Project risk judgment method and device and storage medium

Abstract

The invention provides a project risk judgment method and device and a storage medium. The method comprises the following steps: acquiring historical items; a first risk result information entropy calculation step; a first influence factor information entropy calculation step; a first information gain calculation step; a first node determination step; a current item obtaining step; a current risk result information entropy calculation step; a current influence factor information entropy calculation step; a current information gain calculation step; a gain judgment step; determining a current node; a decision tree generation step; a to-be-judged item obtaining step; a decision tree judgment step; a risk weighted value determination step; and outputting a result. According to the method, traditional subjective judgment is converted into objective decision, and the accuracy of project risk judgment is improved.

Description

Technical Field

[0001] Embodiments of the present invention generally relate to the technical field of vehicle networking function development, and more specifically, to a project risk determination method, device and storage medium. Background Art

[0002] The project management process for the development of Internet of Vehicles functions involves many links and states, which have the following three characteristics: 1. There are many influencing factors, such as the number of software releases, resource adequacy, time to discover problems, severity of problems, etc., which will affect the project progress; 2. The status changes quickly, for example, as problems are constantly discovered during the testing process, software versions are released frequently, etc.; 3. The relationship between factors is complex, for example, whether the remaining number of problems can be fixed depends on the remaining number of software releases, and the number of problems discovered is related to the length of the test, etc.

[0003] Predicting risks based on the status of functional development at different time points can effectively identify and reduce risks in the project. However, in actual work, identifying project risks often requires a lot of project experience, and inexperienced project managers may not be able to identify the existence of risks.

[0004] At present, most of the project risk assessment methods are subjective assessment methods, such as story tree analysis, financial statement method, SWOT analysis method, and scenario analysis method. Take scenario analysis method as an example: according to the diversity of development trends, through systematic analysis of relevant issues inside and outside the system, a variety of possible future prospects are designed, and then the system development situation is described from beginning to end in a scenario and picture similar to writing a movie script. The key points of the above existing methods are to clarify the influencing factors that affect the theme, and the factors are not accurately modeled in the process. Summary of the invention

[0005] In order to solve the above-mentioned problems in the prior art, in a first aspect, an embodiment of the present invention provides a method for determining a target risk, the method comprising: a historical project acquisition step: acquiring the risk result and multiple risk influencing factors of each historical project in a historical project set, wherein each risk result is in one of multiple result types, and each risk influencing factor is in one of multiple factor states; a first risk result information entropy calculation step: calculating the first risk result information entropy according to the probability of each result type in the multiple result types existing in the historical project; a first influencing factor information entropy calculation step: for each risk influencing factor, acquiring the projects in which the risk influencing factor is in multiple different factor states in the historical project set, as multiple first state project sets corresponding to multiple different factor states; calculating the first influencing factor information entropy of the risk influencing factor according to the probability of each result type in the multiple result types existing in the multiple first state project sets; a first information gain calculation step: determining the information gain of each risk influencing factor in the risk influencing factors according to the first risk result information entropy and the first influencing factor information entropy, as the first information gain of each risk influencing factor; a first node determination step: determining the risk influencing factor with the largest first information gain among the risk influencing factors as the first node of the decision tree, and determining the risk influencing factor with the largest first information gain as the first node of the decision tree. The multiple factor states of the risk influencing factors are respectively used as multiple branches under the first node; the current project acquisition step: obtaining from the historical project set a project in which the risk influencing factor represented by the previous node in the decision tree is in the factor state represented by the previous branch as the current project; the current risk result information entropy calculation step: according to the probability of each result type in the multiple result types existing in the current project, calculate the current risk result information entropy; the current influencing factor information entropy calculation step: for each other risk influencing factor except the risk influencing factor represented by the previous node, obtain the project in which the risk influencing factor is in multiple different factor states in the current project as the project related to the multiple different factors a plurality of current state item sets corresponding to the element state; according to the probability of each result type in the plurality of result types existing in the plurality of current state item sets, calculating the current influencing factor information entropy of the risk influencing factor; a current information gain calculation step: according to the current risk result information entropy and the current influencing factor information entropy, determining the information gain of each other risk influencing factor except the risk influencing factor represented by the previous node as the current information gain of each other risk influencing factor; a gain judgment step: judging whether each of the current information gains is lower than a predefined gain threshold; if yes, going to the decision tree generation step; if not, going to the current node determination step;Current node determination step: determine the risk influencing factor with the largest current information gain among the other risk influencing factors as the current node, and use the multiple factor states of the risk influencing factor with the largest current information gain as multiple branches under the current node, and return to the current project acquisition step; decision tree generation step: generate a decision tree according to the determined nodes and branches and the risk results of the historical projects; to-be-determined project acquisition step: obtain the actual factor state of the risk influencing factors of the to-be-determined project; decision tree determination step: determine the nodes passed by the to-be-determined project in the decision tree and the risk results determined by the decision tree according to the decision tree and the actual factor state of the risk influencing factors; risk weighted value determination step: determine the risk weighted value of the to-be-determined project according to the risk influencing factors represented by the passed nodes and the weights of the pre-defined risk influencing factors; result output step: output the risk results determined by the decision tree and the risk weighted value. ;

[0006] In some embodiments, the method includes: for different project stages, based on the historical project sets of different project stages, respectively generating decision trees corresponding to different project stages; and based on the project stage of the project to be determined, using the decision tree corresponding to the project stage to perform risk determination.

