CN103208039A - Method and device for evaluating software project risks - Google Patents
Method and device for evaluating software project risks Download PDFInfo
- Publication number
- CN103208039A CN103208039A CN2012100097525A CN201210009752A CN103208039A CN 103208039 A CN103208039 A CN 103208039A CN 2012100097525 A CN2012100097525 A CN 2012100097525A CN 201210009752 A CN201210009752 A CN 201210009752A CN 103208039 A CN103208039 A CN 103208039A
- Authority
- CN
- China
- Prior art keywords
- risk
- project
- mentioned
- evaluation model
- probability
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention aims at providing a method for evaluating software project risks. According to the method, a software project risk evaluation model with high risk occurrence probability judgment accuracy can be established. The method includes collection: collecting historical data about project indexes and project risks of a plurality of samples; segmentation: segmenting a range mapped by the project indexes into a plurality of sub-ranges and getting a representative value aiming at each sub-range, wherein the range is an m-dimensional vector space; statistics: conducting statistics ratio of the number of project index mapping points occurring with one project risk in each sub-range and the number of all project index mapping points mapped in the sub-range and utilizing the ratio as the occurrence probability of the project risk; and calculation: conducting normalization processing according to the representative value and the risk probability obtained by statistics and utilizing the processed data as training data to manufacture a project risk evaluation model which is used for risk evaluation of a new project.
Description
Technical field
The present invention relates to a kind of software project risk evaluating method and software project risk assessment device, be particularly related to by the quantitative relationship between analysis project index and the project risk, set up risk evaluation model, model be used for is estimated the various risk probability of happening of new projects, and the method and apparatus that model is optimized.
Background technology
In various software project performance historyes through regular meeting various risks appear.Such as, in some project development process, there are situation that main developer leaves office, repeated code error or situation such as demand change frequently frequently.More than the possibility occurrence of these situations that project exploitation is exerted an influence claim project risk.Project risk in the performance history can produce project impacts, and causes some negative effects.Some project risk makes the project schedule postpone, and some project risk increases the project cost of development, and some project risk has reduced software quality.Therefore, hope can be prevented in order to take measures in early days at the probability of happening of each each project risk of project forecast.
Adopting risk evaluation model to come the project risk probability of occurrence is estimated is the effective ways of identification project potential risk in this area.In fact, the availability risk appraisement system for example comprises a plurality of targets, the evaluation of risk probability of happening, the evaluation of the risk order of severity, the prediction of risk time of occurrence, the evaluation of venture influence scope etc.The present invention handles the evaluation of risk probability of happening emphatically.
Existing project risk evaluation method has following several.
For example exist the mode that adopts questionnaire risk probability to be carried out the method for simple evaluation.At non-patent literature 1:(paper) " Software Risk Management:Principles and Practices ", Barry Boehm, IEEE Software has told about a kind of method of carrying out risk assessment with survey in 1991..In the method, by the questionnaire of expert's design specialized, be used for collecting some relevant project information in danger of following the wind.By the survey that project administrator is carried out, the collection management person is to the view scoring of item status.Based on the scoring of collecting, carry out certain analysis-by-synthesis (such as AHP or DELPHI method), draw the probability that risk takes place.After the achievement of this paper author Barry Boehm was delivered, other scholars had carried out a lot of follow-up studies and perfect along same route, all were elementary tactics with the survey.
But this method has bigger uncertain and to the dependence of expert's questionnaire.Because in the process of survey, project administrator joins bias in the scoring easily, has had influence on the accuracy of project evaluation, therefore can bring bigger uncertainty.In addition, the formulation of the questionnaire expertise that places one's entire reliance upon.Must require the expert to work out corresponding questionnaire again after changing a kind of item types, adjust analytical approach.This has formed dependence to the expert, has influenced the scope of application of questionnaire.Therefore, this method exists that human factor is big, accuracy is low and the unmanageable shortcoming of cost.
In addition, the evaluation method that adopts rule-based or model is arranged also, the probability that the methods analyst risk that for example adopts the value of earning to analyze (EVM) occurs.
At patent documentation 2:(United States Patent (USP) 7669180B2) in this a kind of method is disclosed: for every group of risk factors have been created the risk assessment task, for each task definition decision flow diagram, create relevant regular collection for decision flow diagram, automatically risk has been carried out comprehensive evaluation based on project data and regular collection.
In addition, the risk model of mentioning in the patent documentation 2 is set up rule and can be described with certain machine language, and patent documentation 2 related devices provide the API of rule programming usefulness.The user can more freely edit the risk that is applicable to own software project and detect rule.
A kind of rule-based risk evaluating method has been described in the patent documentation 2.These rules can carry out freely editing with certain language or programming API, make this method have favorable expansibility.But this method exists the user to detect regular problem from edlin or risk of selection.In fact, the supvr who lacks experience knows not too which kind of rule is applicable to the risk of their project of detection, does not know how to create effective rule yet.
In addition, at non-patent literature 3:(paper) " Risk management method using data from EVM in software development projects ", A.Hayashi, CIMCA 2008, described a kind of risk evaluating method among the IEEE Computer Society..Specifically, this technology is to adopt EVM methods such as (value of earning analyses), and the various influences that project process is subjected to are quantized into the progress late days, carry out risk assessment according to the progress late days that calculates then.
Summary of the invention
The object of the present invention is to provide a kind of software project risk evaluating method and the software project risk assessment device that can set up the higher software project risk evaluation model of accuracy that the risk probability of occurrence judges.
The present invention is a kind of software project risk evaluating method, comprising: collect step, the historical data about the project indicator and project risk of collecting a plurality of samples; Segmentation step is divided into above-mentioned project indicator institute range of a mapping a plurality of, and the subvalue territory after cutting apart at each replaces tabular value, if when being m with respect to the kind quantity of the project indicator of certain intermediate item risk, above-mentioned codomain is the m gt, and wherein, m is the integer greater than 0; Statistic procedure, according to the historical data of collecting, add up the quantity of the project indicator mapping point of following certain intermediate item risk generation in each subvalue territory with respect to the ratio of the quantity that is mapped to all items index mapping point in this subvalue territory, risk probability as such project risk appearance, and calculation procedure, the above-mentioned risk probability according to after above-mentioned typical value and the statistics carries out normalized, as training data, make the project risk evaluation model.
In addition, the present invention also can be a kind of software project risk assessment device, comprising: collector unit, the historical data about the project indicator and project risk of collecting a plurality of samples; Cutting unit is divided into above-mentioned project indicator institute range of a mapping a plurality of, and the subvalue territory after cutting apart at each replaces tabular value, if when being m with respect to the kind quantity of the project indicator of certain intermediate item risk, above-mentioned codomain is the m gt, and wherein, m is the integer greater than 0; Statistic unit, according to the historical data of collecting, add up the quantity of the project indicator mapping point of following certain intermediate item risk generation in each subvalue territory with respect to the ratio of the quantity that is mapped to all items index mapping point in this subvalue territory, risk probability as such project risk appearance, and computing unit, the above-mentioned risk probability according to after above-mentioned typical value and the statistics carries out normalized, as training data, make the project risk evaluation model.
According to software project risk evaluating method of the present invention and software project risk assessment device, can be under the situation that does not rely on expertise, find out the pests occurrence rule of various risks in the historical data, and it is quantified as risk evaluation model, in new projects, estimate the probability that various risks take place based on the project indicator and evaluation model, thereby improved the accuracy of risk assessment.
