A pre-warning method, device, equipment and medium

Abstract

The application relates to the technical field of trend prediction, in particular to a pre-warning method and device, equipment and medium. In the embodiment of the application, research and development information of a target project to be pre-warned is acquired, wherein the research and development information comprises basic data and process data of a research and development process; a first feature vector is determined according to the basic data; a second feature vector is determined according to the process data; the first feature vector and the second feature vector are input into a trained model, and a risk probability of next research and development of the target project output by the model is acquired. Since the electronic equipment generates the first feature vector and the second feature vector based on the research and development information of the target project, and whether a risk will occur in the next research and development process is predicted according to the first feature vector and the second feature vector, the efficiency and accuracy of prediction are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of trend prediction, and in particular to a pre-warning method and device, equipment and a medium. BACKGROUND

[0002] The development of a software project is related to the progress of the project. In the entire life cycle of the development of the project, a development team needs to identify and evaluate risks in the development process, so as to maximize the reduction of project risks and enable the project to be completed efficiently and with high quality.

[0003] The related art method for predicting and evaluating risks in the development process is to track and evaluate risks manually. Specifically, a technical personnel periodically collects or simulates data in the existing development process, and predicts whether a risk will occur through human experience. This risk prediction method is low in efficiency and poor in accuracy. Moreover, this method requires high experience of the technical personnel, and therefore cannot achieve continuous identification and early warning of risks. SUMMARY

[0004] The present application provides a pre-warning method, device, equipment and medium, to solve the problem of low risk prediction efficiency and poor accuracy in predicting and evaluating risks in the development process in the prior art.

[0005] The present application provides a pre-warning method, which comprises the following steps:

[0006] Obtaining development information of a target project to be pre-warned, wherein the development information comprises basic data and process data of the development process, wherein the basic data at least comprises team size and development language, and the process data at least comprises defect density, iteration completion rate, average defect repair time and build success rate in the development process;

[0007] Determining a first feature vector according to the basic data, pre-stored first values corresponding to each sub-basic data and first positions corresponding to each first value in the first feature vector;

[0008] Determining a second feature vector according to the process data, pre-stored second values corresponding to each sub-process data and second positions corresponding to each second value in the second feature vector;

[0009] Inputting the first feature vector and the second feature vector into a trained model to obtain a risk probability of next development of the target project output by the model.

[0010] Further, the step of determining the first feature vector according to the basic data, pre-stored first values corresponding to each sub-basic data and first positions corresponding to each first value in the first feature vector comprises:

[0011] determining each target sub-process data contained in the process data;

[0012] For each target sub-process data, determining a target second value corresponding to the target sub-process data according to the second value corresponding to each sub-process data pre-stored; determining a target second position corresponding to the target sub-process data according to the second position corresponding to each sub-process data pre-stored in the second feature vector; and writing the target second value into the target second position.

[0013] Further, the determining the second feature vector according to the process data, the second value corresponding to each sub-process data pre-stored and the second position includes:

[0014] determining each target sub-process data contained in the process data;

[0015] For each target sub-process data, determining a target second value corresponding to the target sub-process data according to the second value corresponding to each sub-process data pre-stored; determining a target second position corresponding to the target sub-process data according to the second position corresponding to each sub-process data pre-stored in the second feature vector; and writing the target second value into the target second position.

[0016] Further, the obtaining the R&D information of the target project to be pre-warned includes:

[0017] obtaining candidate R&D information corresponding to each iteration process of the target project pre-stored;

[0018] determining the candidate R&D information corresponding to the latest iteration as the R&D information of the target project.

[0019] Further, the obtaining the R&D information of the target project to be pre-warned includes:

[0020] obtaining candidate R&D information corresponding to each iteration process of the target project pre-stored;

[0021] determining each candidate R&D information corresponding to a preset number of iterations closest to the current time as the R&D information, respectively;

[0022] Before the first feature vector and the second feature vector are input into the trained model, the method further includes:

[0023] calculating a first mean value of the first value corresponding to each first position according to each first feature vector corresponding to each R&D information, determining a first mean value vector corresponding to each first feature vector, and determining the first mean value vector as the first feature vector input into the model;

[0024] According to each second feature vector corresponding to each research and development information, a second mean value of a second numerical value corresponding to each second position is calculated, a second mean value vector corresponding to each second feature vector is determined, and the second mean value vector is determined as the second feature vector input into the model.

[0025] Further, the model is a Wide&Deep model;

[0026] Inputting the first feature vector and the second feature vector into the trained model includes:

[0027] Inputting the first feature vector into a Wide sub-model of the Wide&Deep model;

[0028] Splicing the first feature vector and the second feature vector according to a preset order to obtain a third feature vector, and inputting the third feature vector into a Deep sub-model of the Wide&Deep model.

[0029] Embodiments of the present application also provide a warning device, the device comprising:

[0030] An acquisition module is configured to acquire research and development information of a target project to be warned, wherein the research and development information includes basic data and process data of a research and development process, wherein the basic data at least includes team size and development language, and the process data at least includes defect density, iteration completion rate, defect average repair time, and build success rate in the research and development process;

[0031] A processing module is configured to determine a first feature vector according to the basic data, pre-stored first numerical values corresponding to each sub-basic data, and first positions corresponding to each first numerical value in the first feature vector; and determine a second feature vector according to the process data, pre-stored second numerical values corresponding to each sub-process data, and second positions corresponding to each second numerical value in the second feature vector.

[0032] A prediction module is configured to input the first feature vector and the second feature vector into a trained model, and acquire a risk probability of next research and development of the target project output by the model.

[0033] Further, the processing module is specifically configured to determine each target sub-basic data included in the basic data; for each target sub-basic data, determine a target first numerical value corresponding to the target sub-basic data according to pre-stored first numerical values corresponding to each sub-basic data; determine a target first position corresponding to the target sub-basic data according to pre-stored first positions corresponding to each sub-basic data in the first feature vector; and write the target first numerical value into the target first position.

[0034] Further, the processing module is specifically configured to determine each target sub-process data contained in the process data; for each target sub-process data, determine a target second value corresponding to the target sub-process data according to a second value corresponding to each sub-process data pre-stored; determine a target second position corresponding to the target sub-process data according to a second position corresponding to each sub-process data in the second feature vector; and write the target second value into the target second position.

[0035] Further, the obtaining module is specifically configured to obtain candidate R&D information corresponding to each iteration process of the target project pre-stored; and determine candidate R&D information corresponding to the latest iteration as the R&D information of the target project.

