JP2009187288A - Method and device for detecting project modulation - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a method and a device for detecting project modulation to detect a modulation which occurs in software development by predicting a bug extraction number in the software development based on a planned value and bug extraction results of the past projects, and comparing the predicted bug extraction number with a result value. <P>SOLUTION: The prediction of the bug extraction number for attaining the planned value is performed using the past projects, and the predicted bug extraction number is compared with the result, to detect a predictor occurring in the project. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a project modulation detection method in software development and a project modulation detection apparatus in software development. In particular, the present invention relates to a project modulation detection method in software development and a project modulation detection apparatus in software development for detecting modulation generated in development from the difference between the prediction of the number of bugs extracted in the test process and the actual results.

  In the development of large-scale and complicated software, detecting modulations that occur in a project (especially negative modulations such as the occurrence of problems) earlier and taking countermeasures earlier can reduce the cost of countermeasures, reduce the risk of project failure, etc. Is important from the point of view. This leads to both a development process and a test process in software development (there may be overlapping periods).

  One of the roles of the testing process is to test the developed software and extract and fix bugs until bug extraction converges. This is because the convergence of the number of extracted bugs is important from the viewpoint of quality control.

  A curve in which the horizontal axis represents the number of test days and the number of test steps and the vertical axis represents the total number of bugs extracted is generally called a reliability growth curve (see, for example, Non-Patent Document 1), and is an important reliability factor in the test process. It is used as one. The reliability growth curve is generally used for determining the convergence of bugs. The actual value of bug extraction is fitted to the reliability growth curve model, and when the curve has converged, it is considered that the bug extraction has been completed.

  A technique for detecting the modulation generated in the test process from the reliability growth curve can be easily inferred from a known technique.

  First, a technique for determining whether or not convergence is likely to occur as planned can be considered by predicting the number of future extractions by fitting actual values and models during the test process.

  Next, a technique is conceivable in which the planned value and the actual value are compared and detected as modulation when the difference is large.

As a related technique, a technique for effectively utilizing the reliability evaluation result in evaluating the reliability of the software testing process and the shipping time has been proposed. In the configuration, the number of pass test items, the number of detected bugs, etc., which are obtained every day in the information processing apparatus are input. The cumulative evaluation of bugs is evaluated for reliability, an optimal evaluation model is calculated, and reliability indices such as the total number of predicted bugs and bug detection rates at that time are displayed. The reliability of the number of test items is also evaluated, and the relationship between the total number of potential prediction bugs and the number of test items for detecting them is presented (for example, see Patent Document 1).
Japanese Patent Laid-Open No. 08-076991 Shigeru Yamada. Software Reliability Model-Basics and Applications-. Nikka Giren Publisher, 1994.

  In the related technology mentioned above, there is no concept of calculating the prediction (reference value) from the data up to the present, and using the planned value for generating the reference value, and all of the planned value, past information, and current data are the reference value. It was difficult to make a prediction by comparing with actual values.

  When the prediction is made by fitting the reliability growth curve model from the data at the beginning of the test process, there is a problem that the prediction accuracy is poor. The reason is that sufficient data cannot be obtained only in the early stage of the test process, and the curve overfits the data. This is because, in order to obtain sufficient data, it is necessary to wait until the end of the test process, so that it is difficult to detect the modulation at an early stage from the results of bug extraction.

  Moreover, when comparing a plan value with an actual value, there existed a problem that a detection might become late. The reason for this is that it is difficult to set a plan value in the middle of a process and is generally set only at the end of each partial process in a test process, so that modulation cannot be detected in the middle of the process.

  The present invention has been made to solve the above problems, and predicts the number of bugs to be extracted in software development based on the planned value and the bug extraction results of past projects, and compares them with the actual values. An object of the present invention is to provide a project modulation detection method and a project modulation detection apparatus for detecting a modulation generated in development.

  The project modulation detection method of the present invention is characterized by predicting the number of bugs that can be achieved by using a past project and comparing the result with a track record to detect a sign that occurs in the project.

  In addition, the project modulation detection apparatus of the present invention is characterized by predicting the number of bug extractions that achieve a planned value using a past project, and detecting a sign that occurs in the project by comparing with a past record.

  According to the present invention, it is possible to detect the modulation of the project at an early stage by using the past information to avoid overfitting of the curve and further monitoring the deviation from the plan sequentially.

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.
[First Embodiment]
Referring to FIG. 1, a project modulation detection apparatus 100 according to the first embodiment of the present invention includes a past project information storage apparatus 110, a plan information storage apparatus 120, a bug extraction number prediction means 130, and a modulation detection means. 140 and a modulation output means 150. Further, the project modulation detection apparatus 100 receives the bug extraction data 160 and the bug extraction data 170, and outputs the modulation detection result 180 detected by the project modulation detection means. The project modulation detection apparatus 100 includes a CPU, a memory, and an input / output unit similar to a general information processing apparatus. The CPU controls the device, expands and stores data in the memory, and acquires and outputs data at the input / output unit.