In addition, the present invention has adopted the method for machine learning, finds historical risk occurrence law automatically, is equivalent to " rule " created robotization, and by collecting various types of project risks and index, has enlarged the scope of application of device.
Therefore, according to the present invention, can set up the higher software project risk evaluation model of accuracy, help the various risks in the project administrator discovery software project.Adopt modeling method of the present invention and device effectively to utilize historical data, find out the rule that project risk occurs, the risk that exists in the discovery new projects reduces the dependence to expertise, improve the accuracy rate that risk is found, and the scope of application is extensive.
Description of drawings
Fig. 1 is the process flow diagram of the software project risk evaluating method that relates to of first embodiment;
Fig. 2 is the block diagram of the software project risk assessment device that relates to of first embodiment;
Fig. 3 is the process flow diagram of the software project risk evaluating method that relates to of second embodiment;
Fig. 4 is the block diagram of the software project risk assessment device that relates to of second embodiment;
Fig. 5 is the illustration of project risk record and history item index;
Fig. 6 is the process flow diagram of preparing the risk model training data;
Fig. 7 is the illustration of explanation risk model training data;
Fig. 8 is that explanation is cut apart the illustration that obtains training data by vector space;
Fig. 9 is the process flow diagram that adopts SVM algorithm training risk model;
Figure 10 is the process flow diagram that adopts risk model computational item risk probability;
Figure 11 is the process flow diagram of the degree of correlation between explanation analysis project risk and the project indicator;
Figure 12 is the instance graph of the degree of correlation between analysis project risk and the project indicator;
Figure 13 is the process flow diagram that explanation utilizes verification msg training risk evaluation model;
Figure 14 is that the process flow diagram that vector space is cut apart distance is adjusted in explanation;
Figure 15 is the process flow diagram of adjusting degree of correlation threshold value.
Embodiment
Below, with reference to accompanying drawing, relevant preferred implementation of the present invention is described.
(first embodiment)
Fig. 2 is the block diagram of the software project risk assessment device that relates to of first embodiment.As shown in Figure 2, software project risk assessment device of the present invention comprises collector unit 200, cutting unit 201, statistic unit 203 and computing unit 204.
Wherein, collector unit 200 is preserved the history data collection about the project indicator and project risk of a plurality of samples in database 201.For example, the input that collector unit 200 accepted users are undertaken by input media, perhaps directly import historical data from external device (ED), come automatic/hand to collect historical data and the ongoing new projects data of software project, and data are saved in the database 201.
As one in the database 201 preservation example, can utilize table storage project risk record and history item index.Fig. 5 is the illustration of project risk record and history item index, schematically shows some projects risk and the project indicator in the current software administration in Fig. 5.As shown in Figure 5, preserved the project risk occurrence record in the form " project risk record ".Form the 1st row record entry title, the 2nd row records institute occurrence risk type, the 3rd row and after recorded in certain project certain risk frequency of every day.Wherein, project 1 (Project 1) " pay and postpone (Delivery delay) " risk occurred once on 2011/7/5th, and " pay and postpone " appearred in project 2 equally, and risk once on 2011/7/6th.
Form " history item index " has recorded the various project indicators of project.Form the 1st row record entry title, the collected project indicator of the 2nd row record, the 3rd row and after recorded in certain project certain index value of every day.As shown in Figure 5, form the 2nd, 3 line items " defect concentration " and " demand change scale " two indexs of project 1, the 4th, 5 line items the situation of change of these two indexs of project 2.
At this, because like terms purpose historical data has reference value more to the venture analysis of new projects, therefore suppose that collected data all belong to similar project (such as all belonging to the embedded development project), perhaps classify according to item types (as e-commerce project, mobile phone application item, middleware project etc.).
In addition, only be example as the kind of project risk pointed among Fig. 5 and the project indicator, also comprise other existing project risk and project indicators certainly.
If collector unit is 200 that collect, when being m (m is the integer greater than 0) with respect to the kind quantity of the project indicator of certain intermediate item risk, when the m kind project indicator was namely arranged, the codomain of this m kind project indicator constituted the m gt.The codomain of this m gt of conduct that the just above-mentioned project indicator of cutting unit 201 is shone upon is divided into a plurality of subvalues territory, and the subvalue territory after cutting apart at each replaces tabular value, is used for follow-up processing.
Statistic unit 203 is according to the historical data of collecting, add up the quantity of the project indicator mapping point of following certain intermediate item risk generation in each subvalue territory with respect to the ratio of the quantity that is mapped to all items index mapping point (the m dimensional vector that all items index constitutes in certain project) in this subvalue territory, as the risk probability that such project risk occurs, the details of relevant mapping is narrated in the back.
Computing unit 204 is used for making the project risk evaluation model, specifically, computing unit 204 according to above-mentioned typical value with the statistics after above-mentioned risk probability, carry out normalized, as training data, make the project risk evaluation model, according to the above-mentioned risk probability after above-mentioned typical value and the statistics, carry out normalized, as training data, make the project risk evaluation model.Wherein, above-mentioned normalized can utilize existing normalized algorithm to carry out, and, obtain after the training data, can utilize existing forecast model method for building up to set up the project risk evaluation model, therefore, omit detailed explanation at this.
Below utilize Fig. 1 that the software project risk evaluating method that first embodiment of the present invention relates to is described.Fig. 1 is the process flow diagram of the software project risk evaluating method that relates to of first embodiment.As shown in Figure 1, at first, the historical data (step 100) about the project indicator and project risk that collector unit 200 is collected a plurality of samples.
Then, cutting unit 201 is divided into above-mentioned project indicator institute range of a mapping a plurality of, subvalue territory after cutting apart at each replaces tabular value, establishes kind quantity with respect to the project indicator of certain intermediate item risk when being m, and above-mentioned codomain is m gt (step 101).
Then, statistic unit 203 is according to the historical data of collecting, add up the quantity of the project indicator mapping point of following certain intermediate item risk generation in each subvalue territory with respect to the ratio of the quantity that is mapped to all items index mapping point in this subvalue territory, as the risk probability (step 102) of such project risk appearance.
At last, computing unit 204 carries out normalized according to the above-mentioned risk probability after above-mentioned typical value and the statistics, as training data, makes project risk evaluation model (step 103).
Below describe each step for example in detail.
In general, risk evaluation model is actually the quantization function relation between " risk probability " and " relevant item index ".In order to train the project risk evaluation model, need a large amount of (risk probability, relevant item index) data.Existing history item index can directly find, but corresponding historical risk probability need be calculated in the database.In order to calculate historical risk probability of occurrence, the calculating of historical risk probability is defined as " in a plurality of sample cases under the same or similar condition, certain risk odds " here.Here adding " case under the simulated condition " is to be very difficult because find the case of a plurality of identical conditions in historical data." same or similar condition " in fact refers to " under the same or analogous condition of relevant item index ".For ease of calculating, establishing all items index here all is the relevant item index, thereby only classifies at project kind, certain the category destination data in the database can be brought in the software project risk assessment device of the present invention and handle.
Fig. 6 is the process flow diagram of preparing the risk model training data.Be illustrating the step 101 in the process flow diagram among Fig. 1～103.
