JP7062936B2 - Quality prediction program, quality prediction method and quality prediction device - Google Patents

Description

The present invention relates to a quality prediction program, a quality prediction method and a quality prediction device.

Various tests for software quality evaluation are performed according to the software development process. Tests for software quality evaluation include, for example, program tests, integration tests and system tests. The integration test is performed in the process after the program test, and the system test is performed in the process after the integration test.

As a related technique, a technique related to a project management device that can objectively evaluate the status of a project visually has been proposed (see, for example, Patent Document 1). Further, a technique for improving the prediction performance such as failure prediction of software has been proposed (see, for example, Patent Document 2). Further, a technique relating to a quality evaluation system for accurately detecting an abnormality has been proposed (see, for example, Patent Document 3).

Japanese Unexamined Patent Publication No. 2006-85277 Japanese Unexamined Patent Publication No. 2013-65084 Japanese Unexamined Patent Publication No. 2017-161991

The quality of software is evaluated by an evaluator based on, for example, test results and test items. Therefore, depending on the skill of the evaluator, the accuracy of software quality evaluation may be low.

For example, if a problem is inherent in the test result of a program test, but the test result is determined to be normal, there may be a problem with the quality of the software in a later integration test or system test. It may turn out.

In the above case, since the software development process is in progress, it becomes difficult to identify the problems inherent in the software. Therefore, for various tests in software development, it is desirable to identify problems in software quality evaluation at an early stage.

As one aspect, the present invention aims to predict software problems at an early stage with high accuracy.

In one embodiment, the quality prediction program calculates the test case setting rate and the failure rate based on the first test result at the completion of the past program test of the software, and the program test for the first test result and the software. Machine learning is performed by referring to the storage unit that stores the second test results of the tests to be performed later, and inputting the first test results and the second test results , the test case setting rate, and the failure rate . , A training model is generated, and based on the test result at the completion of the program test of the verification target software and the training model, before the test performed after the program test, the verification target software is performed after the program test. The test results are predicted , the test case setting rate and the range in which the failure rate is normal are displayed on the graph for the test case setting rate and the failure rate, and the prediction is made for the plurality of the verified software. Have the computer perform the process of plotting and displaying the test results .

According to one aspect, software problems can be predicted early and with high accuracy.

It is a functional block diagram which shows an example of a quality prediction apparatus. It is a figure which shows an example of a learning model generation. It is a figure which shows an example of the learning target data. It is a figure which shows an example of quality prediction using a learning model. It is a figure which shows an example of the prediction result list. It is a figure which shows an example of the prediction result graph. It is a flowchart (the 1) which shows an example of a learning model generation. It is a flowchart (2) which shows an example of learning model generation. It is a flowchart which shows an example of the flow of a quality prediction process. It is a flowchart which shows an example of the flow of a plot designation process. It is a figure which shows an example of the hardware composition of a quality prediction apparatus.

<Example of quality prediction device>
Hereinafter, embodiments will be described with reference to the drawings. FIG. 1 is a diagram showing an example of the quality prediction device 1. The quality prediction device 1 includes a control unit 2 and a storage unit 3. Further, the quality prediction device 1 is connected to the display device 4 and the input device 5. The display device 4 is, for example, a display, and the input device 5 is, for example, a mouse, a keyboard, or the like.

The control unit 2 performs various controls in the embodiment. The control unit 2 includes a data acquisition unit 11, a calculation unit 12, a static analysis unit 13, a learning model generation unit 14, a prediction unit 15, and a display control unit 16. Various types of data are stored in the storage unit 3. The learning model generation unit 14 and the prediction unit 15 may be realized by, for example, the R language.

The data acquisition unit 11 acquires the data stored in the storage unit 3. The calculation unit 12 performs various calculations. The static analysis unit 13 performs static analysis on the software source code.

In the embodiment, the static analysis unit 13 analyzes the source code of the software to obtain the number of cyclomatics and the number of methods of the software. The cyclomatic number is also called the cyclomatic complexity, and the cyclomatic number increases as the structure of the software becomes more complicated. For example, as the number of conditional branches contained in software source code increases, the number of cyclomatics increases. Cyclomatic numbers are an example of complexity.