[0007] In some implementations, the project is a vehicle networking function development project, and the project phases include one or more of the following: a requirements phase, a development phase, a testing phase, and a maintenance and update phase.

[0008] In some implementations, the risk influencing factors have different predefined weights for different project stages.

[0009] In some embodiments, the outcome types include at risk and not at risk.

[0010] In some embodiments, the result type includes a plurality of different risk levels.

[0011] In some implementations, the first risk result information entropy and the current risk result information entropy are calculated according to the following formula:

[0012]

[0013] Among them, X represents a set of result types. When calculating the first risk result information entropy, P(x) represents the probability that result type x exists in the historical project, and H(X) represents the first risk result information entropy. When calculating the current risk result information entropy, P(x) represents the probability that result type x exists in the current project, and H(X) represents the current risk result information entropy.

[0014] In some embodiments, the first information gain calculation step further includes: determining the difference between the first risk result information entropy and the first influencing factor information entropy as the first information gain of each risk influencing factor among the risk influencing factors; and the current information gain calculation step further includes: determining the difference between the current risk result information entropy and the current influencing factor information entropy as the current information gain of each other risk influencing factor except the risk influencing factor represented by the previous node.

[0015] In some implementations, the project risk is a project risk for developing Internet of Vehicles functions. In addition, the risk influencing factors include one or more of the following: whether the requirements are clear, whether the interface is clear, the number of remaining software releases, whether the vehicle is fully integrated, the degree of completion of the Internet of Vehicles functions in the background, the number of existing problems with the Internet of Vehicles functions, classification of the severity of the problems, classification of the most existing Internet of Vehicles problems, number of vehicle resources, and number of hardware resources.

[0016] In some embodiments, the result types include risk and no risk. Furthermore, after the risk weighted value determination step, the method further includes: in the case where the risk result determined by the decision tree is risky, sorting the risk influencing factors represented by the nodes that the item to be determined passes through in the decision tree according to the weights; and outputting the first predefined number of risk influencing factors in the weight sorting.

[0017] In a second aspect, an embodiment of the present invention proposes a project risk determination device, the device comprising: a historical project acquisition module, configured to obtain the risk results and multiple risk influencing factors of each historical project in a historical project set, wherein each risk result is in one of multiple result types, and each risk influencing factor is in one of multiple factor states; a first risk result information entropy calculation module, configured to calculate the first risk result information entropy according to the probability of each result type in the multiple result types existing in the historical project; a first influencing factor information entropy calculation module, configured to obtain, for each risk influencing factor, the probability of the risk influencing factor being in multiple different states in the historical project set. Items with the same factor state are taken as a plurality of first state item sets corresponding to a plurality of different factor states; the first influencing factor information entropy of the risk influencing factor is calculated according to the probability of each result type in the plurality of result types existing in the plurality of first state item sets; a first information gain calculation module is configured to determine the information gain of each risk influencing factor in the risk influencing factors according to the first risk result information entropy and the first influencing factor information entropy as the first information gain of each risk influencing factor; a first node determination module is configured to determine the risk influencing factor with the largest first information gain among the risk influencing factors as the first node of the decision tree, and the risk influencing factor with the largest first information gain is determined as the first node of the decision tree. The multiple factor states of the decision tree are respectively used as multiple branches under the first node; the current project acquisition module is configured to acquire, from the historical project set, a project in which the risk influencing factor represented by the previous node in the decision tree is in the factor state represented by the previous branch as the current project; the current risk result information entropy calculation module is configured to calculate the current risk result information entropy according to the probability of each result type in the multiple result types existing in the current project; the current influencing factor information entropy calculation module is configured to, for each other risk influencing factor except the risk influencing factor represented by the previous node, acquire, in the current project, projects in which the risk influencing factor is respectively in multiple different factor states as items related to the multiple different factor states a plurality of current state item sets corresponding to the state; according to the probability of each result type in the plurality of current state item sets respectively existing in the plurality of current state item sets, calculating the current influencing factor information entropy of the risk influencing factor; a current information gain calculation module, configured to determine the information gain of each other risk influencing factor except the risk influencing factor represented by the previous node according to the current risk result information entropy and the current influencing factor information entropy, as the current information gain of each other risk influencing factor; a gain judgment module, configured to judge whether each of the current information gains is lower than a predefined gain threshold; if yes, go to the decision tree generation module; if not, go to the current node determination module;A current node determination module, configured to determine the risk influencing factor with the largest current information gain among the other risk influencing factors as the current node, and to use the multiple factor states of the risk influencing factor with the largest current information gain as multiple branches under the current node, and return to the current project acquisition module; a decision tree generation module, configured to generate a decision tree according to the determined nodes and branches and the risk results of the historical projects; a to-be-determined project acquisition module, configured to obtain the actual factor states of the risk influencing factors of the to-be-determined project; a decision tree determination module, configured to determine the nodes passed by the to-be-determined project in the decision tree and the risk results determined by the decision tree according to the decision tree and the actual factor states of the risk influencing factors; a risk weighted value determination module, configured to determine the risk weighted value of the to-be-determined project according to the risk influencing factors represented by the passed nodes and the weights of the predefined risk influencing factors; and a result output module, configured to output the risk results determined by the decision tree and the risk weighted value. ;

[0018] In a third aspect, an embodiment of the present invention provides a storage medium storing computer-readable instructions, which, when executed by a processor, executes the method according to any of the above embodiments.