As shown in Figure 6, when this risk probability of happening being defined as " under same or similar condition, (possessing the identical or close project indicator); the probability that this risk occurs " (step 800), statistic unit 203 is at certain risk i, find out the record that a plurality of project risk i take place in the database, adopt m class achievement data relevant with risk i in these projects to set up m dimensional vector X{x
1..., x
m; Because each project has many group (risk record, index) data right, can obtain a lot of groups of examples of vectorial X, and the risk record (step 801) of every group of example correspondence; Here " data to " in fact refer to situation that a certain order day risk takes place and the relevant item index on the same day, on September 1st, 2011 for example, " paying and postpone " risk in certain project occurs 1 time, the index of correlation on the same day is " defect concentration=9.2K/LOC; demand change scope=5 module ", corresponding data are to being exactly (risk record=1, indicator vector X=(9.2,5)) (step 801).
Cutting unit 201 is regarded the m dimensional vector as m-dimensional space, on each dimension, the codomain of this dimension is divided into the L section, just whole m-dimensional space is divided into Lm interval, the central point in every interval is designated as { x
1c, x
2c..., x
McAs the typical value (step 802) of each section.
Then, statistic unit 203 is mapped to all sample instance of vectorial X in the m-dimensional space; Be { x at each central point
1c, x
2c..., x
McThe interval in, count and drop on this interval sample size, and can count the sample size that risk has taken place.(risk generation sample number)/(total sample number) as this interval risk probability (step 803).By adding up above-mentioned each interval interior risk probability, can obtain one group of (risk probability, indicator vector) data set, such as (15%, X
1), (88%, X
2) ..., (75%, X
i), X here
iIt is each interval central point (step 804).
At last, computing unit 204 carries out normalization with the vectorial X value of above-mentioned data centralization, and each value of vectorial X is converted into value (step 805) between [0,1].Like this, many groups of (risk probability, the project indicator) records of risk evaluation model training have just been obtained can be used for.
In Fig. 6, the DUAL PROBLEMS OF VECTOR MAPPING that statistic unit 203 follows risk to take place earlier, cutting unit 202 carries out codomain again to be cut apart, but also can carry out codomain earlier cuts apart, and shines upon again.That is to say the embodiment that the execution sequence of each step of the present invention is not limited to exemplify among Fig. 6.Under the situation that does not have the data inheritance relation, the execution sequence of each step is arbitrarily.
In addition, in above-mentioned example, with after cutting apart the section (subvalue territory) the corresponding vector of central point as typical value, in conjunction with the risk probability in this subvalue territory, carry out normalized.But obtained typical value is not limited to central point among the present invention.It also can be the vector of other points.For example will comprise the vector of central point of special pattern (for example circular) of all mapping points in the subvalue territory as typical value.Perhaps also can calculate typical value according to the density degree that the mapping point in the subvalue territory distributes, in a word, the definition mode of typical value is not limited in the concrete example that exemplifies here, can set any vector value in the subvalue territory as typical value according to project environment or emphasis.
Fig. 7 is the illustration of explanation risk model training data.Figure 7 illustrates the example through the resulting training data of each step of Fig. 6, among the figure each line item of form certain vectorial interval central point project indicator value and should the interval in the probability of happening (risk probability, the project indicator) of certain risk.In the example of Fig. 7, the various project indicators have been carried out normalization.
The computation process of risk probability is described in conjunction with Fig. 8.Fig. 8 is that explanation is cut apart the illustration that obtains training data by vector space.For the convenience that illustrates, be that example describes with the simple two-dimensional space.In software project risk evaluating method of the present invention, mainly contain following characteristics.
(1) the m dimension project indicator that will be relevant with project risk is regarded the m gt as.In last legend, " defect concentration (Bug density) " regarded as and " pay and postpone " two kinds of project indicators that risk is relevant with " demand change scale (Require change scale) ", therefore just obtain the two dimensional vector space as figure, " defect concentration " as horizontal ordinate, " demand change scale " is as ordinate.
(2) all the index samples in the database and relevant risk record are mapped in the vector space.The index sample of risky generation represents that with filled circles the index sample that devoid of risk takes place is represented with open circles.In Fig. 8, " defect concentration " of project 1 on the 2011/7/5th in the database and " demand change scale " index is mapped to the sample point (7 in the vector space, 3), because this sample is attended by " pay and postpone " risk, therefore be labeled as filled circles, these two indexs of same 2011/7/6 day project 2 are mapped to sample point (3,4), and are filled circles.On the contrary, two indexs of 2011/7/3 day are mapped to sample point (3,1), but because the same day do not occur " pay and postpone " risk, so this sample point is open circles.
(3) according to the span of sample on each dimension, vector space is divided into many parts.In this example, transverse axis, the longitudinal axis are respectively divided 3 five equilibriums, whole two-dimensional space is divided into 3 * 3 five equilibriums, 9 intervals.About the division principle of boundary value, boundary value is included in forward than in the minizone.For example, in Fig. 8, transverse axis be divided into [0,3], (3,6], (6,9] etc. 3 intervals, the longitudinal axis be divided into [0,2], (2,4], (4,6] etc. 3 intervals.
(4) in the interval of each after division, statistics drops on the sample point sum in this interval, and the sample point number (filled circles number) of risk occurred." risk sample point number/sample point sum " namely as the risk probability in this interval, and calculates this interval central point (star-like symbol among the figure).With (risk probability, the central point vector) in this interval as one group of (risk probability, the project indicator) data.Such as, in the 2nd interval of left column, 2 sample points are arranged among Fig. 8, wherein risk has appearred in 1 sample point, and this interval risk probability is 1/2=50%.According to horizontal ordinate, can calculate this interval central point and be (1.5,3), therefore just obtaining one group (risk probability, project indicator) is recorded as [50%, (1.5,3)].Wherein the project indicator (1.5,3) is 2 dimensional vectors.In like manner, can be in other be interval other two groups of records [0%, (1.5,1)] and [100%, (7.5,3)] among the acquisition figure.
(5) gather (risk probability, the central point vector) result of calculation in all intervals, namely obtain can be used for many groups training data (risk probability, the project indicator) record of risk evaluation model training.
Told about the division of two dimensional vector space in the last example.In fact, this vector space division methods can expand to multidimensional (dimension＞=3) fully.Be that example illustrates the example that hyperspace is cut apart with four-dimensional vector (namely 4 indexs are relevant with certain risk) below.If four-dimensional vector is (x
1, x
2, x
3, x
4), vector space cut apart and interval in the risk probability calculation procedure as follows:
1) at each dimension of vector, from being divided into L section (such as 5 sections) between its peak to peak.Such as, suppose x
1Value between [0,50], then its codomain is divided into x
10, x
11... x
145 sections altogether, x
10The value of section is between [0,10], x
11The value of section between (10,20], and x
14The section value between (40,50].
2) with each section of every dimension, each section with other 3 dimensions carries out unduplicated combination, forms L * L * L * L combination.Each combination is an interval just, such as (x
10, x
21, x
33, x
42).This interval center is for the central point of each dimension on this section, such as (x
10c, x
21c, x
33c, x
42c).
3) in the interval of each after division, search the sample point that drops in the database in this interval, and find the sample point of occurrence risk in these sample points.Such as dropping on (x in the database
10, x
21, x
33, x
42) sample point have 10, the sample point that risk occurs has 4.In this example, just think, in 4 intermediate item indexs near x
10c, x
21c, x
33c, x
42cUnder the situation of (interval central point), risk probability is 4/10=40%.
