9]]> <![CDATA[P 10 ]]> <![CDATA[P 11 ]]> <![CDATA[P 12 ]]> <![CDATA[P 13 ]]> County A 0.21 0.64 0.25 0.31 0.37 0.43 0.00 0.20 0.55 0.40 0.17 0.23 0.18 County B 0.61 0.53 0.19 0.33 0.53 0.50 0.25 0.21 0.77 0.32 0.44 0.60 0.11 County C 0.41 0.93 0.76 0.26 0.50 0.47 0.00 0.21 0.59 0.99 0.10 0.29 0.05 County D 0.39 0.78 0.49 0.35 0.54 0.51 0.00 0.18 0.77 0.26 0.22 0.37 0.13 County E 0.25 0.33 0.06 0.45 0.68 0.40 0.00 0.32 0.50 0.32 0.26 0.53 0.08 County F 0.64 0.68 0.61 0.40 0.67 0.00 0.00 0.05 0.46 0.49 0.25 0.33 0.01 County G 0.22 0.39 0.32 0.29 0.38 0.30 0.50 0.28 0.42 0.19 0.26 0.31 0.01 H County 0.59 0.65 0.59 0.19 0.45 0.05 0.50 0.19 0.25 0.83 0.12 0.24 0.01 County I 0.20 0.64 0.52 0.18 0.54 0.63 0.25 0.10 1.00 0.43 0.17 0.18 0.12 J County 0.38 0.45 0.20 0.17 0.62 0.29 0.00 0.45 0.44 0.36 0.24 0.37 0.12 K County 0.43 0.35 0.25 0.09 0.37 0.56 0.00 0.36 0.55 0.62 0.06 0.21 0.04 L County 0.41 0.31 0.58 0.22 0.51 0.37 0.00 0.17 0.64 0.17 0.17 0.14 0.14 M County 0.61 0.45 0.21 0.45 0.32 0.45 0.50 0.25 0.79 0.87 0.32 0.52 0.01 County N 0.26 0.48 0.31 0.48 0.25 0.00 0.50 0.26 0.09 0.32 0.03 0.09 0.00 O County 0.27 0.74 0.86 0.32 0.61 0.00 0.75 0.61 0.33 0.31 0.02 0.03 0.01 P County 0.32 0.64 0.79 0.40 0.34 0.58 0.50 0.19 0.53 0.12 0.12 0.04 0.01 Q County 0.62 0.72 0.38 0.43 0.41 0.24 0.75 0.74 0.36 0.48 0.09 0.14 0.02 R County 0.44 0.54 0.40 0.83 0.43 0.24 0.75 0.42 0.31 0.16 0.15 0.12 0.01 S County 0.57 0.79 0.80 0.51 0.50 0.31 0.50 0.40 0.50 0.20 0.11 0.05 0.00 T County 0.59 1.00 0.78 0.44 0.49 0.36 0.75 0.00 0.00 0.23 0.15 0.09 0.05

[0141] The comprehensive evaluation value of county traffic safety is calculated based on the RSR method. The steps of the RSR method are as follows: Figure 2 As shown, it is described in detail below in conjunction with the embodiments.

[0142] (1) Ranking. P1 to P9 are cost-type indicators, which are ranked from small to large; P 10 ~P 13 It is a benefit-type indicator, and it is ranked from high to low. The resulting rank matrix R is shown in Table 2.

[0143] Table 2: Rank matrix R of traffic safety evaluation index data for each county in a province

[0144] <![CDATA[P1]]> <![CDATA[P2]]> <![CDATA[P3]]> <![CDATA[P4]]> <![CDATA[P5]]> <![CDATA[P6]]> <![CDATA[P7]]> <![CDATA[P8]]> <![CDATA[P9]]> <![CDATA[P 10 ]]> <![CDATA[P 11 ]]> <![CDATA[P 12 ]]> <![CDATA[P 13 ]]> County A 19 11 15 13 17 8 16.5 13 7 13 11 11 20 County B 3 13 19 11 7 5 11.5 12 4 11 20 20 15 County C 10 2 5 15 9 6 16.5 11 6 20 5 13 13 County D 12 4 10 10 5 4 16.5 16 3 7 14 16 18 County E 17 19 20 5 1 9 16.5 7 10 9 17 19 14 County F 1 7 6 9 2 19 16.5 19 12 16 16 15 5 County G 18 17 13 14 15 13 7.5 8 14 4 18 14 6 H County 5 8 7 17 12 17 7.5 15 18 18 8 12 8 County I 20 9 9 18 6 1 11.5 18 1 14 13 9 16 J County 13 15 18 19 3 14 16.5 3 13 12 15 17 17 K County 9 18 16 20 16 3 16.5 6 8 17 3 10 11 L County 11 20 8 16 8 10 16.5 17 5 3 12 7 19 M County 4 16 17 4 19 7 7.5 10 2 19 19 18 9 County N 16 14 14 3 20 19 7.5 9 19 10 2 5 1 O County 15 5 1 12 4 19 2.5 2 16 8 1 1 4 P County 14 10 3 8 18 2 7.5 14 9 1 7 2 7 Q County 2 6 12 7 14 15 2.5 1 15 15 4 8 10 R County 8 12 11 1 13 16 2.5 4 17 2 9 6 3 S County 7 3 2 2 10 12 7.5 5 11 5 6 3 2 T County 6 1 4 6 11 11 2.5 20 20 6 10 4 12

[0145] (2) Calculation of weight W. The information content V and compatibility C of the 13 indicators are calculated respectively, and then the comprehensive weight W of the 13 indicators is calculated as shown in Table 3.