[0036] Further, the obtaining module is specifically configured to obtain candidate R&D information corresponding to each iteration process of the target project pre-stored; and determine each candidate R&D information corresponding to a preset number of iterations closest to the current time as the R&D information, respectively.

[0037] The processing module is further configured to calculate a first mean value of the first value corresponding to each first position according to each first feature vector corresponding to each R&D information, determine a first mean value vector corresponding to each first feature vector, and determine the first mean value vector as the first feature vector input into the model; and calculate a second mean value of the second value corresponding to each second position according to each second feature vector corresponding to each R&D information, determine a second mean value vector corresponding to each second feature vector, and determine the second mean value vector as the second feature vector input into the model.

[0038] Further, the model is a Wide&Deep model.

[0039] The prediction module is specifically configured to input the first feature vector into a Wide sub-model of the Wide&Deep model; splice the first feature vector and the second feature vector according to a preset order to obtain a third feature vector, and input the third feature vector into a Deep sub-model of the Wide&Deep model.

[0040] Embodiments of the present application also provide an electronic device, which comprises a processor configured to implement steps of the early warning method according to any one of the above embodiments when executing a computer program stored in a memory.

[0041] The embodiment of the present application further provides a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to implement the steps of the early warning method in any one of the above.

[0042] The embodiment of the present application further provides a computer program product, which comprises computer program code, and when the computer program code is run on a computer, the computer program code causes the computer to execute the steps of the early warning method in any one of the above.

[0043] In the embodiment of the present application, the research and development information of a target project to be warned is acquired, wherein the research and development information comprises basic data and process data of a research and development process, wherein the basic data at least comprises team size and development language, and the process data at least comprises defect density, iteration completion rate, average defect repair time and build success rate in the research and development process; a first feature vector is determined according to the basic data, pre-stored first values corresponding to each sub-basic data and first positions corresponding to each first value in the first feature vector; a second feature vector is determined according to the process data, pre-stored second values corresponding to each sub-process data and second positions corresponding to each second value in the second feature vector; and the first feature vector and the second feature vector are input into a trained model to acquire a risk probability of next research and development of the target project output by the model. In the embodiment of the present application, the electronic device generates the first feature vector and the second feature vector based on the research and development information in the previous research and development process of the target project, and predicts whether a risk will occur in the next research and development process according to the first feature vector and the second feature vector, thereby improving the efficiency and accuracy of prediction. BRIEF DESCRIPTION OF DRAWINGS

[0044] In order to more clearly illustrate the technical solutions of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0045] Figure 1 A pre-warning process schematic diagram provided for the embodiment of the present application;

[0046] Figure 2 A research and development information schematic diagram provided for the embodiment of the present application;

[0047] Figure 3 A code quality level schematic diagram provided for the embodiment of the present application;

[0048] Figure 4 A pre-warning device structure schematic diagram provided for the embodiment of the present application;

[0049] Figure 5 An electronic device structure schematic diagram is provided for the embodiments of the present application. DETAILED DESCRIPTION

[0050] In order to make the purposes, technical solutions and advantages of the present application clearer, the present application will be further described in detail below with reference to the drawings. Obviously, the described embodiments are only some of the embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of protection of the present application.

[0051] In order to realize the prediction and evaluation of the risks in the project research and development process and improve the efficiency and accuracy of the prediction, the present application provides a pre-warning method, device, equipment and medium.

[0052] In the embodiments of the present application, the research and development information of a target project to be pre-warned is acquired, wherein the research and development information includes basic data and process data of the research and development process, wherein the basic data at least includes team size and development language, and the process data at least includes defect density, iteration completion rate, average defect repair time and build success rate in the research and development process; a first feature vector is determined according to the basic data, pre-stored first values corresponding to each sub-basic data and first positions corresponding to each first value in the first feature vector; a second feature vector is determined according to the process data, pre-stored second values corresponding to each sub-process data and second positions corresponding to each second value in the second feature vector; the first feature vector and the second feature vector are input into a trained model, and a risk probability of the next research and development of the target project output by the model is acquired.

[0053] Embodiment 1:

[0054] Figure 1 A pre-warning process schematic diagram is provided for the embodiments of the present application, and the process includes:

[0055] S101: Acquire research and development information of a target project to be pre-warned, wherein the research and development information includes basic data and process data of the research and development process, wherein the basic data includes team size and development language, and the process data includes defect density, iteration completion rate, average defect repair time and build success rate in the research and development process.

[0056] The pre-warning method provided by the embodiments of the present application is applied to an electronic device, wherein the electronic device can be a server or a PC, etc.

[0057] In the embodiments of the present application, when the R&D process is prewarned, each project in the R&D process is prewarned. Specifically, the electronic device obtains R&D information of a target project to be prewarned. The R&D information is information generated in the process of having preformed R&D on the target project. The R&D information can be input into and saved in the electronic device previously, or can be directly obtained by the electronic device from other devices. If the R&D information is obtained by the electronic device from other devices, the electronic device saves the R&D information after obtaining the R&D information.

[0058] Specifically, in the embodiments of the present application, the R&D information includes basic data and process data of the R&D process. The basic data is basic configuration information in the R&D process, such as team size and development language, etc. The process data is information generated in the R&D process, such as defect density, iteration completion rate, average defect repair time, build success rate, etc.

[0059] It should be noted that in the embodiments of the present application, the basic data can further include at least one of the following data: industry to which the target project belongs, technical field to which the target project belongs, ratio of development and test personnel, and iteration period, etc. The process data can further include at least one of the following data: code submission frequency, code merging frequency, build and deployment frequency, build and deployment time length, code quality level, unit test coverage, number of online problems, etc.

[0060] In the embodiments of the present application, the original process data collected by the electronic device includes code submission times, code submitter number, code merging times, unit test coverage, unit test pass rate, build and deployment times, build and deployment success times, build and deployment time length, iteration planning user story number, iteration actual completed user story number, iteration internal defect number, defect severity, defect repair time length, code duplication degree, code maintainability level, code reliability level, code security level, and number of online problems, etc. After obtaining the original process data, the electronic device determines the process data required for prewarning according to the original process data.