Add up each interval interior risk probability, and record this interval center point value, namely obtain a large amount of (risk probability, x
1, x
2, x
3, x
4) the data combination.The formed set of this 5 tuple namely can be used for the training of risk evaluation model.Here, the number of combination is exactly the number in spatial division interval in fact, such as L4.Interval division is more many, and each interval is more little, and interval interior sample point is more little apart from the central point distance, and the sample point condition is just more approximate, and the risk probability computational accuracy is just more high, but calculated amount is just more big.Otherwise interval division is more few, and single interval is more big, and the sample point that drops in the single interval is more, but the risk probability precise decreasing, calculated amount also descends.
By the above method based on vector space division statistical history risk probability, can obtain (historical risk probability, respective item index) data set of various risks.These data will be for the evaluation model training of various risks.Model training can adopt support vector machine (SVM), neural network, KNN methods such as (K-Nearest Neighbor algorithm:K nearest neighbor algorithms), and this instructions is that the example explanation utilizes training data to set up the step of forecast model with the SVM algorithm.Training step is referring to accompanying drawing 9.At first, computing unit 204 is the selected kernel function of non-linear SVM, for example:
Here σ is constant (step 1000).Then, be provided for the system of equations of model training, for example:
Here c is constant, b and a
iBe to need the parameter (step 1001) determined in the equation.With off-the-shelf (risk probability, indicator vector) data set as the (y in the equation
i, x
i) substitution, solve equation parameter b and a
iValue (step 1002).After all parameters are determined, create risk evaluation model, such as:
Here, x
iBe the indicator vector { x that is correlated with
1, x
2..., x
m) (1003).So far, the probability assessment model of certain risk draws.The related data that should use various risks is every kind of risk training probability assessment model separately.
In addition, after obtaining the project risk evaluation model, also can carry out risk assessment to new software project in device inside.In this case, software project risk assessment device of the present invention can also comprise evaluation unit, and the project risk evaluation model that utilizes computing unit 204 to calculate by evaluation unit carries out risk assessment to new software project, obtains the forecasting risk probability.
Figure 10 is the process flow diagram that adopts risk model computational item risk probability.As shown in figure 10, the m dimensional vector index relevant with risk i in collecting new projects, with its normalization, and create m dimensional vector X (step 1100) afterwards, evaluation unit adopts the risk evaluation model that trains, the project indicator in the substitution new projects (m dimensional vector X) calculates the risk probability of happening (step 1101) of prediction.Judge the size (step 1102) of risk possibility occurrence according to above result.Simultaneously, the various data relevant with new projects can be preserved into database, as the reference data of subsequent project risk assessment.
And, also can be that after new software project was finished, the information relevant with the project indicator and project risk that collector unit 200 is collected this software project as historical data, was made the project risk evaluation model again.Like this, the sample size of new project risk evaluation model institute foundation increases, thereby the result is more accurate.
According to the present invention, by this codomain is refined as the method in subvalue territory, can find out the rule that project risk occurs simply, the risk that exists in the discovery new projects, minimizing improve the accuracy rate that risk is found, and the scope of application is extensive to the dependence of expertise.
(second embodiment)
In the first embodiment, establishing all items index all is the project indicator relevant with risk, carries out the foundation of risk profile model.But, because establishing all items index all is the relevant item index, ignored the influence of project kind to the project indicator, and the project indicator that the project indicator that the degree of correlation is high and the degree of correlation are low is used with, increase the calculated amount of model training, and might reduce the accuracy of forecast model to a certain extent.Therefore, in second embodiment, increasing the data of collecting are carried out the process handled early stage, is foundation with the degree of correlation, the kind of the project indicator is screened, thereby reduced the calculated amount of model training, and the accuracy that improves forecast model.This second embodiment is especially effective under the bigger situation of sample range.
Apparatus main body formation in second embodiment and the mode of setting up and first embodiment of predictive mode are similar, the main difference point is, the data that collector unit 200 is collected are carried out degree of correlation analysis, only with degree of correlation index the regulation degree of correlation threshold value more than the project indicator as the relevant item index, be reused for segmentation step, being about to a plurality of project indicator limits is the relevant item index, only the relevant item index is used for the generation of the training data of forecast model.
Fig. 4 is the block diagram of the software project risk assessment device that relates to of second embodiment.This software project risk assessment device comprises: data collection module 400, degree of correlation analysis module 402, sample data analysis module 403, model regretional analysis module 404 and risk probability evaluation module 405.
Wherein, data collection module 400 is equivalent to collector unit, collects historical data and the ongoing new projects data of software project, and data are saved in the database 401.
Degree of correlation analysis module 402 is modules of dividing function, statistical function and degree of correlation analytic function with a dimension indicator, according to the central value in each subvalue territory of certain intermediate item index and certain intermediate item risk number of times of following this subvalue territory to take place, calculate the degree of correlation index between this certain intermediate item risk and this certain intermediate item index, only with the project indicator of degree of correlation index more than the degree of correlation threshold value of regulation as the relevant item index, the codomain that is used for the multi-C vector space is cut apart.
Sample data analysis module 403 is equivalent to the module that the function with statistic unit and cutting unit becomes one.Be used for the correlation analysis result according to degree of correlation analysis module 402, extract the project indicator relevant with risk from database 401, the flow process according to first embodiment calculates (historical risk probability, relevant item index) data set again.Wherein, the flow process of said Fig. 6 is finished by sample data analysis module 403 exactly in first embodiment, therefore omits detailed explanation.
Again, for ease of explanation, with statistic unit with cutting unit all is included in degree of correlation analysis module 402 and sample data analysis module 403 describes, but also can only have independently statistic unit, cutting unit and sample data analytic unit, the degree of correlation of carrying out present embodiment by the transmission of data is handled.In this situation, follow the occurrence number of certain intermediate item index that certain intermediate item risk takes place to send to the sample data analytic unit in that statistic unit and cutting unit will come out earlier, each subvalue territory and carry out degree of correlation analysis, accept the result that the degree of correlation is analyzed again, only utilize the relevant item index to carry out again that codomain is cut apart and statistical treatment.Wherein codomain cut apart identical with first embodiment with the process of statistical treatment, so detailed.
Model regretional analysis module 404 is equivalent to computing unit, the data set that the relevant item index that obtains based on sample data analysis module 403 and the risk probability (typical value in subvalue territory) in subvalue territory constitute, as training data, train the risk evaluation model of various risks.The Fig. 9 that illustrates in training flow process and first embodiment is identical, so omits detailed explanation.
Risk probability evaluation module 405 is equivalent to evaluation unit, based on the index of correlation of these models and new projects, calculates the probability of happening of various risks in the new projects.
Fig. 3 is the process flow diagram of the software project risk evaluating method that relates to of second embodiment.At first, data collection module 400 is collected polytype project indicator, and the project risk of record appearance, and the project indicator, the risk of collecting is saved in the database 401 (step 300).
Then, the degree of correlation between degree of correlation analysis module 402 analysis project indexs and the project risk, sample data analysis module 403 and model regretional analysis module 404 are created the evaluation model of project risk probability, this model can the quantization means project risk and index between relation (step 301), concrete degree of correlation analytic process is described in detail in the back.
After determining risk evaluation model, risk probability evaluation module 405 is collected the project indicator of new projects, adopts these project indicators and risk probability model to calculate the probability of happening (step 302) of various risks.
So far, utilize software project risk assessment device of the present invention, project administrator can carry out judgement and the prevention of risk according to the various risk probability of happening that draw.
Can also the project indicator and the practical risk occurrence record of new projects be saved in database by data collection module 400, as required, adopt the Data Update risk probability model (step 303) after upgrading.This is equivalent to the data of new projects are used, and is applied to the risk assessment of following project again.