[0146] Table 3: Weights of the 13 indicators

[0147] <![CDATA[P1]]> <![CDATA[P2]]> <![CDATA[P3]]> <![CDATA[P4]]> <![CDATA[P5]]> <![CDATA[P6]]> <![CDATA[P7]]> <![CDATA[P8]]> <![CDATA[P9]]> <![CDATA[P 10 ]]> <![CDATA[P 11 ]]> <![CDATA[P 12 ]]> <![CDATA[P 13 ]]> Information volume V 0.27 0.49 0.46 0.25 0.23 0.30 0.45 0.25 0.47 0.61 0.60 0.69 1.04 Compatibility C 13.24 15.02 15.52 14.56 11.75 11.22 15.65 13.12 10.56 13.94 14.75 15.55 14.85 Comprehensive weight W 0.04 0.08 0.08 0.04 0.03 0.04 0.08 0.04 0.06 0.10 0.10 0.12 0.18

[0148] (3) Calculation of county-level traffic safety comprehensive evaluation value WRSR. The WRSR values ​​of the 20 counties are calculated by combining the rank matrix R and the comprehensive weight W:

[0149]

[0150] The traffic safety level of counties is rated based on the ordered sample clustering method. The steps of the ordered sample clustering method are as follows: Figure 3 As shown, it is described in detail below in conjunction with the embodiments.

[0151] (1) Construct an ordered sample sequence. Arrange the WRSR values ​​of the 20 counties from high to low, and construct an ordered sample sequence y as follows:

[0152]

[0153] (2) Calculate the classification diameter.

[0154] For any 1≤i<j≤20, calculate the classification diameter D(i,j) respectively, taking D(1,3) as an example:

[0155] D(1,3)=(0.74-0.64) 2 +(0.71-0.64) 2 +(0.70-0.64) 2 =0.0086.

[0156] (3) Find the optimal classification method.

[0157] For all the classification numbers k that satisfy the condition 2≤k<20, solve the minimum classification loss function L[P(20, k)] and the optimal classification method P(20, k) = {G1, G2, .., G k}.

[0158] Take k=3 as an example:

[0159] L[P(20,3)]=D(1,4)+D(5,13)+D(14,20)=0.044,

[0160] G1={y1, y2, y3, y4}, G2={y5, y6, y7, y8, y9, y 10 ,y 11 ,y 12 ,y 13}, G3={y 14 ,y 15 ,y 16 ,y 17 ,y 18 ,y 19 ,y20}.

[0161] (4) Determine the optimal number of categories

[0162] Draw the minimum classification loss function as the number of classifications changes. Figure 4 As shown, the inflection point k=4 is selected as the optimal number of classifications, that is, the 20 counties are divided into 4 levels. The classification results are shown in Table 4.

[0163] Table 4: Traffic safety rating examples

[0164]

[0165] The county traffic safety evaluation method provided by the present application first selects evaluation indicators that can be quantified based on objective data according to actual needs; collects relevant data, calculates the indicator values ​​of each county, and establishes an original data matrix; uses the RSR method to rank the original data matrix, establishes a rank matrix, calculates the comprehensive weights of each indicator according to the information content of the indicator and the compatibility of the indicator, and calculates the comprehensive traffic safety evaluation value WRSR of each county by combining the rank matrix and the weight; constructs an ordered sequence with the WRSR value of each county, and uses the ordered sample clustering method to classify it to obtain the classification result of the county traffic safety level. The county traffic safety evaluation method provided by the present invention can evaluate the traffic safety level of counties within a certain range and divide the traffic safety level into grades, which can help traffic management departments identify the key points of county traffic safety risk prevention and improve the refined level of county management within the jurisdiction in a targeted manner.

[0166] The above are only preferred embodiments of the present invention, and are not intended to limit the present invention in any form. Although the present invention has been disclosed as a preferred embodiment as above, it is not intended to limit the present invention. Any technician familiar with the profession can make some changes or modifications to equivalent embodiments of equivalent changes using the technical contents disclosed above without departing from the scope of the technical solution of the present invention. However, any simple modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention are still within the scope of the technical solution of the present invention.