[0061] Specifically, in the embodiments of the present application, the electronic device determines the ratio of the iteration internal defect number to the iteration internal user story number as the defect density; determines the ratio of the actual completed story number to the planned user story number as the iteration completion rate; determines the ratio of the build and deployment success times to the build and deployment times as the build success rate; determines the daily code submission times per person as the code submission frequency; determines the ratio of the iteration internal code merging times to the iteration internal user story number as the code merging frequency; and determines the code quality level according to the code duplication degree, the code maintainability level, the code reliability level, the code security level, and a pre-defined rule, etc.

[0062] Figure 2 The schematic diagram of the research and development information provided for the embodiments of the present application is shown in FIG. 1. Figure 2 The research and development information shown includes basic data and process data, the basic data including the industry to which the target project belongs, the technical field to which the target project belongs, the team size, the development language, and the iteration period, and the process data including the code submission frequency, the code merging frequency, the build and deployment frequency, the deployment time length, the average defect repair time length, the code quality level, the unit test coverage, the build success rate, the defect density, the number of online problems, the iteration completion rate, the defect risk, and the delay risk.

[0063] Among them, the industry to which the target project belongs can be the Internet, finance, e-commerce, medical treatment, automobile, or other, the technical field to which the target project belongs can be big data, artificial intelligence, blockchain, cloud computing, or other, the team size can be less than 5 people, 5-10 people, 10-20 people, or more than 20 people, the development language can be Java, C / C++, Python, Golong, or other, and the iteration period can be 1 week, 2 weeks, 2-4 weeks, or more than 4 weeks.

[0064] In the process data, the code submission frequency is the number of code submissions per day per person, the code merging frequency is the ratio of the number of code merging to the number of user stories, the build and deployment frequency is the number of build and deployment per day, the deployment time length is the average time for one build and deployment, the average defect repair time length is the average time for repairing one defect, the code quality level can be A level, B level, C level, D level, or E level, etc., the unit test coverage is the ratio of the code covered by the unit test, the build success rate is the ratio of the number of successful build to the total number of build, the defect density is the ratio of the number of defects to the number of user stories, the iteration completion rate is the ratio of the actual number of completed user stories to the planned number of user stories, the defect risk can be at risk and no risk, wherein whether there is a defect risk is determined according to the defect density, if the defect density is not lower than a preset defect density threshold, it is determined that there is a risk, otherwise it is determined that there is no risk, and the delay risk can be at risk and no risk, wherein whether there is a delay risk is determined according to the iteration completion rate, if the iteration completion rate is less than a preset iteration completion rate threshold, it is determined that there is a risk, otherwise it is determined that there is no risk.

[0065] Figure 3 The schematic diagram of the code quality level provided for the embodiments of the present application is shown in FIG. 2. Figure 3As shown, the code quality level can be determined by the code repetition degree, the code maintainability level, the code reliability level and the code security level. Specifically, if the code repetition degree is less than or equal to 30%, and the code maintainability level, the code reliability level and the code security level are all A level, the code quality level is determined as A level; if the code repetition degree is less than or equal to 40%, and the code maintainability level, the code reliability level and the code security level are all B level, the code quality level is determined as B level; if the code repetition degree is less than or equal to 50%, and the code maintainability level, the code reliability level and the code security level are all C level, the code quality level is determined as C level; if the code repetition degree is less than or equal to 70%, and the code maintainability level, the code reliability level and the code security level are all D level, the code quality level is determined as D level; if the code repetition degree is greater than 70%, and the code maintainability level, the code reliability level and the code security level are all E level, the code quality level is determined as E level.

[0066] S102: Determine the first feature vector according to the basic data, the first value corresponding to each sub-basic data pre-stored and the first position corresponding to each first value in the first feature vector.

[0067] In the embodiment of the present application, after the electronic device determines the R&D information of the target project, the electronic device determines the first feature vector according to the basic data in the R&D information, the first value corresponding to each sub-basic data pre-stored and the first position corresponding to each first value in the first feature vector. Wherein, because the basic data includes team size and development language, the team size is one sub-basic data of the basic data, and the development language is another sub-basic data of the basic data.

[0068] In the embodiment of the present application, the electronic device pre-stores the first value corresponding to each sub-basic data, wherein the first value corresponding to different types of sub-basic data can be the same.

[0069] S103: Determine the second feature vector according to the process data, the second value corresponding to each sub-process data pre-stored and the second position corresponding to each second value in the second feature vector.

[0070] In the embodiment of the present application, after the electronic device determines the research and development information of the target project, the electronic device determines a second feature vector according to process data in the research and development information, a second value corresponding to each sub-process data pre-stored, and a second position corresponding to each second value in the second feature vector. Wherein, because the process data includes defect density, iteration completion rate, average defect repair time, and build success rate, the defect density is a sub-process data of the process data, the iteration completion rate is a sub-process data of the process data, the average defect repair time is a sub-process data of the process data, and the build success rate is a sub-process data of the process data.

[0071] S104: input the first feature vector and the second feature vector into the trained model, and obtain a risk probability of next research and development of the target project output by the model.

[0072] In the embodiment of the present application, after the electronic device determines the first feature vector and the second feature vector, the electronic device inputs the first feature vector and the second feature vector into the trained model. The model can output a risk probability of next research and development of the target project according to the first feature vector and the second feature vector. The first feature vector and the second feature vector can be input respectively, or the first feature vector and the second feature vector can be spliced to obtain a third feature vector, and the third feature vector is input into the trained model.

[0073] In the embodiment of the present application, the electronic device generates the first feature vector and the second feature vector based on the research and development information in the previous research and development process of the target project, and predicts whether a risk will occur in the next research and development process according to the first feature vector and the second feature vector, thereby improving the efficiency and accuracy of prediction.

[0074] Embodiment 2:

[0075] In order to determine the first feature vector so that the electronic device can make prediction according to the first feature vector and improve the efficiency and accuracy of prediction, on the basis of the above-mentioned embodiment, in the embodiment of the present application, the determination of the first feature vector according to the basic data, the first value corresponding to each sub-basic data pre-stored, and the first position corresponding to each first value in the first feature vector comprises:

[0076] determining each target sub-basic data contained in the basic data;

[0077] For each target sub-basic data, a target first value corresponding to the target sub-basic data is determined according to a pre-stored first value corresponding to each sub-basic data; a target first position corresponding to the target sub-basic data is determined according to a pre-stored first position corresponding to each sub-basic data in the first feature vector; and the target first value is written into the target first position.