The bug extraction data 160 is data related to the number of bug extractions obtained from the start of the test process of the project to be monitored to the prediction, and is [x _i , t _i ], where t _i is the number of test days and test man-hours. X _i represents the total number of bugs extracted. x _i need not simply be the total number of bugs picked up, for example, the total number of bugs picked up around the scale divided by the scale of the project, or the amount obtained by performing some operation on the total number of bugs picked up. Absent.

  The bug extraction data 170 is data in the same format as the bug extraction data 160, but compared to the bug extraction data 160 used for learning for prediction, the bug extraction data 170 is the bug extraction data 160 or later. It differs in that it is data to be monitored. The bug extraction data 170 can be monitored by inputting it sequentially for each test day or test man-hour.

The past project information storage device 110 stores information related to the number of bugs extracted from past project data. In the following, the data of the j-th past project is expressed as [d ^j _i , t ^j _i ].

  The plan information storage device 120 stores bug extraction plan values planned at the time of project planning or corrected in the middle. The planned value is, for example, a value indicating how many bugs are picked up at a certain time and man-hour at the end of a partial process of the test process.

  The bug extraction number prediction means 130 stores means for predicting the number of bug extractions according to the plan value using the bug extraction data 160, past project information, and plan information. This is because, for example, a past project that fits well with the bug extraction data 160 and converges near the plan information is selected from the bug extraction data in the past project, and the number of extractions after the bug extraction data 160 is the actual value of the past project. A method of prediction based on the above can be considered.

  The modulation detection unit 140 is a unit for comparing the predicted value predicted by the bug extraction number prediction unit 130 and the bug extraction data 170 to detect modulation. For example, a method of detecting the modulation when the deviation between the predicted value and the actually measured value exceeds a certain threshold value can be considered. Further, for example, a method of detecting a change point as a modulation by applying a time series change point detection technique to a residual time series of a predicted value and an actual measurement value is also conceivable.

  If the predicted value and actual value according to the plan are different, if the actual value is small, for example, there is a concern that omission of bugs due to insufficient testing may occur, and if the actual value is large, for example, the quality in the development process may have been poor. Concerned. In addition, both may have a defect in the bug extraction plan at the planning stage, and in this case, feedback can be returned to the plan.

  The modulation output unit 150 outputs the modulation detected by the modulation detection unit 140. The output destination may be an output device such as a display connected to the project modulation detection device 100, or may be an output device or a terminal device connected via a network.

  Referring to FIG. 2, the project modulation detection apparatus 100 first inputs bug extraction data before a point of time to be predicted from the bug extraction data 160 (S100), and from the past project information storage device 110 and the plan information storage device 120, The past project information and plan information are read (S101).

  Next, the bug extraction number prediction means 130 predicts the future bug extraction number according to the plan information (S102). The prediction may be performed sequentially, or may be corrected according to a predetermined rule, for example, every time a partial process of the test process is completed.

  Next, the bug extraction data 170 is input (S103), and the modulation detection means 140 compares the predicted value calculated in step S102 and the input actual value to detect the modulation (S104). Is output by the modulation output means 150 (S105). When the project ends, the process ends. When the project continues, the modulation detection flow (S103 to S105) is sequentially performed.

[Second Embodiment]
Referring to FIG. 3, the project modulation detection apparatus 200 according to the second embodiment of the present invention is compared with the project modulation detection apparatus 100 according to the first embodiment shown in FIG. The difference is that a past project information storage device 210 is provided instead of the storage device 110 and a past project model learning device 220 is connected.

  The past project model learning device 220 is a device that learns a reliability growth curve model suitable for a past project.

  The reliability growth curve model includes various models such as an exponential inhomogeneous Poisson process model represented by (1), a logistic curve model represented by (2), and a Gompertz curve model represented by (3) and their parameters. A learning method has been proposed.

  The past project information storage device 210 stores model parameters learned by the past project model learning device 220, project scale, test days, man-hours, and the like. The stored past project model is represented by (4). p is an index of the stored past project model.

  The operation of the project modulation detection apparatus 200 according to the present embodiment is shown in FIG. 1 except that the past project information read in step S101 in FIG. 2 is information related to the model learned by the past project model learning apparatus 220. The operation is the same as that of the project modulation detection apparatus 100 according to the first embodiment.

  Although past projects have deviated from the reliability growth curve model due to various factors, by adapting the model once as in this embodiment, noise specific to the past project is removed, and along the model It is possible to make predictions.

[Third Embodiment]
Referring to FIG. 4, the project modulation detection apparatus 300 according to the second embodiment of the present invention predicts the number of bugs to be extracted as compared with the project modulation detection apparatus 200 according to the second embodiment shown in FIG. The difference is that a bug extraction number prediction means 310 is provided instead of the means 130. Predict current project as a superposition of past projects.