The learning model generation unit 14 performs machine learning using the teacher data and generates a learning model. The learning model may be generated based on machine learning by, for example, a support vector machine or a neural network. Further, as the support vector machine, for example, a support vector machine using a Radial basis function kernel (RBF) kernel may be used.

The prediction unit 15 predicts the test result of the software to be verified after the program test based on the generated learning model and the test result at the completion of the program test of the software to be verified.

In the embodiment, it is assumed that there are three tests for performing software quality evaluation: a program test, an integration test, and a system test. The tests for quality evaluation of software are not limited to the above three tests. The integration test is performed after the program test, and the system test is performed after the integration test. Program tests are also referred to as unit tests or program unit tests.

The prediction unit 15 predicts the test result of the integration test or the system test before the integration test is performed. An integration test or system test is an example of a test performed after a program test. The prediction unit 15 stores the prediction result in the storage unit 3.

The display control unit 16 controls the display device 4 to display the prediction result by the prediction unit 15 in a table format or a graph format. The prediction result may be displayed on the display device 4 in a format other than the table format or the graph format.

<Example of learning model generation>
An example of learning model generation will be described with reference to FIG. The past results of the program test are the results showing the test results of the software performed in the past. In the embodiment, the past record of the program test includes the software name, the number of development scale steps, the number of tests, and the number of defects. The past results of the program test are an example of the first test results.

The software name is a name that identifies the software. Information other than the software name may be adopted as long as the information can identify the software. The number of development scale steps is the number of lines of software source code. The number of tests is the number of program tests performed on the software in the past. The number of defects indicates the number of defects that have occurred. The defect may be referred to as a failure.

The calculation unit 12 calculates the test rate and the failure rate. The test rate indicates the ratio of the number of tests performed to the number of development scale steps. The test rate is also referred to as a test case setting rate, a test item setting rate, or the like. The failure rate indicates the rate of extracted defects (failures) with respect to the number of development scale steps. That is, the failure rate indicates the rate at which a failure has occurred (failure occurrence rate).

For example, the test rate is calculated by the formula "test rate = number of tests / (number of development scale steps / 1000) x test rate reference value". In the case of the formula, the number of development scale steps is expressed in kilosteps. The test rate reference value is, for example, a value based on an empirical rule, and is a reference value for determining whether or not the test rate is appropriate.

For example, if the test rate reference value is "90.9" and the range of 20% before and after is the appropriate range of the test rate, the calculated test rate is within the range of "72.7 to 109.1". If so, it can be determined that the quality of the software is normal. On the other hand, if the calculated test rate is out of the above range, the quality of the software can be determined to be defective.

The failure rate is calculated by, for example, the formula “failure rate = number of defects / (number of development scale steps / 1000) × failure rate reference value”. The failure rate reference value is also, for example, a value based on an empirical rule, and is a reference value for determining whether or not the failure rate is appropriate.

For example, when the failure rate reference value is "9.2" and the range of 20% before and after is the appropriate range of the failure rate, the calculated failure rate is within the range of "7.4 to 11.0". If so, it can be determined that the quality is normal. On the other hand, if the calculated failure rate is out of the above range, it can be determined that the quality of the software is defective.

However, if either or both of the above test rate reference value and the failure rate reference value are not set to appropriate values, the accuracy of software quality evaluation will be low.

The static analysis unit 13 analyzes the source code of the software to obtain the number of cyclomatics and the number of methods.

The software name, the number of development scale steps, the number of tests, and the number of defects for a plurality of software included in the past results of the program test are input to the learning model generation unit 14. Further, the test rate and the failure rate calculated by the calculation unit 12 are input to the learning model generation unit 14 for each software. It is assumed that the past results and source code of the program test are stored in the storage unit 3.