[0078] In the embodiments of the present application, the electronic device stores a first value corresponding to each sub-basic data and a first position corresponding to each sub-basic data in the first feature vector. The first values corresponding to sub-basic data of different types can be the same or different. The first values corresponding to sub-basic data of the same type are different. The first positions corresponding to sub-basic data of the same type are the same.

[0079] In the embodiments of the present application, before determining the first feature vector, the electronic device acquires each target sub-basic data included in the basic data. For each target sub-basic data, the electronic device searches for a target first value corresponding to the target sub-basic data in pre-stored first values corresponding to each sub-basic data. Meanwhile, the electronic device also searches for a target first position corresponding to the target sub-basic data in pre-stored first positions corresponding to each sub-basic data.

[0080] The electronic device writes the target first value corresponding to each target sub-basic data into the corresponding target first position to obtain the first feature vector.

[0081] For example, the basic data includes two sub-basic data, i.e., team size and development language, where the team size includes a size of less than 5 people, and the development language is Python. The electronic device stores a first value corresponding to a size of less than 5 people, i.e., 001, and a first position corresponding to the size of less than 5 people, i.e., the first right position. The electronic device also stores a first value corresponding to Python, i.e., 003, and a first position corresponding to Python, i.e., the second right position. The electronic device determines the first feature vector as {003, 001}.

[0082] Embodiment 3

[0083] To determine the second feature vector, so that the electronic device can make a prediction according to the second feature vector, and improve the efficiency and accuracy of the prediction, on the basis of the above embodiments, in the embodiments of the present application, the determining the second feature vector according to the process data, the pre-stored second value corresponding to each sub-process data, and the second position includes:

[0084] Each target sub-process data included in the process data is determined.

[0085] For each target sub-process data, a target second value corresponding to the target sub-process data is determined according to a second value corresponding to each sub-process data pre-stored; a target second position corresponding to the target sub-process data is determined according to a second position corresponding to each sub-process data in the second feature vector in the pre-stored; and the target second value is written into the target second position.

[0086] In the embodiments of the present application, the electronic device stores a first value corresponding to each sub-process data and a second position corresponding to each sub-process data in the second feature vector, wherein the second values corresponding to the sub-process data of different types may be the same or different, the second values corresponding to the sub-process data of the same type are different, and the second positions corresponding to the sub-process data of the same type are the same position.

[0087] In the embodiments of the present application, the electronic device obtains each target sub-process data contained in the process data. For each target sub-process data, the electronic device searches for a target second value corresponding to the target sub-process data in the second value corresponding to each sub-process data pre-stored. Meanwhile, the electronic device also searches for a target second position corresponding to the target sub-process data in the second position corresponding to each sub-process data pre-stored.

[0088] The electronic device writes the target second value corresponding to each target sub-process data into the target second position to obtain the second feature vector.

[0089] For example, the process data includes four sub-process data, i.e., defect density, iteration completion rate, average defect repair time, and build success rate, wherein the defect density is 0.3, the iteration completion rate is 50%, the average defect repair time is 3 hours, and the build success rate is 60%. The electronic device stores that the second value corresponding to the defect density of 0.3 is 0030 and the second position is the first right number, the second value corresponding to the iteration completion rate of 50% is 0050 and the second position is the second right number, the second value corresponding to the average defect repair time of 3 hours is 0003 and the second position is the third right number, and the second value corresponding to the build success rate of 60% is 0060 and the second position is the fourth right number. Then, the electronic device determines that the second feature vector is {0060, 0003, 0050, 0030}.

[0090] Embodiment 4

[0091] In order to better predict risks based on R&D information, on the basis of the above embodiments, in the embodiments of the present application, the R&D information of the target project to be pre-warned is obtained, including:

[0092] The candidate R&D information corresponding to each iteration process of the target project is obtained.

[0093] determine the candidate R&D information corresponding to the latest iteration as the R&D information of the target project.

[0094] In the embodiments of the present application, the electronic device stores candidate R&D information corresponding to each iteration process of a target project, and the electronic device can determine the candidate R&D information corresponding to the latest iteration as the R&D information of the target project when determining the R&D information of the target project.

[0095] To better predict risks based on R&D information, on the basis of the above embodiments, in the embodiments of the present application, the R&D information of the target project to be warned is obtained as follows:

[0096] obtain the stored candidate R&D information corresponding to each iteration process of the target project;

[0097] determine each candidate R&D information corresponding to the preset number of iterations closest to the current time as the R&D information, respectively;

[0098] Before inputting the first feature vector and the second feature vector into the trained model, the method further includes:

[0099] According to each first feature vector corresponding to each R&D information, a first mean value of each first value corresponding to each first position is calculated, a first mean value vector corresponding to each first feature vector is determined, and the first mean value vector is determined as the first feature vector input into the model;

[0100] According to each second feature vector corresponding to each R&D information, a second mean value of each second value corresponding to each second position is calculated, a second mean value vector corresponding to each second feature vector is determined, and the second mean value vector is determined as the second feature vector input into the model.

[0101] In the embodiments of the present application, when determining the R&D information of the target project, the electronic device can also determine each candidate R&D information corresponding to the preset number of iterations closest to the current time as the R&D information, respectively. That is, the electronic device can determine a preset number of R&D information, and determine the first feature vector and the second feature vector corresponding to each R&D information.

[0102] In the embodiments of the present application, for each first position in the first feature vector, the electronic device calculates a first mean value of the first value corresponding to the first position. The electronic device determines a first mean value vector corresponding to each first feature vector according to each first position and the first mean value corresponding to each first position, and determines the first mean value vector as the first feature vector input into the model.

[0103] In the embodiments of the present application, the electronic device calculates, for each second position in the second feature vector, a second mean value of the second numerical value corresponding to the second position. The electronic device determines, according to each second position and the second mean value corresponding to each second position, a second mean value vector corresponding to each second feature vector, and determines the second mean value vector as the second feature vector input into the model.

[0104] Generally, in the embodiments of the present application, the preset number of iterations is three.

[0105] Embodiment 5

[0106] In order to improve the efficiency and accuracy of risk prediction, on the basis of the above embodiments, in the embodiments of the present application, the model is a Wide&Deep model.

[0107] Inputting the first feature vector and the second feature vector into the trained model includes:

[0108] Inputting the first feature vector into a Wide sub-model of the Wide&Deep model.

[0109] Splicing the first feature vector and the second feature vector according to a preset order to obtain a third feature vector, and inputting the third feature vector into a Deep sub-model of the Wide&Deep model.