The bug extraction number predicting means 310 inputs the bug extraction data 160 and reads the plan value information from the plan value information storage device 120 so as to achieve the plan value while conforming to the bug extraction data 160 from the past project. Predict by combining multiple learned models. As a combination method, for example, a combination of curves expressed by addition and subtraction of past project models stored as in (4) is comprehensively calculated, and a combination that best matches the planned value and the bug extraction data 160 is selected. Can be considered. As another example, a method of expressing a prediction curve as a linear combination of past project models as shown in (5) and calculating a coupling coefficient w _p is also conceivable. As an example of the calculation method of the coupling coefficient, an arbitrary loss function (for example, a square error) is defined between the bug extraction data 160 and the plan value information and the curve expressed by (5), which is minimized. Thus, it can be calculated by obtaining w _p as follows.

  The operation of the project modulation detection apparatus 300 according to the present embodiment is calculated in step S102 that the past project information read in step S101 in FIG. 2 is information on the model learned by the past project model learning apparatus 220. 1 is the same as the operation of the project modulation detection apparatus 100 according to the first embodiment shown in FIG.

  Each of the above-described embodiments is a preferred embodiment of the present invention, and various modifications can be made without departing from the scope of the present invention. For example, a process for realizing the function of the project modulation detection device may be performed by causing the project modulation detection device to read and execute a program for realizing the function of the project modulation detection device. Further, the program is transmitted to another computer system by a transmission wave via a computer-readable recording medium such as a CD-ROM or a magneto-optical disk, or via a transmission medium such as the Internet or a telephone line. Also good. Further, a mode in which functions are distributed by an additional device is also within the scope of the present invention.

It is a block diagram which shows the structure of the project modulation | alteration detection apparatus concerning the 1st Embodiment of this invention. It is a flowchart which shows the process example of the project modulation | alteration detection apparatus concerning the 1st Embodiment of this invention. It is a block diagram which shows the structure of the project modulation | alteration detection apparatus concerning the 2nd Embodiment of this invention. It is a block diagram which shows the structure of the project modulation | alteration detection apparatus concerning the 3rd Embodiment of this invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 100 Project modulation | alteration detection apparatus 110 Past project information storage apparatus 120 Plan information storage apparatus 130 Bug extraction number prediction means 140 Modulation detection means 150 Modulation output means 160, 170 Bug extraction data 180 Modulation detection result

Claims (10)

  A project modulation detection method characterized by predicting the number of bug extractions that achieve a planned value using a past project, and detecting a sign that occurs in the project by comparing with a result. A project modulation detection method in a project modulation detection apparatus,
A step of inputting bug extraction data before the time point to be predicted from the first bug extraction data;
Reading past project information and plan information;
Predicting the future number of bugs to be extracted according to the plan information;
Inputting second bug extraction data, comparing the predicted value calculated in the predicting step with the input actual value, and detecting modulation;
The project modulation detection method according to claim 1, further comprising a step of outputting the result. The first bug extraction data is data relating to the number of bug extractions obtained from a predetermined point in time of a project to be monitored until a prediction is made,
The second bug extraction data is data to be monitored after the first bug extraction data,
The past project information is information on the number of bugs extracted from past project data,
The plan information is a bug extraction plan value planned at the time of project planning or corrected in the middle.
3. The project modulation detection method according to claim 2, wherein the number of extracted bugs is predicted according to a plan value using the first bug extraction data, the past project information, and the plan information.   The project modulation detection method according to any one of claims 1 to 3, wherein data of a past project is used after being fitted to a model that fits once.   The project modulation detection method according to any one of claims 1 to 4, wherein a plurality of past projects are combined in the prediction.   A project modulation detection apparatus characterized by predicting the number of bugs that can be achieved by using a past project and comparing the result with the actual results to detect signs of the project. Means for inputting bug extraction data prior to the point of prediction from the first bug extraction data;
Means for reading past project information and plan information;
A means of predicting the number of bugs to be extracted in the future according to the plan information;
Means for inputting second bug extraction data, detecting the modulation by comparing the predicted value calculated by the means for predicting and the input actual value;
The project modulation detecting apparatus according to claim 6, further comprising: means for outputting the result. The first bug extraction data is data relating to the number of bug extractions obtained from a predetermined point in time of a project to be monitored until a prediction is made,
The second bug extraction data is data to be monitored after the first bug extraction data,
The past project information is information on the number of bugs extracted from past project data,
The plan information is a bug extraction plan value planned at the time of project planning or corrected in the middle.
8. The project modulation detection apparatus according to claim 7, wherein the number of extracted bugs is predicted according to a plan value using the first bug extraction data, the past project information, and the plan information.   9. The project modulation detection apparatus according to claim 6, wherein data of a past project is used after being fitted to a model that fits once.   The project modulation detection apparatus according to claim 6, wherein a plurality of past projects are combined at the time of prediction.

Images