Below the degree of correlation analysis that relates in second embodiment is elaborated.
Before training the risk evaluation model of user-specific, at first to find out the factor of influence in the model, just find out the relevant project indicator in danger of following the wind.This process of identifying project index and the risk degree of correlation is illustrated in accompanying drawing 11.
Degree of correlation analysis module 402 is carried out the flow process shown in Figure 11.As shown in figure 11, for the degree of correlation of identify project index and risk, need pursue analyzing the project indicator and risk.At first, select a pair of risk R
iWith project indicator M
j(step 600) analyzed its degree of correlation.With project indicator M
jCodomain be divided into L interval, the mid point that each is interval is designated as V
m, the bound that each is interval is designated as V
d, V
u(step 601).Statistical indicator M
jBe positioned at each interval (V
d, V
u) time risk number of times (step 602) that takes place.According to above-mentioned statistics, draw L group (risk number of times, M
j) data are to (step 603), and employing
Scheduling algorithm (also can adopt the net correlation analytical algorithm to eliminate the repeatability influence of similar index), calculation risk occurrence number and project indicator M
jBetween relevance degree W
j(step 604).Here adopting Pearson model result calculated is the value that is positioned between [1,1].Judge then | W
j| deep and remote helping imprisoned not greater than threshold value C
r, C
rIt is a threshold value constant (step 605).If greater than threshold value C
r, risk R is described
iWith index M
jBetween significant association (step 606) is arranged; Otherwise, risk R then is described
iWith index M
jBetween association can ignore (step 607).Continue to analyze the degree of correlation between sundry item risk and the index, up to having analyzed all risks and index combination (step 608).By this analytic process, can find out with every kind of several project indicators that project risk is associated, these indexs will become the factor of influence of respective risk evaluation model.
Here, a pair of risk R
iWith project indicator M
jHandle, cut apart at a certain project indicator on one dimension, this processing also can be finished by cutting unit, perhaps the partial function of cutting unit and statistic unit is incorporated in the degree of correlation analytic unit and is carried out.Formally to carry out the process of training data in extracting in method and first embodiment with statistics similar but cut apart, and difference is that degree of correlation analysis is one dimension, and formal training data to extract in the flow process be the vector space that m ties up.
And the density degree that the codomain of carrying out in order to carry out degree of correlation analysis is cut apart can to extract the density degree that the codomain of carrying out in the flow process cuts apart identical with formal training data, also can be different.
Figure 12 is the instance graph of the degree of correlation between analysis project risk and the project indicator.As shown in figure 12, selected a pair of risk i and project indicator M
jAfter, with project indicator M
jCodomain be divided into L interval, the mid point that each is interval is designated as V
m, the bound that each is interval is designated as V
d, V
uM
jItself be the curve of a variation, its codomain is the scope between maximal value and the minimum value just.Statistical indicator M
jBe positioned at each interval (V
d, V
u) time risk number of times that takes place, this interval intermediate value is V
mEach interval can obtain this two values, therefore, can obtain L " risk i frequency and project indicator M altogether
j" data right.The data of for example " paying and postponing " risk occurrence number and the formation of " defect concentration " index are right.Next, adopt the degree of correlation of Pearson model calculation risk i and Metricsj.Here, E is risk i occurrence number or index M
jExpectation value.Determine risk i and M according to degree of correlation threshold value
jBetween the degree of correlation whether enough remarkable.
According to the degree of correlation analysis of first embodiment, improved the project indicator on the whole with the degree of correlation of risk, therefore can be that the evaluation result of software project risk assessment device is more accurate.
(first variation)
Basic structure and the main flow process of the software project risk evaluating method that the present invention relates to and device more than have been described, but the present invention is not limited in above-mentioned embodiment, can also carry out various distortion and optimization on the basis of main frame, these all comprise within the scope of the invention.Below be described in detail at several important variation.
At first, need to create the probability assessment model of various risks among the present invention.In order to verify the accuracy of these risk evaluation models, software project risk assessment device of the present invention can also comprise authentication unit, this authentication unit with a part of data in the historical data as verification msg, substitution project risk evaluation model is verified uses the forecasting risk probability, and calculate the difference of the forecasting risk probability of this verification msg and real history risk probability with respect to the number percent of real risk probability, be lower than at above-mentioned number percent under the situation of regulation verification threshold, do not use this project risk evaluation model to carry out risk assessment.
Figure 13 is the process flow diagram that explanation utilizes verification msg training risk evaluation model.As shown in figure 13, at first, with the historical data collected at first or advanced degree of correlation analysis and project risk and achievement data with significant correlation is divided into two groups according to project, one group is used for model training, and one group is used for modelling verification (step 700).
More than divide can project for according to dividing, for example with 7 one-tenth historical datas as training data, all the other 3 one-tenth historical datas are as verification msg.Set up the sample that is used for model training, sample comprises historical risk probability and corresponding project indicator data (step 701).
Next, adopt the method that illustrates in above first, second embodiment to extract training data, and adopt regression algorithm such as SVM to train every kind of risk evaluation model (step 702) separately.
At last, adopt the verification msg that keeps in advance to be updated to these risk evaluation models, whether estimated risk evaluation model (step 703) is tested the model accuracy that trains and can be accepted.
Specifically, according to the historical statistics in the training data, calculate the real risk probability of verification msg, with the project indicator difference substitution risk evaluation model of training data, obtain risk probability and predict the outcome again; To predict the outcome and compare with the real risk probability.If both standard deviations greater than certain percentage threshold of real risk probability (such as greater than real risk 20%), then think this model accuracy deficiency, abandoned; Otherwise then accepting this model is new risk evaluation model.
By above-mentioned verification step, can find the bigger risk probability model of accuracy to be used, thereby improve the precision of software project risk assessment device.
(second variation)
In above-mentioned historical risk probability statistical method, vector space is divided and is being played the part of important role.But how preventing that spatial division is overstocked or dredging excessively is a problem.Therefore, cutting unit carries out the judgement of spatial division density when also can codomain cutting apart in advance, determines optimum partitioning scheme.
Figure 14 has illustrated a kind of mechanism of control vector spatial division quantity, and is can the control vector spatial division overstocked or thin excessively.At first cutting unit 202 is selected a kind of risk R
iM dimension indicator vector X (step 900) with relevant tries to divide, with every dimension X of vectorial X
jCodomain is divided into 10 sections.Dividing line is interweaved, thereby forms 10 in the m gt
mIndividual interval (step 901).
Then, the risky total sample number of institute is designated as N
t, the risk number of samples in the space that the risk number of samples is maximum is designated as N
m(step 902).Judge N
mWhether less than (N
t/ 100), at N
mLess than (N
t/ 100) time, be judged as the interval too much, divide too intensive (step 903).Flow process enters step 904.
In step 904, with the each minimizing 10% of the division hop count on every dimension, judge again, be false until above-mentioned inequality.
If for negating, then enter step 905 in step 903.Judge whether N
mWhether greater than (N
t/ 2) (step 905).Under sure situation, be judged as interval very fewly, to divide too sparsely, flow process enters step 906.After the each increase by 10% of the division hop count on every dimension, judge again, be false until above-mentioned inequality.
If for negating, then enter step 907 in step 905.Through above adjustment, can access the vector space dividing mode that is not too intensive or too sparse.
The above-mentioned number percent that goes forward one by one increase and decrease is not limited to 10%, can adjust according to the calculating of equipment.Thus, can determine the distance of cutting apart of vector space the best.