[0110] In order to better predict the risk in the research and development process, the model for predicting the risk in the embodiments of the present application can be a Wide&Deep model, wherein the Wide&Deep model combines the memorization of a linear model and the generalization of a Deep Neural Networks (DNN) model. Specifically, in the embodiments of the present application, the Wide&Deep model is composed of a Wide sub-model and a Deep sub-model, wherein the Wide sub-model is a linear model, and the Deep sub-model is a DNN model.

[0111] Specifically, in the embodiments of the present application, the electronic device inputs the first feature vector into the Wide sub-model, the Cross Product Transformation structure inside the Wide sub-model processes the first feature vector to obtain a combined feature, and predicts the risk based on the combined feature to obtain a first prediction result.

[0112] The electronic device concatenates the first feature vector and the second feature vector in a preset order to obtain a third feature vector, and inputs the third feature vector into a Deep model, in which a low-dimensional real vector corresponding to each position in the third feature vector is pre-stored, and the real vector is referred to as an embedding of the feature vector. The Deep model can make a prediction according to the numerical value of each position in the third feature vector and the real vector corresponding to each position to obtain a second prediction result.

[0113] In the embodiment of the present application, the Wide&Deep model connects the output of the Wide sub-model with the output of the Deep sub-model composed of the Embedding layer and the multiple hidden layers, and inputs the outputs of the two sub-models into a final output layer, and then outputs the risk probability of the next research and development of the target item by the final output layer. The Wide sub-model is good at processing a large number of sparse feature vectors, and the Deep sub-model is characterized by strong expression ability of neural networks and deep feature cross of the third feature vector.

[0114] In the embodiment of the present application, when constructing the training sample set of the Wide&Deep model, the first sample research and development information corresponding to a preset number of iterations except the latest iteration and the second sample research and development information of the latest iteration are obtained for each sample item. For each sample item, the sample first feature vector and the sample second feature vector corresponding to each first sample research and development information are determined, and the sample first mean vector corresponding to each sample first feature vector and the sample second mean vector corresponding to each sample second feature vector are determined, and whether a risk occurs in the research and development process is determined according to the second sample research and development information, and the sample risk probability corresponding to the second sample research and development information is determined according to the pre-stored corresponding relationship between whether a risk occurs and the sample risk probability. In the embodiment of the present application, the sample first mean vector, the sample second mean vector and the sample risk probability are a group of sample data in the training sample set.

[0115] When training the Wide&Deep model, the electronic device obtains a group of sample data in the training sample set, and inputs the sample data into the Wide&Deep model to be trained. The Wide&Deep model makes a risk prediction according to the first mean vector and the second mean vector in the sample data, and outputs a predicted risk probability of the next research and development. The electronic device calculates a loss value of the Wide&Deep model according to the predicted risk probability and the sample risk probability carried in the sample data, and adjusts the internal parameters of the Wide&Deep model to be trained according to the loss value.

[0116] In the embodiments of the present application, the electronic device collects, analyzes, extracts features and constructs a feature vector for the research and development information of the research and development process of a large number of projects based on deep learning technology and relying on data analysis and AI technology, and constructs and trains a model according to the constructed feature vector. When there is a new target project, the electronic device constructs a first feature vector and a second feature vector corresponding to the research and development information of the research and development process by collecting the research and development information of the research and development process in real time, and predicts the probability of risk occurrence of the target project in the next iteration and gives an alarm using the model according to the first feature vector and the second feature vector.

[0117] Embodiment 6:

[0118] Figure 4 A schematic structural diagram of a pre-warning device provided in the embodiments of the present application, the device comprising:

[0119] The acquisition module 401 is configured to acquire research and development information of a target project to be pre-warned, wherein the research and development information comprises basic data and process data of a research and development process, wherein the basic data at least comprises team size and development language, and the process data at least comprises defect density, iteration completion rate, defect average repair time and build success rate in the research and development process.

[0120] The processing module 402 is configured to determine a first feature vector according to the basic data, a pre-stored first value corresponding to each sub-basic data and a first position corresponding to each first value in the first feature vector, and determine a second feature vector according to the process data, a pre-stored second value corresponding to each sub-process data and a second position corresponding to each second value in the second feature vector.

[0121] The prediction module 403 is configured to input the first feature vector and the second feature vector into a trained model to obtain a risk probability of next research and development of the target project output by the model.

[0122] In a possible implementation, the processing module 402 is specifically configured to determine each target sub-basic data contained in the basic data, for each target sub-basic data, determine a target first value corresponding to the target sub-basic data according to a pre-stored first value corresponding to each sub-basic data, determine a target first position corresponding to the target sub-basic data according to a pre-stored first position corresponding to each sub-basic data in the first feature vector, and write the target first value to the target first position.

[0123] In a possible implementation, the processing module 402 is specifically configured to determine each target sub-process data included in the process data; for each target sub-process data, determine a target second value corresponding to the target sub-process data according to a second value corresponding to each sub-process data pre-stored; determine a target second position corresponding to the target sub-process data according to a second position corresponding to each sub-process data in the second feature vector; and write the target second value into the target second position.

[0124] In a possible implementation, the obtaining module 401 is specifically configured to obtain candidate R&D information corresponding to each iteration process of the target project pre-stored; and determine candidate R&D information corresponding to the latest iteration as the R&D information of the target project.

[0125] In a possible implementation, the obtaining module 401 is specifically configured to obtain candidate R&D information corresponding to each iteration process of the target project pre-stored; and determine each candidate R&D information corresponding to a preset number of iterations closest to the current time as the R&D information.

[0126] The processing module is further configured to calculate a first mean value of the first value corresponding to each first position according to each first feature vector corresponding to each R&D information, determine a first mean value vector corresponding to each first feature vector, and determine the first mean value vector as the first feature vector input into the model; calculate a second mean value of the second value corresponding to each second position according to each second feature vector corresponding to each R&D information, determine a second mean value vector corresponding to each second feature vector, and determine the second mean value vector as the second feature vector input into the model.

[0127] In a possible implementation, the model is a Wide&Deep model.

[0128] The prediction module 403 is specifically configured to input the first feature vector into a Wide sub-model of the Wide&Deep model; splice the first feature vector and the second feature vector according to a preset order to obtain a third feature vector, and input the third feature vector into a Deep sub-model of the Wide&Deep model.