Further, in the learning model generation unit 14, the past number of defects stored in the storage unit 3 is input. The actual number of defects in the past shows the number of defects (failures) extracted in the integration test and the system test for each software. It is assumed that the past record of the number of defects is stored in the storage unit 3. In FIG. 2, the actual number of defects in the past is represented as the number of IT / ST errors. The past number of defects is an example of the second test.

The learning model generation unit 14 generates learning target data based on various input data. The learning model generation unit 14 sets the teacher data based on the past number of defects for each software as the learning target data for each software.

When the past failure record in the integration test and the system test is "0", the learning model generation unit 14 sets the teacher data of the corresponding software as "OK" and sets it as the learning target data.

When the past failure record in the integration test and the system test is "1 case" or more, the learning model generation unit 14 sets the teacher data of the corresponding software as "NG" and sets it as the learning target data.

FIG. 3 shows an example of learning target data to which teacher data is added. It is assumed that the learning target data is stored in the storage unit 3. As shown in the example of FIG. 3, in the learning target data, when the IT / ST error number is “0”, the teacher data “OK” is set and the IT / ST error number is “1” for each software. In the case of "case" or more, the teacher data "NG" is set. Further, data other than the teacher data is input to the learning model generation unit 14 as described above.

As shown in the example of FIG. 2, the learning model generation unit 14 performs machine learning using a support vector machine using an RBF kernel, a neural network, or the like, using the learning target data in which the teacher data is set as an input. Generate a learning model. The generated learning model is stored in the storage unit 3.

As described above, the learning model generation unit 14 inputs the past record of the program test and the past failure record (IT / ST error number) of the integration test and the system test for each software, and the number of IT / ST errors. Machine learning is performed using the teacher data based on. This makes it possible to generate a learning model with high prediction accuracy.

Further, the learning model generation unit 14 performs machine learning in consideration of elements such as the number of cyclomatics and the number of methods for each software, and generates a learning target model. Since the number of cyclomatics and the number of methods affect the accuracy of prediction, it is possible to generate a learning model with higher prediction accuracy by performing machine learning in consideration of the number of cyclomatics and the number of methods.

Further, the learning model generation unit 14 performs machine learning using a learning target model in which factors such as a test rate and a failure rate for each software are taken into consideration. Since the test rate and the failure rate affect the accuracy of the evaluation prediction, it is possible to generate a learning model with higher prediction accuracy by performing machine learning in consideration of the test rate and the failure rate.

The learning model generation unit 14 sets the probability of occurrence of a defect with respect to the number of IT / ST errors. Since the number of IT / ST errors indicates the number of defects that have occurred in the integration test and the system test, the number of IT / ST errors has a correlation with the probability of defects (failures) that occur during the integration test and the system test. be.

For example, the learning model generation unit 14 may set the number of IT / ST errors and the probability of failure occurrence as follows. In the following, as the number of IT / ST errors increases, the probability of failure occurrence increases.
When the number of IT / ST errors = "0", the probability of failure occurrence is set to "5%".
When the number of IT / ST errors = "1 to 2", the probability of failure is set to "15%".
When the number of IT / ST errors = "3", the probability of failure occurrence is set to "30%".
When the number of IT / ST errors = "4", the probability of failure occurrence is set to "50%".
When the number of IT / ST errors = "5", the probability of failure occurrence is set to "65%".
When the number of IT / ST errors = "6", the probability of failure occurrence is set to "75%".
When the number of IT / ST errors = "7", the probability of failure occurrence is set to "90%".
When the number of IT / ST errors = "8 or more", the probability of failure occurrence is set to "100%".

The learning model generation unit 14 performs machine learning and generates a learning model. The relationship between the number of IT / ST errors and the probability of failure occurrence also changes as machine learning is performed. As a result, the probability of failure in the integration test and the system test is predicted with high accuracy. The learning model includes a prediction result of the probability of occurrence of a defect (hereinafter referred to as a probability of occurrence of a defect prediction).

<An example of quality prediction using a learning model>
An example of quality prediction using a learning model will be described with reference to the example of FIG. When a program test is performed on a given software, test results are obtained. The predetermined software is software to be verified (software to be verified).

The software to be verified is the software for which the program test has been completed, and is the software for predicting whether or not a defect will occur in the integration test and the system test.

In the example of FIG. 4, the data at the time of program test completion is data based on the test result when the program test of the software to be verified is completed, and includes the software name, the number of development scale steps, the number of tests, and the number of defects. In FIG. 4, the data at the time of completing the program test is expressed as "data at the time of completing PT".

The static analysis unit 13 analyzes the source code of the software to be verified to obtain the number of cyclomatics and the number of methods. The calculation unit 12 calculates the test rate and the failure rate using the above-mentioned formulas based on the test results for the software to be verified.

The prediction unit 15 acquires the above-mentioned various data. Further, the prediction unit 15 acquires the learning model stored in the storage unit 3. The prediction unit 15 predicts the test results of the integration test and the system test for the verification target software based on various data and the training model for the verification target software.

The prediction accuracy of the generated learning model (accuracy of prediction of failure in integration test and system test) is high. The prediction unit 15 predicts the test results of the integration test and the system test for the software to be verified by using the learning model, so that the quality of the software can be evaluated with high accuracy at an early stage before the integration test. can.

The prediction unit 15 generates a prediction result list with a prediction result and a defect prediction occurrence probability for each verification target software. FIG. 5 is a diagram showing an example of a list of prediction results. As shown in the example of FIG. 5, the prediction result list includes the above-mentioned various data, the prediction result, and the defect prediction occurrence probability.

The prediction unit 15 generates a prediction result list when the program test is completed. It is assumed that the generated prediction result list is stored in the storage unit 3. For example, the display control unit 16 may display the prediction result list shown in the example of FIG. 5 in a table format on the display device 4.

By displaying the prediction result list on the display device 4, it is possible to present the prediction result with high accuracy to the evaluator. If the prediction result indicates "NG", there is a high possibility that the corresponding software has an inherent problem.

By presenting the software predicted to be "NG" with high accuracy when the program test is completed, the evaluator can recognize that there is a high possibility that the software has a problem. In other words, the evaluator can recognize that the quality of the software corresponding to "NG" is likely to be low in the prediction result at an early stage before the integration test.

The display control unit 16 may highlight the fact that the prediction result is “NG” and the software name whose prediction result is “NG” and display it on the display device 4. For example, the display control unit 16 may display the software name whose prediction result is “NG” and the software name whose prediction result is “OK” in different colors on the display device 4.

As described above, the learning model includes the probability of failure prediction that machine learning has been performed regarding the relationship between the number of IT / ST errors and the probability of failure occurrence. The prediction unit 15 adds the defect prediction occurrence probability to the prediction result list for each verification target software.

By displaying the prediction result list to which the defect prediction occurrence probability is added on the display device 4, the evaluator can recognize the possibility of failure (failure) as a quantitative numerical value in the integration test and the system test. ..

FIG. 6 is a diagram showing an example of a prediction result graph. The prediction result graph is a graph in which the prediction results are plotted and displayed when the test rate is on the horizontal axis and the failure rate is on the vertical axis. Black circles indicate that the prediction result is "OK", and "x" indicates that the prediction result is "NG".

When the display control unit 16 accepts the designation of the above-mentioned "x" plot among the plots displayed on the display device 4, it corresponds to the "x" plot (the plot whose prediction result is "NG"). Information indicating the software to be used may be displayed on the display device 4.

For example, suppose that the evaluator operates the mouse as the input device 5, puts the mouse cursor on the plot of "x" in the central quadrant, and clicks. The control unit 2 accepts the operation.

The display control unit 16 refers to the prediction result list stored in the storage unit 3, acquires the name of the software corresponding to the designated “x” plot (plot for which the operation is accepted), and obtains the name of the software, and the display device 4 May display the name of the software. For example, the display control unit 16 may display the name of the software on the display device 4 in a pop-up manner.

Therefore, the name of the software that may be predicted to be "NG" in the integration test and the system test is displayed on the display device 4, and the evaluator determines that the test result is "NG". Recognize possible software names.