(the 3rd variation)
In the above-described 2nd embodiment, the degree of correlation between project risk and the project indicator is estimated, determined that the project indicator that the degree of correlation is higher than certain threshold value carries out risk assessment.But, along with variation and the development of environment, be identical project risk and the project indicator exactly, the degree of correlation between project risk and the project indicator also is in continuous change procedure.Therefore, degree of correlation analytic unit also can regularly carry out the correction of degree of correlation threshold value, makes degree of correlation threshold value more near actual conditions.
In this variation, each phase certain intermediate item risk and the project indicator abandoned is reused for the foundation of evaluation model relatively regularly of degree of correlation analytic unit, set up interim evaluation model, and the risk assessment result that will utilize interim evaluation model to obtain compares with the risk assessment result who utilizes former evaluation model to obtain, the risk assessment result that the risk assessment result who obtains at interim evaluation model obtains than former evaluation model is adjusted into the corresponding degree of correlation of this project indicator of abandoning with the degree of correlation threshold value of stipulating during more near historical risk probability.
Figure 15 is the process flow diagram of adjusting degree of correlation threshold value.As shown in figure 15, can revise index degree of correlation threshold value according to the precision of risk evaluation model.Main design is whether increase the necessity of coming judgment threshold to adjust according to adjusting index of correlation postevaluation precision.
At first, to risk i, calculating the degree of correlation of all items index and this risk, press the order of magnitude ordering, is that boundary is divided into two groups (step 1200) with all items index with default threshold t.
Degree of correlation threshold value according to acquiescence is carried out choosing of risk index of correlation, based on index set (x
1..., x
m), finish the training (step 1201) of master pattern.
Then, according to the descending order of the degree of correlation, the initial index x that abandons
M+1Add and advance effective range, based on extended counter collection (x
1..., x
m, x
M+1), training pattern (step 1202) again.
The precision (step 1203) that compares new model and master pattern.Ratio of precision can be carried out based on verification msg.At first according to historical data, calculate the real risk probability (for example trying to achieve by the step 101,102 of Fig. 1) of verification msg; With the various project indicators of verification msg, distinguish substitution new model and master pattern more then, draw risk probability respectively and predict the outcome; Two kinds predicted the outcome compares with the real risk probability, sees which is more near true probability.If have the evaluation model of multiple risk or the verification msg of a plurality of projects, when comparing new model and master pattern order of accuarcy, carry out according to the variance or the standard deviation that predict the outcome with true probability.
If the precision of new model is higher, then degree of correlation threshold value is modified to index x
M+1The degree of correlation, and replace master pattern (step 1205) with new model, the degree of correlation index that continues again to attempt the next one is abandoned is expanded into index set (step 1206); If the new model precision is lower, illustrates and abandon index x
M+1Be reasonably, degree of correlation threshold value and master pattern constant (step 1204).
Should be noted in the discussion above that increases degree of correlation threshold value correction step, can help to improve the risk assessment precision, but under the more situation of calculated amount, the particularly project indicator that can aggrandizement apparatus.This is because each model training calculated amount is all bigger, present threshold value correction strategy be with repeatedly again training pattern be prerequisite.Under actual conditions, for the big consideration of calculated amount, preferably limit the index number relevant with risk.
In sum, the present invention can be applied in the software project risk assessment in various fields, helps the various risks in the project administrator discovery software project better.Adopt modeling method of the present invention and device effectively to utilize historical data, find out the rule that project risk occurs, the risk that exists in the discovery new projects reduces the dependence to expertise, improve the accuracy rate that risk is found, and the scope of application is extensive.
Claims (18)
1. a software project risk evaluating method is characterized in that,
Comprise:
Collect step, the historical data about the project indicator and project risk of collecting a plurality of samples;
Segmentation step is divided into above-mentioned project indicator institute range of a mapping a plurality of, and the subvalue territory after cutting apart at each replaces tabular value, if when being m with respect to the kind quantity of the project indicator of certain intermediate item risk, above-mentioned codomain is the m gt, and wherein, m is the integer greater than 0;
Statistic procedure, according to the historical data of collecting, add up the quantity of the project indicator mapping point of following certain intermediate item risk generation in each subvalue territory with respect to the ratio of the quantity that is mapped to all items index mapping point in this subvalue territory, as the risk probability of such project risk appearance, and
Calculation procedure, the above-mentioned risk probability according to after above-mentioned typical value and the statistics carries out normalized, as training data, makes the project risk evaluation model.
2. software project risk evaluating method according to claim 1 is characterized in that,
In above-mentioned segmentation step, the sample size that is mapped in certain subvalue territory is compared the size in the subvalue territory that adjustment is cut apart according to comparative result with the amount threshold of regulation.
3. software project risk evaluating method according to claim 1 is characterized in that,
Also comprise verification step, with a part of data in the historical data as verification msg, substitution project risk evaluation model is verified uses the forecasting risk probability, and calculate this checking with the difference of the historical risk probability of forecasting risk probability and the verification msg number percent with respect to historical risk probability, be lower than at above-mentioned number percent under the situation of regulation verification threshold, do not use this project risk evaluation model to carry out risk assessment.
4. software project risk evaluating method according to claim 1 is characterized in that,
Also comprise evaluation procedure, utilize the project risk evaluation model that calculates in the above-mentioned calculation procedure that new software project is carried out risk assessment, obtain the forecasting risk probability.
5. software project risk evaluating method according to claim 1 is characterized in that,
Also comprise feedback step, after new software project is finished, collect the information feedback relevant with the project indicator and project risk of this software project and give above-mentioned collection step, as historical data, make the project risk evaluation model again.
6. software project risk evaluating method according to claim 1 is characterized in that,
In above-mentioned segmentation step, get the central value in each subvalue territory as the typical value in this subvalue territory.
7. software project risk evaluating method according to claim 1 is characterized in that,
Also comprise degree of correlation analytical procedure, according to the occurrence number that comes out in above-mentioned statistic procedure, follow certain intermediate item index that certain intermediate item risk takes place in each subvalue territory, calculate the degree of correlation index between this certain intermediate item risk and this certain intermediate item index, only the project indicator of degree of correlation index more than the degree of correlation threshold value of regulation is used for segmentation step as the relevant item index
In above-mentioned segmentation step, only above-mentioned relevant item index institute range of a mapping is cut apart as the m gt.
8. software project risk evaluating method according to claim 7 is characterized in that,
In above-mentioned degree of correlation analytical procedure, each is phase certain intermediate item risk and the project indicator abandoned is reused for the foundation of evaluation model relatively regularly, set up interim evaluation model, and relatively with historical data respectively the interim evaluation model of substitution and former evaluation model obtain and the risk assessment result, during more near historical risk probability, the degree of correlation threshold value of stipulating is being adjusted into the corresponding degree of correlation of this project indicator of abandoning as the risk assessment result's of former evaluation model forecasting risk probability as the risk assessment result's of interim evaluation model forecasting risk likelihood ratio.
9. software project risk evaluating method according to claim 1 is characterized in that,
In above-mentioned calculation procedure, utilize support vector machine, neural network or k nearest neighbor algorithm KNN, training data is carried out modeling handle, make the project risk evaluation model.