[0129] Embodiment 7

[0130] On the basis of the above embodiments, the embodiments of the present application further provide an electronic device, Figure 5 A structural schematic diagram of an electronic device provided by the embodiments of the present application is as follows, Figure 5As shown, the device comprises a processor 51, a communication interface 52, a memory 53 and a communication bus 54, wherein the processor 51, the communication interface 52 and the memory 53 communicate with each other through the communication bus 54;

[0131] The memory 53 stores a computer program, which, when executed by the processor 51, causes the processor 51 to perform the following steps:

[0132] Obtain the R&D information of the target project to be warned, wherein the R&D information comprises basic data and process data of the R&D process, wherein the basic data at least comprises team size and development language, and the process data at least comprises defect density, iteration completion rate, average defect repair time and build success rate in the R&D process;

[0133] Determine the first feature vector according to the basic data, the first value corresponding to each sub-basic data and the first position corresponding to each first value in the first feature vector;

[0134] Determine the second feature vector according to the process data, the second value corresponding to each sub-process data and the second position corresponding to each second value in the second feature vector;

[0135] Input the first feature vector and the second feature vector into the trained model to obtain the risk probability of the next R&D of the target project output by the model.

[0136] In a possible implementation, the processor is further configured to:

[0137] Determine each target sub-basic data contained in the basic data;

[0138] For each target sub-basic data, determine a target first value corresponding to the target sub-basic data according to the first value corresponding to each sub-basic data, and determine a target first position corresponding to the target sub-basic data according to the first position corresponding to each sub-basic data in the first feature vector; and write the target first value into the target first position.

[0139] In a possible implementation, the processor is further configured to:

[0140] Determine each target sub-process data contained in the process data;

[0141] For each target sub-process data, a target second value corresponding to the target sub-process data is determined according to a second value corresponding to each sub-process data pre-stored; a target second position corresponding to the target sub-process data is determined according to a second position corresponding to each sub-process data in the second feature vector pre-stored; and the target second value is written into the target second position.

[0142] In a possible implementation, the processor is further configured to:

[0143] obtain candidate R&D information corresponding to each iteration process of the target project pre-stored;

[0144] determine the candidate R&D information corresponding to the latest iteration as the R&D information of the target project.

[0145] In a possible implementation, the processor is further configured to:

[0146] obtain candidate R&D information corresponding to each iteration process of the target project pre-stored;

[0147] determine each candidate R&D information corresponding to a preset number of iterations closest to the current time as the R&D information, respectively;

[0148] The processor is further configured to:

[0149] calculate a first mean value of each first value corresponding to each first position according to each first feature vector corresponding to each R&D information, determine a first mean value vector corresponding to each first feature vector, and determine the first mean value vector as a first feature vector input into the model;

[0150] calculate a second mean value of each second value corresponding to each second position according to each second feature vector corresponding to each R&D information, determine a second mean value vector corresponding to each second feature vector, and determine the second mean value vector as a second feature vector input into the model.

[0151] In a possible implementation, the model is a Wide&Deep model;

[0152] The processor is further configured to:

[0153] inputting the first feature vector and the second feature vector into the trained model includes:

[0154] inputting the first feature vector into a Wide sub-model of the Wide&Deep model;

[0155] The first feature vector and the second feature vector are spliced in a preset order to obtain a third feature vector, and the third feature vector is input into a Deep sub-model of the Wide & Deep model.

[0156] Since the principle of solving the problem of the electronic device and the early warning method are similar, the implementation of the electronic device can refer to the implementation of the method, and the repeated parts will not be repeated.

[0157] The communication bus mentioned in the electronic device can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. The communication bus can be divided into an address bus, a data bus, a control bus, etc. For the convenience of representation, only one thick line is used in the figure, but it does not mean that there is only one bus or one type of bus. The communication interface 52 is used for communication between the electronic device and other devices. The memory can include a Random Access Memory (RAM) and can also include a Non-Volatile Memory (NVM), such as at least one disk memory. Optionally, the memory can also be at least one storage device located away from the aforementioned processor.

[0158] The processor mentioned above can be a general-purpose processor, including a central processing unit, a network processor (NP), etc.; it can also be a Digital Signal Processing (DSP), an application-specific integrated circuit, a field programmable gate array or other programmable logic device, a discrete gate or transistor logic device, a discrete hardware component, etc.

[0159] Embodiment 8:

[0160] On the basis of the above-mentioned embodiments, the embodiment of the present application further provides a computer readable storage medium, and the computer readable storage medium stores a computer program executable by a processor. When the program runs on the processor, the processor executes the following steps:

[0161] Obtain the research and development information of the target project to be warned, wherein the research and development information includes basic data and process data of the research and development process, wherein the basic data at least includes team size and development language, and the process data at least includes defect density, iteration completion rate, average defect repair time, and build success rate in the research and development process;

[0162] determine the first feature vector according to the basic data, the first values corresponding to each sub-basic data pre-stored, and the first positions corresponding to each first value in the first feature vector;

[0163] determine the second feature vector according to the process data, the second values corresponding to each sub-process data pre-stored, and the second positions corresponding to each second value in the second feature vector;

[0164] input the first feature vector and the second feature vector into the trained model, and obtain the risk probability of the next research and development of the target project output by the model.

[0165] In a possible implementation, the determining the first feature vector according to the basic data, the first values corresponding to each sub-basic data pre-stored, and the first positions corresponding to each first value in the first feature vector includes:

[0166] determine each target sub-basic data included in the basic data;

[0167] for each target sub-basic data, determine a target first value corresponding to the target sub-basic data according to the first values corresponding to each sub-basic data pre-stored, determine a target first position corresponding to the target sub-basic data according to the first positions corresponding to each sub-basic data in the first feature vector pre-stored, and write the target first value into the target first position.

[0168] In a possible implementation, the determining the second feature vector according to the process data, the second values corresponding to each sub-process data pre-stored, and the second positions includes:

[0169] determine each target sub-process data included in the process data;

[0170] for each target sub-process data, determine a target second value corresponding to the target sub-process data according to the second values corresponding to each sub-process data pre-stored, determine a target second position corresponding to the target sub-process data according to the second positions corresponding to each sub-process data in the second feature vector pre-stored, and write the target second value into the target second position.

[0171] In a possible implementation, the obtaining the research and development information of the target project to be prewarned includes:

[0172] obtain the candidate research and development information corresponding to each iteration process of the target project pre-stored;

[0173] determine the candidate research and development information corresponding to the latest iteration as the research and development information of the target project.