[0019] The implementation method of the present invention realizes a project risk identification method based on a decision tree. The identification of project experience is converted into a model, and the actual status of each time node can be scientifically estimated whether there is a risk in the result. Through digital modeling, traditional subjective judgments are converted into objective data decisions, which improves the accuracy of project risk judgments, and inexperienced project managers can also make judgments based on the model and then make decisions. The implementation method proposed by the present invention builds a project risk assessment model establishment method based on a decision tree algorithm, constructs a project risk model through historical data, and finally achieves the result of automatically determining the project risk level by inputting the current project status. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] The above and other objects, features and advantages of the embodiments of the present invention will become readily understood by reading the following detailed description with reference to the accompanying drawings. In the accompanying drawings, several embodiments of the present invention are shown in an exemplary and non-limiting manner, in which:

[0021] Figure 1 A flowchart showing a project risk determination method according to an embodiment of the present invention is shown;

[0022] Figure 2 A schematic diagram showing an example of a decision tree according to an embodiment of the present invention;

[0023] Figure 3A block diagram of a project risk determination device according to an embodiment of the present invention is shown.

[0024] In the drawings, the same or corresponding reference numerals represent the same or corresponding parts. DETAILED DESCRIPTION

[0025] The principle and spirit of the present invention will be described below with reference to several exemplary embodiments. It should be understood that these embodiments are provided only to enable those skilled in the art to better understand and implement the present invention, and are not intended to limit the scope of the present invention in any way.

[0026] In one aspect, an embodiment of the present invention provides a project risk determination method. Figure 1 , which shows a flow chart of a project risk determination method 100 according to an embodiment of the present invention. Figure 1 As shown, the method 100 includes steps S101-S116.

[0027] In step S101, i.e., the historical project acquisition step, the risk result and multiple risk influencing factors of each historical project in the historical project set are acquired, wherein each risk result is in one of multiple result types, and each risk influencing factor is in one of multiple factor states.

[0028] As one embodiment of the present invention, the result type may include risk and no risk. As another embodiment of the present invention, the result type may include multiple different risk levels, such as high risk, medium risk, low risk, no risk, and risk levels 1-5, etc.

[0029] As an embodiment of the present invention, project risk is the risk of the Internet of Vehicles function development project. Risk influencing factors may include one or more of the following: whether the requirements are clear (Boolean variable); whether the interface is clear (Boolean variable); the number of remaining software releases (integer), the number of remaining software releases, that is, the number of versions of the software that can be updated for a certain function in the future. The more remaining software releases, the lower the risk; whether the vehicle has been integrated (Boolean variable), that is, whether the software of the function on the vehicle side is already available, TRUE for completion and FALSE for incompleteness, and the risk is higher for incompleteness; the degree of completion of the Internet of Vehicles function on the background side (Boolean variable), that is, the degree of completion of the Internet of Vehicles function on the background side The completion status of the work on the backend, if completed it is TRUE, if not completed it is FALSE; the number of existing problems with the Internet of Vehicles function (integer), that is, the number of problems with the Internet of Vehicles function that currently exist. The threshold for the number of problems can be set in advance. For example, it can be divided into two situations: more than 20 problems and less than 20 problems; classification of problem severity (enumeration type); classification of the most existing Internet of Vehicles problems, for example, which category is A, B, or C with the most existing Internet of Vehicles problems? According to the severity of the problems, they are A, B, and C from severe to least severe; number of vehicle resources (integer); number of hardware resources (integer).

[0030] Based on the results of previous projects, we sorted out the corresponding relationships between risks and various factors in past projects. The method is to first find the risk time points in the project and record the status of the influencing factors at the corresponding time points. Finally, we organize the above data into a matrix or list as a training sample set.

[0031] The following example uses the result type of risky or risk-free in the scenario of Internet of Vehicles function development, and the risk influencing factors include the remaining software release number, whether the vehicle has been integrated, the degree of completion of the Internet of Vehicles function work on the backend, the number of existing problems with the Internet of Vehicles function, and the classification of the most existing Internet of Vehicles problems, a total of five factors, as an example to illustrate the implementation of the present invention. However, it should be noted that the application scenario of the implementation of the present invention is not limited to the Internet of Vehicles function development scenario, and the result type and risk influencing factors are not limited to the types listed in this example.

[0032] Referring to Table 1, an example of a history item collection is shown. Each row in the table represents a history item.

[0033] Table 1. Risk data table of Internet of Vehicles project

[0034]

[0035]

[0036] In step S102, namely the first risk result information entropy calculation step, the first risk result information entropy is calculated according to the probability of each result type among the multiple result types existing in the historical projects.