Here, the prediction result graph of the example of FIG. 6 is divided into nine quadrants. The central quadrant is within a predetermined range before and after the test rate reference value and within a predetermined range before and after the failure rate reference value. The central quadrant is an example of a range in which the set rate and failure rate are normal. For example, in the above-mentioned example, the test rate reference value is "90.9" and the failure rate reference value is "9.2".

The prediction result graph of the example of FIG. 6 shows the prediction result when the test rate reference value "90.9" is set to "1.0" and the failure rate reference value "9.2" is set to "1.0". The graph displayed in plot is shown. As described above, if the test rate and the failure rate are within the predetermined ranges, it can be determined that the quality is normal.

The central quadrant in the example of FIG. 6 indicates the quadrant at which quality can be determined to be normal. Therefore, it is assumed that all plots within the range of the central quadrant are originally black circles indicating that the prediction result is "OK".

In this regard, if the prediction result contains an "x" indicating "NG" within the range of the central quadrant, the quality of the software to be verified is normal in the judgment based on the test rate and the failure rate, but the prediction result. As for, there is a high possibility that a defect (failure) will occur in the integration test and the system test.

For example, it is conceivable to evaluate the quality of software based on the failure rate and the test rate obtained from the past results of program testing. However, one or both of the test rate reference value and the failure rate reference value may not be set to appropriate values, in which case the accuracy of software quality evaluation is reduced.

Therefore, as shown in the example of FIG. 6, even software that is shown to be normal in the judgment based on the test rate and the failure rate is likely to have a defect (failure) in the integration test and the system test. The prediction result indicating that can be presented to the evaluator.

<Example of processing flow of the embodiment>
Next, an example of the processing flow of the embodiment will be described. 7 and 8 are flowcharts showing an example of learning model generation. As shown in the flowchart of FIG. 7, the data acquisition unit 11 acquires the past results of the program test for the plurality of software from the storage unit 3 (step S1).

The calculation unit 12 calculates the test rate and the failure rate for each software based on the acquired past results of the program test by the above-mentioned formula (step S2). The static analysis unit 13 statically analyzes the source code of each software stored in the storage unit 3 and obtains the number of cyclomatics and the number of methods for each software (step S3).

The data acquisition unit 11 acquires the number of IT / ST errors for each software from the past failure records stored in the storage unit 3 (step S4). By the processing of steps S1 to S4, various data about each software are acquired.

The learning model generation unit 14 selects one of the software (step S5). The learning model generation unit 14 determines whether the number of IT / ST errors corresponding to the selected software is “0” (step S6).

If YES in step S6, the number of IT / ST errors is "0", so "OK" is set as teacher data in the selected software (step S7). If NO in step S6, the number of IT / ST errors is 1 or more, so “NG” is set as the teacher data in the selected software (step S7).

The learning model generation unit 14 determines whether or not the teacher data setting has been completed for all the software (step S9). If NO in step S9, the process proceeds to step S5. If YES in step S9, the process proceeds from "A" to step S10.

The processing after "A" will be described with reference to the flowchart of FIG. As described above, the learning model generation unit 14 sets the failure occurrence probability for the learning model with respect to the number of IT / ST errors (step S10).

The learning model generation unit 14 generates learning target data in which the above-mentioned various data, teacher data, and defect occurrence probability are represented in a table format (multidimensional array format) for each software (step S11).

The learning model generation unit 14 uses the teacher data of the training target data to perform machine learning using a support vector machine using an RBF kernel, a neural network, or the like, and generates a learning model (step S12). The generated learning model is stored in the storage unit 3 (step S13).

Next, an example of the flow of quality prediction processing will be described with reference to the flowchart of FIG. The data acquisition unit 11 acquires the test results at the completion of the program test of the plurality of verification target software from the storage unit 3 (step S21).

The calculation unit 12 calculates the test rate and the failure rate for each verification target software based on the acquired test result at the completion of the program test by the above-mentioned formula (step S22).

The static analysis unit 13 statically analyzes the source code of each verification target software stored in the storage unit 3 and obtains the number of cyclomatics and the number of methods for each verification control software (step S23).

The prediction unit 15 generates a multidimensional array for each of the acquired various data of the verification target software (step S24). The multidimensional array is a multidimensional array having no prediction result and defect prediction occurrence probability in the prediction result list shown in the example of FIG.

The prediction unit 15 predicts the quality of each software to be verified based on the generated learning model and the multidimensional array generated in step S24 (step S25). As a result, the prediction result for each verification target software can be obtained. The prediction result is added to the multidimensional array.

The prediction unit 15 adds the above-mentioned defect prediction occurrence probability to each verification target software of the multidimensional array, and generates a prediction result list shown in the example of FIG. 5 (step S26). The generated prediction result list shows the prediction result for each verification target software, and the prediction unit 15 may feed back (add) the prediction result list to the learning target data.

The prediction result list generated in step S26 is stored in the storage unit 3. The display control unit 16 causes the display device 4 to display the prediction result list stored in the storage unit 3 in a table format or a graph format (step S28).

Next, the flow of the plot designation process will be described with reference to the flowchart of FIG. When the prediction result graph is displayed on the display device 4, the control unit 2 accepts the designation of any plot by the mouse as the input device 5 (step S31). For example, when the evaluator performs an operation of moving the mouse cursor to the plot of "x" and clicking, the control unit 2 accepts the operation.

The display control unit 16 refers to the prediction result list stored in the storage unit 3 and identifies the verification target software corresponding to the designated plot (step S32). The display control unit 16 causes the display device 4 to display the specified software to be verified (step S33).

<Example of hardware configuration of quality prediction device>
Next, an example of the hardware configuration of the quality prediction device 1 will be described with reference to the example of FIG. As shown in the example of FIG. 11, the processor 111, the Random Access Memory (RAM) 112, and the Read Only Memory (ROM) 113 are connected to the bus 100. Further, the auxiliary storage device 114 and the medium connection unit 115 are connected to the bus 100.

The processor 111 executes the program expanded in the RAM 112. As the program to be executed, a quality prediction program that performs the processing in the embodiment may be applied.

The ROM 113 is a non-volatile storage device that stores the quality prediction program developed in the RAM 112. The auxiliary storage device 114 is a storage device that stores various information, and for example, a hard disk drive, a semiconductor memory, or the like may be applied. The medium connection unit 115 is provided so as to be connectable to the portable recording medium 115M.

As the portable recording medium 115M, a portable memory (for example, an optical disk, a semiconductor memory, etc.) may be applied. A determination program for performing the processing of the embodiment may be recorded on the portable recording medium 115M. The control unit 2 described above may be realized by the processor 111 executing a given quality prediction program.

The RAM 112, ROM 113, auxiliary storage device 114, and portable recording medium 115M are all examples of computer-readable tangible storage media. These tangible storage media are not temporary media such as signal carriers.

<Others>
The present embodiment and the modified examples are not limited to the embodiments described above, and various configurations or embodiments can be taken within a range that does not deviate from the gist of the present embodiment and the modified examples.

1 Quality prediction device 2 Control unit 3 Storage unit 4 Display device 5 Input device 11 Data acquisition unit 12 Calculation unit 13 Static analysis unit 14 Learning model generation unit 15 Prediction unit 16 Display control unit 111 Processor 112 RAM
113 ROM

Claims (8)

Calculate the test case setting rate and failure rate based on the first test results when the software's past program test is completed.
With reference to the storage unit that stores the first test results and the second test results of the tests performed after the program test for the software, the first test results and the second test results, the test case setting rate, and the test case setting rate Using the above failure rate as an input, machine learning is performed to generate a learning model.
Based on the test result at the completion of the program test of the software to be verified and the learning model, the test result to be performed after the program test for the software to be verified is predicted before the test performed after the program test.
The test case setting rate and the range in which the failure rate is normal are displayed on the graph for the test case setting rate and the failure rate, and the predicted test results for the plurality of software to be verified are plotted and displayed. ,
A quality prediction program characterized by having a computer perform processing. The complexity and the number of methods of the software, which is the target of the first test results, are analyzed.
Further, the machine learning is performed by inputting the complexity and the number of methods, and the learning model is generated.
The quality prediction program according to claim 1, wherein the processing is executed by the computer. When the designation of any of the test results displayed as plots in the graph is accepted, the information indicating the software to be verified corresponding to the test results is displayed.
The quality prediction program according to claim 1 or 2, wherein the processing is executed by the computer. Refer to the storage unit that stores the first test results when the past program test of the software is completed and the second test results of the tests performed after the program test for the software, and the first test results and the second test results. Is used as an input to perform machine learning, generate a learning model, and
Based on the test result at the completion of the program test of the software to be verified and the learning model, the test result to be performed after the program test for the software to be verified is predicted before the test performed after the program test.
The display device is controlled to display the verification target software and the predicted test result in association with each other, and the value set according to the number of defects of the software indicated by the second test result is used. The prediction of the probability of failure occurring in the verification target software obtained by inputting to the learning model is displayed in association with the verification target software and the test result.
A quality prediction program characterized by having a computer perform processing. Calculate the test case setting rate and failure rate based on the first test results when the software's past program test is completed.
With reference to the storage unit that stores the first test results and the second test results of the tests performed after the program test for the software, the first test results and the second test results, the test case setting rate, and the test case setting rate Using the above failure rate as an input, machine learning is performed to generate a learning model.
Based on the test result at the completion of the program test of the software to be verified and the learning model, the test result to be performed after the program test for the software to be verified is predicted before the test performed after the program test.
The test case setting rate and the range in which the failure rate is normal are displayed on the graph for the test case setting rate and the failure rate, and the predicted test results for the plurality of software to be verified are plotted and displayed. ,
A quality prediction method characterized by a computer performing processing. Refer to the storage unit that stores the first test results when the past program test of the software is completed and the second test results of the tests performed after the program test for the software, and the first test results and the second test results. Is used as an input to perform machine learning, generate a learning model, and
Based on the test result at the completion of the program test of the software to be verified and the learning model, the test result to be performed after the program test for the software to be verified is predicted before the test performed after the program test.
The display device is controlled to display the verification target software and the predicted test result in association with each other, and the value set according to the number of defects of the software indicated by the second test result is used. The prediction of the probability of failure occurring in the verification target software obtained by inputting to the learning model is displayed in association with the verification target software and the test result.
A quality prediction method characterized by a computer performing processing. A calculation unit that calculates the test case setting rate and failure rate based on the first test results when the software's past program test is completed,
With reference to the storage unit that stores the first test results and the second test results of the tests performed after the program test for the software, the first test results and the second test results, the test case setting rate, and the test case setting rate A learning model generation unit that performs machine learning and generates a learning model using the failure rate as an input.
Based on the test result at the completion of the program test of the software to be verified and the learning model, the prediction of predicting the test result of the software to be verified after the program test before the test performed after the program test. Department and
The test case setting rate and the range in which the failure rate is normal are displayed on the graph for the test case setting rate and the failure rate, and the predicted test results for the plurality of software to be verified are plotted and displayed. Display and
A quality prediction device characterized by being equipped with. Refer to the storage unit that stores the first test results when the past program test of the software is completed and the second test results of the tests performed after the program test for the software, and the first test results and the second test results. A learning model generator that performs machine learning and generates a learning model using
Based on the test result at the completion of the program test of the software to be verified and the learning model, a prediction unit that predicts the test result of the software to be verified after the program test at the completion of the program test of the software to be verified. ,
The display device is controlled to display the verification target software and the predicted test result in association with each other, and the value set according to the number of defects of the software indicated by the second test result is used. A display control unit that displays the prediction of the probability of failure occurring in the verification target software obtained by inputting to the learning model in association with the verification target software and the test result.
A quality prediction device characterized by being equipped with.

Images