[0174] In a possible implementation, the obtaining the R&D information of the target project to be warned includes:

[0175] obtaining candidate R&D information corresponding to each iteration process of the target project;

[0176] determining each candidate R&D information corresponding to a preset number of iterations closest to the current time as the R&D information respectively;

[0177] Before the first feature vector and the second feature vector are input into the trained model, the method further includes:

[0178] calculating a first mean value of the first numerical value corresponding to each first position according to each first feature vector corresponding to each R&D information, determining a first mean value vector corresponding to the first feature vector, and determining the first mean value vector as the first feature vector input into the model;

[0179] calculating a second mean value of the second numerical value corresponding to each second position according to each second feature vector corresponding to each R&D information, determining a second mean value vector corresponding to the second feature vector, and determining the second mean value vector as the second feature vector input into the model.

[0180] In a possible implementation, the model is a Wide&Deep model.

[0181] inputting the first feature vector and the second feature vector into the trained model includes:

[0182] inputting the first feature vector into a Wide sub-model of the Wide&Deep model;

[0183] splicing the first feature vector and the second feature vector according to a preset order to obtain a third feature vector, and inputting the third feature vector into a Deep sub-model of the Wide&Deep model.

[0184] Since the principle of solving the problem by the above computer readable storage medium is similar to the warning method, the implementation of the above computer readable storage medium can refer to the embodiments of the method, and the repeated parts will not be described herein.

[0185] Embodiment 9

[0186] The application also provides a computer program product, which, when executed by a computer, implements the warning method described in any of the method embodiments applied to the electronic device.

[0187] In the embodiments described above, all or some of the embodiments can be implemented by hardware, software, firmware or any combination of them. All or some of the embodiments can be implemented as computer program products. The computer program products include one or more computer program instructions. When the computer program instructions are loaded into and executed by a computer, all or some of the embodiments as described in the specification are all or partially produced.

[0188] Those skilled in the art should understand that the embodiments of the present application can be provided as methods, systems or computer program products. Therefore, the present application can take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to magnetic disks, CD-ROMs, optical storage media, etc.) having computer usable program code embodied thereon.

[0189] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems) and computer program products according to the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device produce a device that implements the functions specified in the flowcharts and / or block diagrams. Figure 1 one or more flows and / or blocks Figure 1 means for performing the functions specified in the flowchart

[0190] These computer program instructions can also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce a manufactured product including instruction means that implement the functions specified in the flowcharts and / or block diagrams. Figure 1 one or more flows and / or blocks Figure 1 means for performing the functions specified in the flowchart

[0191] These computer program instructions can also be loaded onto a computer or other programmable data processing device to cause a series of operational steps to be performed on the computer or other programmable device to produce a computer implemented process, so that the instructions executed on the computer or other programmable device provide a process for implementing the functions specified in the flowcharts and / or block diagrams. Figure 1 one or more flows and / or blocks Figure 1 means for performing the functions specified in the flowchart

[0192] Obviously, many modifications and variations of the present application are possible in light of the above teachings. It is, therefore, to be understood that within the scope of the appended claims and their equivalents, the application can be practiced otherwise than as specifically described.

Claims

1. A method of early warning, characterized in that, The method comprises: obtaining research and development information of a target project to be warned, wherein the research and development information comprises basic data and process data of a research and development process, wherein the basic data at least comprises team size and development language, and the process data at least comprises defect density, iteration completion rate, defect average repair time length, and build success rate in the research and development process; wherein the basic data further comprises at least one of the following data: an industry to which the target project belongs, a technical field to which the target project belongs, a ratio of development and test personnel, and an iteration cycle; and the process data further comprises at least one of the following data: code submission frequency, code merging frequency, build and deployment frequency, build and deployment time length, code quality level, unit test coverage, and online problem quantity; determining a first feature vector according to the basic data, a first value corresponding to each sub-basic data and a first position corresponding to each first value in the first feature vector; determining a second feature vector according to the process data, a second value corresponding to each sub-process data and a second position corresponding to each second value in the second feature vector; inputting the first feature vector and the second feature vector into a trained model to obtain a risk probability of next research and development of the target project output by the model; wherein the model is a Wide&Deep model comprising a Wide sub-model, a Deep sub-model, and an output layer; the inputting the first feature vector and the second feature vector into the trained model to obtain the risk probability of next research and development of the target project output by the model comprises: inputting the first feature vector into the Wide sub-model, processing the first feature vector by a cross product transformation structure in the Wide sub-model to obtain combined features, and predicting a risk based on the combined features to obtain a first prediction result; splicing the first feature vector and the second feature vector according to a preset order to obtain a third feature vector, and inputting the third feature vector into the Deep sub-model; wherein the Deep sub-model has pre-stored low-dimensional real number vectors corresponding to each position in the third feature vector, and the Deep sub-model predicts according to a value of each position in the third feature vector and the real number vector corresponding to each position to obtain a second prediction result; connecting the first prediction result and the second prediction result to input into the output layer, and outputting the risk probability by the output layer; wherein the determination process of the process data comprises: Collecting original process data, the original process data comprising: code commit times, code committer numbers, code merge times, unit test coverage, unit test pass rates, build deployment times, build deployment success times, build deployment time lengths, iteration planning user story numbers, iteration actually completed user story numbers, iteration inner user story numbers, iteration inner defect numbers, defect severity, defect repair time lengths, code duplication degrees, code maintainability levels, code reliability levels, code security levels, online problem numbers; According to the original process data, determining the process data; Wherein, according to the original process data, determining the process data comprises: The ratio of the iteration inner defect number to the iteration inner user story number is determined as the defect density; The ratio of the iteration actually completed story number to the iteration planning user story number is determined as the iteration completion rate; The ratio of the build deployment success number to the build deployment number is determined as the build success rate; According to the code commit times and the code committer numbers, the daily per person code commit times are determined as the code commit frequency; The ratio of the iteration inner code merge times to the iteration inner user story number is determined as the code merge frequency; According to the code duplication degree, the code maintainability level, the code reliability level, the code security level and the pre-defined rule, the code quality level is determined.

2. The method of claim 1, wherein, According to the basis data, the pre-saved first value corresponding to each sub basis data and the first position corresponding to each first value in the first feature vector, the first feature vector is determined. Determine each target sub basis data contained in the basis data; For each target sub basis data, according to the pre-saved first value corresponding to each sub basis data, the target first value corresponding to the target sub basis data is determined; According to the pre-saved first position corresponding to each sub basis data in the first feature vector, the target first position corresponding to the target sub basis data is determined; The target first value is written into the target first position.