10. a software project risk assessment device is characterized in that,
Comprise:
Collector unit, the historical data about the project indicator and project risk of collecting a plurality of samples;
Cutting unit is divided into above-mentioned project indicator institute range of a mapping a plurality of, and the subvalue territory after cutting apart at each replaces tabular value, if when being m with respect to the kind quantity of the project indicator of certain intermediate item risk, above-mentioned codomain is the m gt, and wherein, m is the integer greater than 0;
Statistic unit, according to the historical data of collecting, add up the quantity of the project indicator mapping point of following certain intermediate item risk generation in each subvalue territory with respect to the ratio of the quantity that is mapped to all items index mapping point in this subvalue territory, as the risk probability of such project risk appearance, and
Computing unit, the above-mentioned risk probability according to after above-mentioned typical value and the statistics carries out normalized, as training data, makes the project risk evaluation model.
11. software project risk assessment device according to claim 10 is characterized in that,
The sample size that above-mentioned cutting unit will be mapped in certain subvalue territory is compared with the amount threshold of regulation, the size in the subvalue territory that adjustment is cut apart according to comparative result.
12. software project risk assessment device according to claim 10 is characterized in that,
Also comprise authentication unit, with a part of data in the historical data as verification msg, substitution project risk evaluation model is verified uses the forecasting risk probability, and calculate this checking with the difference of the historical risk probability of forecasting risk probability and the verification msg number percent with respect to historical risk probability, be lower than at above-mentioned number percent under the situation of regulation verification threshold, do not use this project risk evaluation model to carry out risk assessment.
13. software project risk assessment device according to claim 10 is characterized in that,
Also comprise evaluation unit, the project risk evaluation model that utilizes above-mentioned computing unit to calculate carries out risk assessment to new software project, obtains the forecasting risk probability.
14. software project risk assessment device according to claim 10 is characterized in that,
After new software project was finished, above-mentioned collector unit was collected the information relevant with the project indicator and project risk of this software project, as historical data, makes the project risk evaluation model again.
15. software project risk assessment device according to claim 10 is characterized in that,
Above-mentioned cutting unit is got the central value in each subvalue territory as the typical value in this subvalue territory.
16. software project risk assessment device according to claim 10 is characterized in that,
Also comprise degree of correlation analytic unit, follow the occurrence number of certain intermediate item index that certain intermediate item risk takes place in that above-mentioned degree of correlation analytic unit comes out according to above-mentioned statistic unit, each subvalue territory, calculate the degree of correlation index between this certain intermediate item risk and this certain intermediate item index, with degree of correlation index the regulation degree of correlation threshold value more than the project indicator as the relevant item index
Above-mentioned cutting unit is only cut apart above-mentioned relevant item index institute range of a mapping as the m gt.
17. software project risk assessment device according to claim 16 is characterized in that,
Each phase certain intermediate item risk and the project indicator abandoned is reused for the foundation of evaluation model relatively regularly of above-mentioned degree of correlation analytic unit, set up interim evaluation model, and the risk assessment result that will utilize interim evaluation model to obtain compares with the risk assessment result who utilizes former evaluation model to obtain, the risk assessment result that the risk assessment result who obtains at interim evaluation model obtains than former evaluation model is adjusted into the corresponding degree of correlation of this project indicator of abandoning with the degree of correlation threshold value of stipulating during more near historical risk probability.
18. software project risk assessment device according to claim 10 is characterized in that,
Above-mentioned computing unit utilizes support vector machine, neural network or k nearest neighbor algorithm KNN, training data is carried out modeling handle, and makes the project risk evaluation model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210009752.5A CN103208039B (en) | 2012-01-13 | 2012-01-13 | Method and device for evaluating software project risks |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210009752.5A CN103208039B (en) | 2012-01-13 | 2012-01-13 | Method and device for evaluating software project risks |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103208039A true CN103208039A (en) | 2013-07-17 |
| CN103208039B CN103208039B (en) | 2017-05-03 |
Family
ID=48755256
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201210009752.5A Expired - Fee Related CN103208039B (en) | 2012-01-13 | 2012-01-13 | Method and device for evaluating software project risks |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103208039B (en) |
Cited By (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105721410A (en) * | 2014-12-05 | 2016-06-29 | 北京奇虎科技有限公司 | Method and device for obtaining network safety state |
| CN106600007A (en) * | 2016-11-29 | 2017-04-26 | 上海亿账通互联网科技有限公司 | Activity quality evaluation method and device |
| WO2017167071A1 (en) * | 2016-03-30 | 2017-10-05 | 阿里巴巴集团控股有限公司 | Application program project evaluation method and system |
| CN107480435A (en) * | 2017-07-31 | 2017-12-15 | 广东精点数据科技股份有限公司 | A kind of automatic searching machine learning system and method applied to clinical data |
| WO2018120726A1 (en) * | 2016-12-30 | 2018-07-05 | 平安科技（深圳）有限公司 | Data mining based modeling method, system, electronic device and storage medium |
| CN108449307A (en) * | 2017-02-16 | 2018-08-24 | 上海行邑信息科技有限公司 | A method of risk equipment for identification |
| CN109376929A (en) * | 2018-10-24 | 2019-02-22 | 北京小度信息科技有限公司 | Dispense determination method for parameter, determining device, storage medium and electronic equipment |
| CN109816221A (en) * | 2019-01-07 | 2019-05-28 | 平安科技（深圳）有限公司 | Decision of Project Risk method, apparatus, computer equipment and storage medium |
| CN109978330A (en) * | 2019-02-20 | 2019-07-05 | 华电电力科学研究院有限公司 | A kind of department's safety management level measurement system |
| CN110399302A (en) * | 2019-07-26 | 2019-11-01 | 中国工商银行股份有限公司 | Risk Identification Method, device, electronic equipment and the medium of Software Testing Project |
| CN111143212A (en) * | 2019-12-24 | 2020-05-12 | 中国航空工业集团公司西安飞机设计研究所 | Functional logic function library verification method under module integrated software architecture |
| CN112363928A (en) * | 2020-11-10 | 2021-02-12 | 网易（杭州）网络有限公司 | Test case processing method and device, processor and electronic device |
| CN112379913A (en) * | 2020-11-20 | 2021-02-19 | 上海复深蓝软件股份有限公司 | Software optimization method, device, equipment and storage medium based on risk identification |
| CN112882959A (en) * | 2021-03-30 | 2021-06-01 | 中信银行股份有限公司 | Project risk judgment method, device and equipment and readable storage medium |
| CN113297333A (en) * | 2021-03-17 | 2021-08-24 | 无锡极数宝大数据科技有限公司 | Data processing method, device, server and storage medium |
| CN113296836A (en) * | 2021-06-08 | 2021-08-24 | 北京百度网讯科技有限公司 | Method for training model, testing method, device, electronic equipment and storage medium |