[0037] As an embodiment of the present invention, the first risk result information entropy is calculated according to the following formula (1):

[0038] H(X)=-∑ x∈X P(x)log2P(x) (1)

[0039] Where X represents the set of result types, P(x) represents the probability of the existence of result type x in historical projects, and H(X) represents the information entropy of the first risk result.

[0040] For example, the information entropy of the data in Table 1 is calculated according to formula (1). According to the risk results in the table, there are two types, risky and risk-free, represented by Yes and No respectively, then the result type set X = [Yes, No]. There are 15 rows of test data, of which there are 6 Yes types and 9 No types. The probability of the existence of a risk item is H(X) = -6 / 15*LOG2(6 / 15)-9 / 15*LOG2(9 / 15) = 0.97.

[0041] In step S103, i.e., the first influencing factor information entropy calculation step, for each risk influencing factor, projects in which the risk influencing factor is in multiple different factor states are obtained in the historical project set as multiple first state project sets corresponding to the multiple different factor states; according to the probability that each of the multiple result types exists in the multiple first state project sets, the first influencing factor information entropy of the risk influencing factor is calculated.

[0042] Taking the data in Table 1 as an example, the information entropy of each risk factor is calculated: for the first column of Table 1, "Number of remaining software releases" has two values, greater than 3 and less than 3, respectively. Therefore, H(X|A = "Number of remaining software releases > 3") = -2 / (2+4)*LOG2(2 / (2+4))-4 / (4+5)*LOG(4 / (4+5)) = 0.918; H(X|A = "Number of remaining software releases < 3") = -4 / (4+5)*LOG2(4 / (4+5))-5 / (4+5)*LOG(5 / (4+5)) = 0.991; H(X1) = 0.4; H(X2) = 0.6. The entropy is calculated as: 0.918*0.4+0.991*0.6=0.961; similarly, the entropy of whether the vehicle integration is completed is 0.961; similarly, the entropy of whether the background is completed is 0.9709; similarly, the entropy of the number of problems is 0.9709; similarly, the entropy of the maximum classification of problems is 0.716.

[0043] In step S104, i.e., the first information gain calculation step, the information gain of each risk influencing factor in the risk influencing factors is determined according to the first risk result information entropy and the first influencing factor information entropy as the first information gain of each risk influencing factor.

[0044] As an embodiment of the present invention, the first information gain calculation step further includes: determining the difference between the first risk result information entropy and the first influencing factor information entropy as the first information gain of each risk influencing factor in the risk influencing factors.

[0045] That is, the first information gain is calculated by the following formula (2):

[0046] G(X,A)=H(X)-H(x|A) (2)

[0047] Among them, H(X) represents the first risk result information entropy, H(x|A) represents the first influencing factor information entropy, and G(X,A) represents the first information gain.

[0048] Taking the data in Table 1 as an example, the information gain of the remaining software releases is 0.97-0.961=0.0089, the information gain of whether the vehicle integration is completed is 0.97-0.961=0.0089, the information gain of whether the background is completed is 0, the information gain of the number of problems is 0, and the information gain of the category with the most problems is 0.253.

[0049] In step S105, i.e., the first node determination step, the risk influencing factor with the largest first information gain among the risk influencing factors is determined as the first node of the decision tree, and multiple factor states of the risk influencing factor with the largest first information gain are respectively used as multiple branches under the first node.

[0050] In the above specific example combining the data in Table 1, the item with the maximum gain is taken. It can be seen that the "most problematic categories" is the first child node of the tree, that is, the "most problematic categories" factor has the greatest impact on risk judgment.

[0051] Next, the other nodes of the decision tree are derived in the same way to obtain the decision tree model.

[0052] In step S106, ie, the current project acquisition step, a project whose risk influencing factor represented by a previous node in the decision tree is in the factor state represented by a previous branch is acquired from the historical project set as the current project.

[0053] In step S107, ie, the step of calculating the current risk result information entropy, the current risk result information entropy is calculated according to the probability of each of the multiple result types existing in the current project.

[0054] As an embodiment of the present invention, similar to the calculation of the first risk result information entropy, the current risk result information entropy is calculated by the following formula (3):

[0055] H(X)=-∑ x∈X P(x)log2P(x) (3)

[0056] Among them, X represents the set of result types, P(x) represents the probability that result type x exists in the current project, and H(X) represents the information entropy of the current risk result.

[0057] In step S108, i.e., the current influencing factor information entropy calculation step, for each other risk influencing factor except the risk influencing factor represented by the previous node, the projects in which the risk influencing factor is in multiple different factor states are obtained in the current project as multiple current state project sets corresponding to the multiple different factor states; according to the probability that each of the multiple result types exists in the multiple current state project sets, the current influencing factor information entropy of the risk influencing factor is calculated.

[0058] In step S109, i.e., the current information gain calculation step, the information gain of each other risk influencing factor except the risk influencing factor represented by the previous node is determined according to the current risk result information entropy and the current influencing factor information entropy as the current information gain of each other risk influencing factor.

[0059] As an embodiment of the present invention, the current information gain calculation step further includes: determining the difference between the current risk result information entropy and the current influencing factor information entropy as the current information gain of each other risk influencing factor except the risk influencing factor represented by the previous node.

[0060] In step S110, i.e., the gain determination step, it is determined whether each of the current information gains is lower than a predefined gain threshold; if yes, the process proceeds to the decision tree generation step; if no, the process proceeds to the current node determination step. The predefined gain threshold is, for example, zero.

[0061] In step S111, i.e., the current node determination step, the risk influencing factor with the largest current information gain among other risk influencing factors is determined as the current node, and multiple factor states of the risk influencing factor with the largest current information gain are respectively used as multiple branches under the current node, and return to the current project acquisition step.

[0062] In step S112, ie, the decision tree generation step, a decision tree is generated according to the determined nodes and branches and the risk results of the historical projects.

[0063] It should be noted that in the process of generating a decision tree, the above iterative steps need to be performed for each branch of each previous node until the information gain under the branch is lower than the gain threshold, and then further calculation of the branch is stopped.

[0064] The decision tree generated according to the above steps S101-S112 is a decision analysis method that is based on the known probability of occurrence of various situations and uses a decision tree to obtain the probability that the expected value of the net present value is greater than or equal to zero, evaluate project risks, and judge its feasibility. It is a graphical method that intuitively uses probability analysis. A decision tree is a prediction model that represents a mapping relationship between object attributes and object values.

[0065] refer to Figure 2 , which shows a schematic diagram of an example of a decision tree according to an embodiment of the present invention.

[0066] After obtaining the decision tree model, the current project status can be included in it for prediction. The impact of the current input factors on the project can be determined based on the decision tree model generated above, and the final output is whether there is a risk.

[0067] In step S113, namely the step of obtaining the project to be determined, the actual factor status of the risk influencing factor of the project to be determined is obtained.

[0068] In step S114, ie, the decision tree determination step, the nodes that the item to be determined passes through in the decision tree and the risk result determined by the decision tree are determined according to the decision tree and the actual factor states of the risk influencing factors.

[0069] In step S115, ie, the risk weighted value determination step, the risk weighted value of the project to be determined is determined according to the risk influencing factors represented by the passed nodes and the weights of the predefined risk influencing factors.

[0070] Predefined weights are added to the risk item, such as a weight of 1.0 for the remaining software release rights and a weight of 0.8 for whether vehicle integration is completed. The weights can be set according to the severity of each input factor, and then the sum is calculated. When the result is that there is a risk, the weight can be considered. The larger the weight, the higher the risk and the more difficult it is to resolve.

[0071] In step S116, i.e., the result output step, the risk result determined by the decision tree and the risk weighted value are output.

[0072] For example, when Figure 2After the decision tree shown is formed, if a Class A problem occurs in the project to be judged, the remaining software release number is greater than 3, and the number of problems is greater than 20, then it can be judged that the project is at risk, and the sum of the risk weights is (setting: problem classification 0.5, remaining software releases 1.5, number of problems 0.7) 2.7; similarly, if a Class A ticket occurs and the number of software releases is less than 3, there is also a risk, and the risk coefficient is 2; it can be seen that the more influencing factors there are, the greater the sum of the risk weights will be.

[0073] Optionally, the risk determination results and risk weighted values ​​output by the method can be displayed through the agile development project tracking dashboard. For example, a risk information assessment column is set up in the agile development project tracking dashboard as the data input of the risk algorithm module. The risk assessment system automatically reads the updated data from the agile development factor assessment column and calculates the risk assessment results. When the risk assessment results show that there is a risk, there will be a warning message and the factors with the greatest current risk impact will be given. Finally, the decision tree model and the degree of impact of each risk factor are displayed on the display page.

[0074] By using the agile development dashboard, you can maximize visualization and solve team communication barriers. Adding risk information to the dashboard as a risk assessment module can be combined with other dashboard data to predict project risks.

[0075] The implementation method of the present invention realizes a project risk identification method based on a decision tree. The identification of project experience is converted into a model, and the actual status of each time node can be scientifically estimated whether the result has risks. Through digital modeling, the traditional subjective judgment is converted into objective data decision, which improves the accuracy of project risk judgment. Inexperienced project managers can also make judgments based on the model and then make decisions. The implementation method proposed by the present invention builds a project risk assessment model establishment method based on a decision tree algorithm, constructs a project risk model through historical data, and finally achieves the result of automatically determining the project risk level by inputting the current project status. Project influencing factors can be set according to the characteristics of different projects. Furthermore, with the update of historical project data, it is convenient to improve the accuracy of the model through continuous training of the model.

[0076] As an embodiment of the present invention, the project is a vehicle networking function development project, and the project phases may include one or more of the following: a demand phase, a development phase, a testing phase, and a maintenance and update phase.

[0077] Optionally, for different project stages, decision trees corresponding to different project stages can be generated based on historical project sets of different project stages; and risk determination can be performed using the decision tree corresponding to the project stage according to the project stage of the project to be determined.

[0078] Alternatively, risk influencing factors have different predefined weights for different project phases. As an example, a project can be divided into three major phases: requirements phase, development phase, testing phase, and maintenance and update phase. In each phase, the weight of each factor will change. In the demand stage, if the factors in the risk model are selected, the weights of whether the demand is clear (Boolean variable) and whether the interface is clear (Boolean variable) can be set to 0.8, 0.9, and 1.0 in this stage, and other factors are given weights between 0 and 0.8; in the development stage, whether the vehicle has completed integration (Boolean variable) and whether the background function is complete (Boolean variable) should be given weights between 0.8 and 1.0, and other factors are given weights between 0 and 0.8; in the testing stage, the number of existing problems (integer), problem severity classification (enumeration type), number of vehicle resources (integer), and number of hardware resources (integer) should be 0.8, 0.9, and 1.0, and other factors are given weights between 0 and 0.8; in the maintenance and update stage, the remaining software releases (integer), the number of existing problems (integer), and the problem severity classification (enumeration type) are set to 0.8, 0.9, and 1.0, and other factors are given weights between 0 and 0.8.

[0079] By assigning different weights to each factor at different project stages, the most important factors affecting project risk can be reflected.

[0080] Both of the above methods can achieve differentiated risk assessment for different project stages. The two methods can be implemented independently or in combination.

[0081] As an embodiment of the present invention, when the result types include risky and non-risky, after the risk weighted value determination step, the method may also include: when the risk result determined by the decision tree is risky, sorting the risk influencing factors represented by the nodes passed by the project to be determined in the decision tree according to the weights; and outputting a predefined number of risk influencing factors in the weight sorting.

[0082] In another aspect, an embodiment of the present invention provides a project risk determination device. Figure 3 , which shows a block diagram of a project risk determination device according to an embodiment of the present invention. The device includes modules 301-316.

[0083] The historical project acquisition module 301 can be configured to obtain the risk results and multiple risk influencing factors of each historical project in the historical project set, wherein each risk result is in one of multiple result types, and each risk influencing factor is in one of multiple factor states.

[0084] The first risk result information entropy calculation module 302 may be configured to calculate the first risk result information entropy according to the probability of each result type in the multiple result types existing in the historical project.

[0085] The first influencing factor information entropy calculation module 303 can be configured to obtain, for each risk influencing factor, projects in the historical project set in which the risk influencing factor is in multiple different factor states, as multiple first state project sets corresponding to the multiple different factor states; and calculate the first influencing factor information entropy of the risk influencing factor according to the probability of each of the multiple result types existing in the multiple first state project sets.

[0086] The first information gain calculation module 304 may be configured to determine the information gain of each risk influencing factor among the risk influencing factors according to the first risk result information entropy and the first influencing factor information entropy as the first information gain of each risk influencing factor.

[0087] The first node determination module 305 can be configured to determine the risk influencing factor with the largest first information gain among the risk influencing factors as the first node of the decision tree, and use multiple factor states of the risk influencing factor with the largest first information gain as multiple branches under the first node.

[0088] The current project acquisition module 306 may be configured to acquire, from the historical project set, a project in which the risk influencing factor represented by a previous node in the decision tree is in the factor state represented by a previous branch as the current project.

[0089] The current risk result information entropy calculation module 307 may be configured to calculate the current risk result information entropy according to the probability that each result type among the multiple result types exists in the current project.

[0090] The current influencing factor information entropy calculation module 308 can be configured to obtain, for each other risk influencing factor except the risk influencing factor represented by the previous node, projects in the current project in which the risk influencing factor is in multiple different factor states, as multiple current state project sets corresponding to the multiple different factor states; and calculate the current influencing factor information entropy of the risk influencing factor according to the probability that each of the multiple result types exists in the multiple current state project sets.

[0091] The current information gain calculation module 309 can be configured to determine the information gain of each other risk influencing factor except the risk influencing factor represented by the previous node according to the current risk result information entropy and the current influencing factor information entropy as the current information gain of each other risk influencing factor.

[0092] The gain judgment module 310 may be configured to judge whether each of the current information gains is lower than a predefined gain threshold; if yes, then go to the decision tree generation module; if no, then go to the current node determination module.

[0093] The current node determination module 311 can be configured to determine the risk influencing factor with the largest current information gain among other risk influencing factors as the current node, and use the multiple factor states of the risk influencing factor with the largest current information gain as multiple branches under the current node, and return it to the current project acquisition module.

[0094] The decision tree generation module 312 may be configured to generate a decision tree according to the determined nodes and branches and risk results of historical projects.

[0095] The to-be-determined project acquisition module 313 may be configured to acquire the actual factor status of the risk influencing factor of the to-be-determined project.

[0096] The decision tree determination module 314 may be configured to determine the nodes that the item to be determined passes through in the decision tree and the risk result determined by the decision tree according to the decision tree and the actual factor status of the risk influencing factors.

[0097] The risk weighted value determination module 315 may be configured to determine the risk weighted value of the project to be determined based on the risk impact factors represented by the passed nodes and the weights of the predefined risk impact factors.

[0098] The result output module 316 may be configured to output the risk result determined by the decision tree and the risk weight value.

[0099] It should be noted that the functions implemented by each module in the project risk assessment device proposed in the embodiment of the present invention correspond one-to-one to the various steps of the project risk assessment method described above. For its specific implementation methods, examples and beneficial effects, please refer to the above description of the method.

[0100] In yet another aspect, an embodiment of the present invention provides a storage medium storing computer-readable instructions, which, when executed by a processor, executes the project risk determination method described in any of the above embodiments.

[0101] The implementation method proposed in the present invention realizes a project risk identification method based on a decision tree. The identification of project experience is converted into a model, and according to the actual status of each time node, it is possible to scientifically estimate whether there is a risk in the result. Through digital modeling, traditional subjective judgments are converted into objective data decisions, thereby improving the accuracy of project risk judgments. Inexperienced project managers can also make judgments based on the model and then make decisions. The implementation method proposed in the present invention builds a project risk assessment model establishment method based on a decision tree algorithm, constructs a project risk model through historical data, and finally achieves the result of automatically determining the project risk level by inputting the current project status. Furthermore, with the update of historical project data, the model accuracy can be improved by continuous training of the model.

[0102] The project risk determination method, device and storage medium proposed in the embodiment of the present invention establish a model based on a big data learning algorithm. As the input amount of model data increases, the model can be continuously refined, and finally the function of automatically determining project risk warnings based on the input of existing project status factors is achieved, thereby improving the prediction of project warnings. At the same time, the factors with the greatest current risk impact can be output, thereby breaking through the project risk points.

[0103] For the purpose of illustration, the foregoing description of the embodiments of the present invention has been given, which is not exhaustive nor intended to limit the present invention to the disclosed exact form. It will be appreciated by those skilled in the art that various changes can be made without departing from the scope of the present invention, and that the elements therein can be replaced with equivalents. In addition, without departing from the basic scope of the present invention, many modifications can be made so that specific situations or materials are adapted to the teachings of the present invention. Therefore, the present invention is not intended to be limited to the specific embodiments disclosed as the best mode for realizing the present invention, and the present invention will include all embodiments falling within the scope of the appended claims.

Claims

1. A project risk determination method, characterized in that: The method comprises: Historical project acquisition step: acquiring risk results and multiple risk influencing factors of each historical project in the historical project set, wherein each risk result is in one of multiple result types, and each risk influencing factor is in one of multiple factor states; The first risk result information entropy calculation step: calculating the first risk result information entropy according to the probability of each result type in the multiple result types existing in the historical project; The first influencing factor information entropy calculation step: for each risk influencing factor, obtaining items in the historical item set in which the risk influencing factor is in a plurality of different factor states respectively, as a plurality of first state item sets corresponding to the plurality of different factor states; calculating the first influencing factor information entropy of the risk influencing factor according to the probability that each of a plurality of result types exists in the plurality of first state item sets respectively; First information gain calculation step: determining the information gain of each risk influencing factor in the risk influencing factors according to the first risk result information entropy and the first influencing factor information entropy as the first information gain of each risk influencing factor; First node determination step: determining the risk influencing factor with the largest first information gain among the risk influencing factors as the first node of the decision tree, and using multiple factor states of the risk influencing factor with the largest first information gain as multiple branches under the first node respectively; Current project acquisition step: acquiring, from the historical project set, a project whose risk influencing factor represented by a previous node in the decision tree is in the factor state represented by a previous branch, as the current project; Current risk result information entropy calculation step: calculating the current risk result information entropy according to the probability of each result type in the current project; The current influencing factor information entropy calculation step: for each other risk influencing factor except the risk influencing factor represented by the previous node, in the current project, the items of the risk influencing factor in a plurality of different factor states are obtained as a plurality of current state item sets corresponding to the plurality of different factor states; according to the probability that each of the plurality of result types exists in the plurality of current state item sets, the current influencing factor information entropy of the risk influencing factor is calculated; Current information gain calculation step: determining the information gain of each other risk influencing factor except the risk influencing factor represented by the previous node according to the current risk result information entropy and the current influencing factor information entropy as the current information gain of each other risk influencing factor; Gain judgment step: judging whether each of the current information gains is lower than a predefined gain threshold; if yes, proceeding to the decision tree generation step; if no, proceeding to the current node determination step; Current node determination step: determining the risk influencing factor with the largest current information gain among the other risk influencing factors as the current node, taking multiple factor states of the risk influencing factor with the largest current information gain as multiple branches under the current node, and returning to the current project acquisition step; Decision tree generation step: generating a decision tree according to the determined nodes and branches and the risk results of the historical projects; Step of obtaining the project to be determined: obtaining the actual factor status of the risk influencing factors of the project to be determined; Decision tree determination step: determining the nodes that the item to be determined passes through in the decision tree and the risk result determined by the decision tree according to the decision tree and the actual factor status of the risk influencing factors; The risk weighted value determination step: determining the risk weighted value of the project to be determined according to the risk influencing factors represented by the passed nodes and the weights of the predefined risk influencing factors; Result output step: output the risk result determined by the decision tree and the risk weighted value.

2. The method according to claim 1, characterized in that The method comprises: For different project stages, decision trees corresponding to different project stages are generated according to the historical project sets of different project stages; According to the project stage of the project to be judged, the risk judgment is carried out using the decision tree corresponding to the project stage.

3. The method according to claim 2, characterized in that The project is a vehicle networking function development project, and the project stages include one or more of the following: a demand stage, a development stage, a testing stage, and a maintenance and update stage.

4. The method according to claim 1 or 2, characterized in that: For different project stages, the risk influencing factors have different pre-defined weights.

5. The method according to claim 1, characterized in that The outcome types include risky and no risk.

6. The method according to claim 1, characterized in that The result types include a number of different risk levels.

7. The method according to claim 1, characterized in that The first risk result information entropy and the current risk result information entropy are calculated according to the following formula: Among them, X represents a set of result types. When calculating the first risk result information entropy, P(x) represents the probability that result type x exists in the historical project, and H(X) represents the first risk result information entropy. When calculating the current risk result information entropy, P(x) represents the probability that result type x exists in the current project, and H(X) represents the current risk result information entropy.

8. The method according to claim 1, characterized in that The first information gain calculation step further includes: determining the difference between the first risk result information entropy and the first influencing factor information entropy as the first information gain of each risk influencing factor among the risk influencing factors; and The current information gain calculation step further includes: determining the difference between the current risk result information entropy and the current influencing factor information entropy as the current information gain of each other risk influencing factor except the risk influencing factor represented by the previous node.

9. The method according to claim 1, characterized in that: The project risk is the risk of the Internet of Vehicles function development project, and, The risk influencing factors include one or more of the following: whether the requirements are clear, whether the interfaces are clear, the number of remaining software releases, whether the vehicle has been integrated, the degree of completion of the Internet of Vehicles function in the background, the number of existing problems with the Internet of Vehicles function, classification of problem severity, classification of the most existing Internet of Vehicles problems, the number of vehicle resources, and the number of hardware resources.

10. The method according to claim 1, characterized in that The result types include risk and no risk, and after the risk weighted value determination step, the method further includes: If the risk result determined by the decision tree is risky, the risk influencing factors represented by the nodes passed by the item to be determined in the decision tree are sorted according to the weights; Output a predefined number of risk influencing factors in weight ranking.

11. A project risk determination device, characterized in that: The device comprises: A historical project acquisition module configured to acquire a risk result and a plurality of risk influencing factors of each historical project in the historical project set, wherein each risk result is in one of a plurality of result types, and each risk influencing factor is in one of a plurality of factor states; A first risk result information entropy calculation module, configured to calculate the first risk result information entropy according to the probability of each result type in the multiple result types existing in the historical project; A first influencing factor information entropy calculation module is configured to obtain, for each risk influencing factor, items in the historical item set in which the risk influencing factor is in a plurality of different factor states, as a plurality of first state item sets corresponding to the plurality of different factor states; and calculate the first influencing factor information entropy of the risk influencing factor according to the probability that each of a plurality of result types exists in the plurality of first state item sets; A first information gain calculation module, configured to determine the information gain of each of the risk influencing factors according to the first risk result information entropy and the first influencing factor information entropy as the first information gain of each risk influencing factor; A first node determination module is configured to determine the risk influencing factor with the largest first information gain among the risk influencing factors as the first node of the decision tree, and to use multiple factor states of the risk influencing factor with the largest first information gain as multiple branches under the first node; A current project acquisition module configured to acquire, from the historical project set, a project in which the risk influencing factor represented by a previous node in the decision tree is in the factor state represented by a previous branch, as the current project; A current risk result information entropy calculation module, configured to calculate the current risk result information entropy according to the probability of each result type in the multiple result types existing in the current project; The current influencing factor information entropy calculation module is configured to obtain, for each other risk influencing factor except the risk influencing factor represented by the previous node, items in the current item in which the risk influencing factor is in a plurality of different factor states respectively, as a plurality of current state item sets corresponding to the plurality of different factor states; calculate the current influencing factor information entropy of the risk influencing factor according to the probability that each of the plurality of result types exists in the plurality of current state item sets respectively; a current information gain calculation module, configured to determine, according to the current risk result information entropy and the current influencing factor information entropy, the information gain of each other risk influencing factor except the risk influencing factor represented by the previous node as the current information gain of each other risk influencing factor; A gain judgment module, configured to judge whether each of the current information gains is lower than a predefined gain threshold; if yes, then go to the decision tree generation module; if no, then go to the current node determination module; A current node determination module is configured to determine the risk influencing factor with the largest current information gain among the other risk influencing factors as the current node, and use multiple factor states of the risk influencing factor with the largest current information gain as multiple branches under the current node, and return them to the current project acquisition module; A decision tree generation module configured to generate a decision tree according to the determined nodes and branches and the risk results of the historical projects; A module for obtaining projects to be determined, configured to obtain the actual factor status of risk influencing factors of projects to be determined; A decision tree determination module, configured to determine the nodes passed by the item to be determined in the decision tree and the risk result determined by the decision tree according to the decision tree and the actual factor status of the risk influencing factor; A risk weighted value determination module, configured to determine the risk weighted value of the project to be determined according to the risk influencing factors represented by the passed nodes and the weights of the predefined risk influencing factors; The result output module is configured to output the risk result determined by the decision tree and the risk weighted value.

12. A storage medium storing computer-readable instructions, wherein when the instructions are executed by a processor, the method according to any one of claims 1 to 10 is executed.

Images