3. The method of claim 1, wherein, According to the process data, the pre-saved second value corresponding to each sub process data and the second position, the second feature vector is determined. Determine each target sub process data contained in the process data; For each target sub process data, according to the pre-saved second value corresponding to each sub process data, the target second value corresponding to the target sub process data is determined; According to the pre-saved second position corresponding to each sub process data in the second feature vector, the target second position corresponding to the target sub process data is determined; The target second value is written into the target second position.

4. The method of claim 1, wherein, The research and development information of the target project to be warned comprises: Obtaining the candidate research and development information corresponding to each iteration process of the target project saved; The candidate research and development information corresponding to the latest iteration is determined as the research and development information of the target project.

5. The method of claim 1, wherein, The research and development information of the target project to be warned comprises: obtain candidate R&D information corresponding to each iteration process of the target project saved; determine each candidate R&D information corresponding to a preset number of iterations closest to the current time as the R&D information respectively; Before the first feature vector and the second feature vector are input into the trained model, the method further comprises: determine a first mean value vector corresponding to each first feature vector according to a first mean value of each first position corresponding to each first feature vector, and determine the first mean value vector as the first feature vector input into the model; determine a second mean value vector corresponding to each second feature vector according to a second mean value of each second position corresponding to each second feature vector, and determine the second mean value vector as the second feature vector input into the model.

6. A warning device, characterized in that The device comprises: An acquisition module is configured to acquire R&D information of a target project to be warned, wherein the R&D information comprises basic data and process data of a R&D process, wherein the basic data at least comprises team size and development language, and the process data at least comprises defect density, iteration completion rate, defect average repair time length, and build success rate in the R&D process; wherein the basic data further comprises at least one of the following data: industry to which the target project belongs, technical field to which the target project belongs, ratio of development and testing personnel, and iteration cycle; and the process data further comprises at least one of the following data: code submission frequency, code merging frequency, build and deployment frequency, build and deployment time length, code quality level, unit test coverage, and online problem number; A processing module is configured to determine a first feature vector according to the basic data, a first value corresponding to each sub-basic data pre-stored, and a first position corresponding to each first value in the first feature vector; and determine a second feature vector according to the process data, a second value corresponding to each sub-process data pre-stored, and a second position corresponding to each second value in the second feature vector; A prediction module is configured to input the first feature vector and the second feature vector into a trained model, and obtain a risk probability of next R&D of the target project output by the model; The model is a Wide&Deep model, comprising a Wide sub-model, a Deep sub-model, and an output layer. The prediction module is specifically configured to input the first feature vector into the Wide sub-model, perform processing on the first feature vector by a dot product transformation structure inside the Wide sub-model to obtain combined features, and perform prediction on a risk based on the combined features to obtain a first prediction result; splice the first feature vector and the second feature vector according to a preset order to obtain a third feature vector, and input the third feature vector into the Deep sub-model; wherein, the Deep sub-model has pre-stored low-dimensional real vectors corresponding to each position in the third feature vector, and the Deep sub-model performs prediction according to a numerical value of each position in the third feature vector and the real vector corresponding to each position to obtain a second prediction result; the first prediction result and the second prediction result are connected and input into the output layer, and the output layer outputs the risk probability. The determination process of the process data includes: Collecting original process data, the original process data including: code commit times, code committer number, code merge times, unit test coverage, unit test pass rate, build and deployment times, build and deployment success times, build and deployment duration, iteration planning user story quantity, iteration actual completed user story quantity, iteration internal user story quantity, iteration internal defect quantity, defect severity, defect repair duration, code duplication degree, code maintainability level, code reliability level, code security level, and online problem quantity; Determining the process data according to the original process data; The determination of the process data according to the original process data includes: Determining the ratio of the iteration internal defect quantity to the iteration internal user story quantity as the defect density, determining the ratio of the iteration actual completed story quantity to the iteration planning user story quantity as the iteration completion rate, determining the ratio of the build and deployment success times to the build and deployment times as the build success rate, determining the daily code commit times per person as the code commit frequency according to the code commit times and the code committer number, determining the ratio of the iteration internal code merge times to the iteration internal user story quantity as the code merge frequency, and determining the code quality level according to the code duplication degree, the code maintainability level, the code reliability level, the code security level, and a pre-defined rule.

7. The apparatus of claim 6, wherein, The processing module is specifically configured to determine each target sub-basic data included in the basic data, for each target sub-basic data, determine a target first value corresponding to the target sub-basic data according to a pre-stored first value corresponding to each sub-basic data, determine a target first position corresponding to the target sub-basic data according to a pre-stored first position of each sub-basic data in the first feature vector, and write the target first value into the target first position.

8. The apparatus of claim 6, wherein, The processing module is specifically used to determine each target subprocess data contained in the process data; for each target subprocess data, determine the target second value corresponding to the target subprocess data according to the second value corresponding to each pre-saved subprocess data; determine the target second position corresponding to the target subprocess data according to the second position corresponding to each pre-saved subprocess data in the second feature vector; and write the target second value into the target second position.

9. The apparatus of claim 6, wherein, The acquisition module is specifically used to acquire the candidate R&D information corresponding to each iteration of the target project; and to determine the candidate R&D information corresponding to the latest iteration as the R&D information of the target project.

10. The apparatus of claim 6, wherein, The acquisition module is specifically used to acquire the candidate R&D information corresponding to each iteration process of the target project; and to determine each candidate R&D information corresponding to the preset number of iterations closest to the current time as the R&D information. The processing module is further configured to: calculate the first mean of the first value corresponding to each first position based on each first feature vector corresponding to each R&D information; determine the first mean vector corresponding to each first feature vector; and determine the first mean vector as the first feature vector input into the model; calculate the second mean of the second value corresponding to each second position based on each second feature vector corresponding to each R&D information; determine the second mean vector corresponding to each second feature vector; and determine the second mean vector as the second feature vector input into the model.

11. An electronic device, comprising: The electronic device includes a processor that executes a computer program stored in a memory to implement the steps of the early warning method as described in any one of claims 1-5.

12. A computer-readable storage medium, characterized in that, It stores a computer program that, when executed by a processor, implements the steps of the early warning method as described in any one of claims 1-5.

13. A computer program product, characterised in that, The computer program product includes: computer program code, which, when run on a computer, causes the computer to perform the steps of the early warning method as described in any one of claims 1-5.

Images