Claims

1. A county traffic safety evaluation method combining RSR method and ordered sample clustering method, characterized in that: It includes the following steps: S1: Construct traffic safety evaluation indicators for the county to be evaluated; Suppose: There are n counties to be evaluated, and each county to be evaluated corresponds to m traffic safety evaluation indicators; The traffic safety evaluation indicators are objective data indicators; S2: Collect relevant data of the traffic safety evaluation indicators and establish an original data matrix X with n rows and m columns; Among them, x nm represents the value of the mth traffic safety evaluation index of the nth county to be evaluated; S3: Use the RSR method to rank the original data matrix X and establish a rank matrix R; Among them, R ij Represents x ij In x 1j , x 2j ,···,x nj rank in; S3: Express the information amount of the traffic safety evaluation indicator by the coefficient of variation V of this indicator to obtain the indicator information amount V; Express the compatibility of the traffic safety evaluation indicator by the sum C of the Spearman correlation coefficients between this indicator and other indicators to obtain the indicator compatibility C; The amount of information V corresponding to the jth indicator j for: in, The index compatibility C corresponding to the jth index j for: Among them, m is the total number of traffic safety evaluation indicators, c ij is the Spearman correlation coefficient between the i-th traffic safety evaluation index and the j-th traffic safety evaluation index; S4: Calculate the comprehensive weight W of m traffic safety evaluation indicators according to the indicator information amount V and the indicator compatibility C respectively; The comprehensive weight W of the jth traffic safety evaluation index j The calculation method is: S5: Combine the rank matrix R and the comprehensive weight W to calculate the comprehensive traffic safety evaluation value WRSR of n counties; The comprehensive traffic safety evaluation value WRSR corresponding to the i-th county to be evaluated i The calculation method is: Among them, m is the total number of traffic safety evaluation indicators; S6: Construct an ordered sequence with the WRSR values of the counties to be evaluated, and use the method of clustering ordered samples to classify the ordered sequence to obtain the optimal classification result of the traffic safety levels of n counties to be evaluated.

2. The county traffic safety evaluation method combining RSR method and ordered sample clustering method according to claim 1 is characterized by: In step S6, it specifically includes the following steps: a1: Construct an ordered sample sequence; Arrange the WRSR values of n counties from high to low to form a sequence y of n ordered samples: y(1), y(2), …, y(n); a2: Calculate the classification diameters of all existing classification methods; Among them, D(i,j) represents the classification {y (i) ,y (i+1) ,…y (j) } diameter, Represents the mean of all samples in the classification; a3: Solve the optimal classification method of the WRSR value; Construct a loss function L[b(n,k)], where L[b(n,k)] represents the loss function for dividing n samples into k categories; Among them, i t Represents the first element of the tth category, i1=1, i k+1 -1 = n; Solve for the loss function to obtain the minimum loss function L[P(n,k)] and the optimal classification method P(n,k); a4: Determine the optimal number of classifications: For any number of categories k that satisfies 2≤k≤n, plot a curve of L[P(n,k)] varying with k, and take the inflection point of the curve as the optimal category k. b ; Obtain the optimal classification result of the traffic safety levels of n counties to be evaluated.

3. The county traffic safety evaluation method combining RSR method and ordered sample clustering method according to claim 2 is characterized by: Step a3 includes the following detailed steps: b1: Calculate the minimum loss function L[P(n,k)]; When k = 2, its calculation method is as follows: When 2 < k ≤ n, its calculation method is as follows: Among them, P(n,k) represents the classification method that makes the loss function reach the minimum; b2: find the dividing point j k , so that the recursive formula reaches a minimum; L[P(n,k)]=L[P(j k -1,k-1)]+D(j k ,n); Get the kth class G k ={j k ,j k +1,…,n}; b3: Find j k-1 , so that it satisfies: L[P(j k-1 ,k-1)]=L[P(j k-1 -1,k-2)]+D(j k-1 ,j k -1); Get the k-1th class G k-1 ={j k-1 ,j k-1 +1,…,j k -1}, we can get all classes G1, G2,…, G k , which is the optimal solution with the minimum loss function: P(n,k)={G1,G2,…,G k }。 4. The county traffic safety evaluation method combining RSR method and ordered sample clustering method according to claim 1 is characterized by: The traffic safety evaluation indicators include: P1 road traffic accident death rate per 100 kilometers, P2 traffic accident death rate per 100,000 people, P3 traffic accident death rate per million vehicles, P4 proportion of traffic accidents caused by major violations, P5 proportion of deaths and injuries among key groups, P6 accident rate per 100 vehicles of road transport enterprises, P7 number of accident-prone sections, P8 rate of unprocessed violations of more than 3 motor vehicles, P9 violation rate per 100 vehicles of road transport enterprises, P 10 Helmet use rate of motorcycle and electric bike riders, P 11 Traffic management personnel staffing rate per 100 kilometers of road, P 12 Front-end sensing equipment installation rate and P per 100 km highway 13 Strengthen the investigation and punishment of key illegal activities.

5. The county traffic safety evaluation method combining RSR method and ordered sample clustering method according to claim 1 is characterized by: When calculating the rank, the rank of the benefit-type indicator is obtained by ranking the indicator data from low to high, and the rank of the cost-type indicator is obtained by ranking the indicator data from high to low. For those with the same indicator data, the average rank is assigned.

6. The county traffic safety evaluation method combining RSR method and ordered sample clustering method according to claim 1 is characterized by: The Spearman correlation coefficient c ij The calculation method is:

Images