| CN114742351A (en) * | 2022-03-09 | 2022-07-12 | 中国国际金融股份有限公司 | Project risk prediction method and module and computing device |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101283317A (en) * | 2005-02-11 | 2008-10-08 | 伏特资讯科学公司 | Project work change in plan/scope administrative and business information synergy system and method |
| CN101339596A (en) * | 2008-08-26 | 2009-01-07 | 腾讯科技（深圳）有限公司 | Method and device for protecting computer software system |
| US7669180B2 (en) * | 2004-06-18 | 2010-02-23 | International Business Machines Corporation | Method and apparatus for automated risk assessment in software projects |
| CN101826202A (en) * | 2009-08-26 | 2010-09-08 | 张艳红 | Patch space theory system, model and implementation after project management risk line breakage |
-
2012
- 2012-01-13 CN CN201210009752.5A patent/CN103208039B/en not_active Expired - Fee Related
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7669180B2 (en) * | 2004-06-18 | 2010-02-23 | International Business Machines Corporation | Method and apparatus for automated risk assessment in software projects |
| CN101283317A (en) * | 2005-02-11 | 2008-10-08 | 伏特资讯科学公司 | Project work change in plan/scope administrative and business information synergy system and method |
| CN101339596A (en) * | 2008-08-26 | 2009-01-07 | 腾讯科技（深圳）有限公司 | Method and device for protecting computer software system |
| CN101826202A (en) * | 2009-08-26 | 2010-09-08 | 张艳红 | Patch space theory system, model and implementation after project management risk line breakage |
Cited By (26)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105721410B (en) * | 2014-12-05 | 2020-05-01 | 北京奇虎科技有限公司 | Method and device for acquiring network security condition |
| CN105721410A (en) * | 2014-12-05 | 2016-06-29 | 北京奇虎科技有限公司 | Method and device for obtaining network safety state |
| WO2017167071A1 (en) * | 2016-03-30 | 2017-10-05 | 阿里巴巴集团控股有限公司 | Application program project evaluation method and system |
| CN107292463A (en) * | 2016-03-30 | 2017-10-24 | 阿里巴巴集团控股有限公司 | A kind of method and system that the project evaluation is carried out to application program |
| CN106600007A (en) * | 2016-11-29 | 2017-04-26 | 上海亿账通互联网科技有限公司 | Activity quality evaluation method and device |
| WO2018120726A1 (en) * | 2016-12-30 | 2018-07-05 | 平安科技（深圳）有限公司 | Data mining based modeling method, system, electronic device and storage medium |
| CN108449307A (en) * | 2017-02-16 | 2018-08-24 | 上海行邑信息科技有限公司 | A method of risk equipment for identification |
| CN108449307B (en) * | 2017-02-16 | 2020-12-29 | 上海行邑信息科技有限公司 | Method for identifying risk equipment |
| CN107480435A (en) * | 2017-07-31 | 2017-12-15 | 广东精点数据科技股份有限公司 | A kind of automatic searching machine learning system and method applied to clinical data |
| CN107480435B (en) * | 2017-07-31 | 2020-12-08 | 广东精点数据科技股份有限公司 | Automatic search machine learning system and method applied to clinical data |
| CN109376929A (en) * | 2018-10-24 | 2019-02-22 | 北京小度信息科技有限公司 | Dispense determination method for parameter, determining device, storage medium and electronic equipment |
| CN109816221A (en) * | 2019-01-07 | 2019-05-28 | 平安科技（深圳）有限公司 | Decision of Project Risk method, apparatus, computer equipment and storage medium |
| CN109816221B (en) * | 2019-01-07 | 2024-01-02 | 平安科技（深圳）有限公司 | Project risk decision method, apparatus, computer device and storage medium |
| CN109978330A (en) * | 2019-02-20 | 2019-07-05 | 华电电力科学研究院有限公司 | A kind of department's safety management level measurement system |
| CN110399302A (en) * | 2019-07-26 | 2019-11-01 | 中国工商银行股份有限公司 | Risk Identification Method, device, electronic equipment and the medium of Software Testing Project |
| CN111143212A (en) * | 2019-12-24 | 2020-05-12 | 中国航空工业集团公司西安飞机设计研究所 | Functional logic function library verification method under module integrated software architecture |
| CN111143212B (en) * | 2019-12-24 | 2023-06-23 | 中国航空工业集团公司西安飞机设计研究所 | Functional logic function library verification method under module integrated software architecture |
| CN112363928B (en) * | 2020-11-10 | 2023-08-22 | 网易（杭州）网络有限公司 | Processing method and device of test case, processor and electronic device |
| CN112363928A (en) * | 2020-11-10 | 2021-02-12 | 网易（杭州）网络有限公司 | Test case processing method and device, processor and electronic device |
| CN112379913A (en) * | 2020-11-20 | 2021-02-19 | 上海复深蓝软件股份有限公司 | Software optimization method, device, equipment and storage medium based on risk identification |
| CN112379913B (en) * | 2020-11-20 | 2022-01-07 | 上海复深蓝软件股份有限公司 | Software optimization method, device, equipment and storage medium based on risk identification |
| CN113297333A (en) * | 2021-03-17 | 2021-08-24 | 无锡极数宝大数据科技有限公司 | Data processing method, device, server and storage medium |
| CN112882959A (en) * | 2021-03-30 | 2021-06-01 | 中信银行股份有限公司 | Project risk judgment method, device and equipment and readable storage medium |
| CN112882959B (en) * | 2021-03-30 | 2024-05-17 | 中信银行股份有限公司 | Project risk judging method, device, equipment and readable storage medium |
| CN113296836A (en) * | 2021-06-08 | 2021-08-24 | 北京百度网讯科技有限公司 | Method for training model, testing method, device, electronic equipment and storage medium |
| CN114742351A (en) * | 2022-03-09 | 2022-07-12 | 中国国际金融股份有限公司 | Project risk prediction method and module and computing device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103208039B (en) | 2017-05-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103208039A (en) | Method and device for evaluating software project risks | |
| US20190370684A1 (en) | System for automatic, simultaneous feature selection and hyperparameter tuning for a machine learning model | |
| Ye et al. | Research on quantitative assessment of climate change risk at an urban scale: Review of recent progress and outlook of future direction | |
| JP7090936B2 (en) | ESG-based corporate evaluation execution device and its operation method | |
| Yaacob et al. | Predicting student drop-out in higher institution using data mining techniques | |
| Piad et al. | Predicting IT employability using data mining techniques | |
| CN111444247A (en) | KPI (Key performance indicator) -based root cause positioning method and device and storage medium | |
| CN106067094A (en) | A kind of dynamic assessment method and system | |
| CN115688404B (en) | Rainfall landslide early warning method based on SVM-RF model | |
| CN102930495B (en) | Steganography evaluation based steganalysis method | |
| CN116228021A (en) | Mine ecological restoration evaluation analysis method and system based on environment monitoring | |
| Liu et al. | Data adaptive functional outlier detection: Analysis of the Paris bike sharing system data | |
| CN103678683A (en) | Precision agriculture-oriented weighted spatial fuzzy clustering method and device | |
| CN112767126A (en) | Collateral grading method and device based on big data | |
| CN109615018B (en) | User personalized behavior evaluation method and device, computer equipment and storage medium | |
| Cao | Design and optimization of a decision support system for sports training based on data mining technology | |
| CN116366277A (en) | Network security situation assessment method for information fusion | |
| CN104298751A (en) | Method and device for obtaining internet virtual identity characteristic information | |
| Malhotra et al. | A neuro-fuzzy classifier for website quality prediction | |
| David et al. | Crime Forecasting using Interpretable Regression Techniques | |
| Sadi-Nezhad et al. | A new fuzzy clustering algorithm based on multi-objective mathematical programming | |
| Iliyasu et al. | Predictive Analysis of Road Traffic Accidents in Katsina State, Nigeria Using Machine Learning Algorithms: A Study on Factors and Mitigation Strategies | |
| Siraj et al. | Data mining and neural networks: the impact of data representation | |
| Obuandike et al. | Classification of Crime Data for Crime Control Using C4. 5 and Naïve Bayes Techniques | |
| White | An investigation of nonparametric data mining techniques for acquisition cost estimating |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20170503 Termination date: 20200113 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |