JP2019028871A - Project management support device, project management support method and program - Google Patents

Abstract

To modify estimation of man-hours and required quality created by a user on the basis of past actual result.SOLUTION: A project management support device includes: an input part for receiving inputs of information representing an extent inputted by a user and information showing the type of use of a product about the scale of a project or quality required to the product of the project; and an estimation part for modifying the information representing the extent inputted by the user by using a learned model which is caused to learn features of information representing an actual result of the scale of the project or the quality required to the product by using a self-encoder using a neural network and the information showing the type of use of the product, and generating information representing an estimation value of the scale of the project or the quality required to the product of the project.SELECTED DRAWING: Figure 1

Description

  The present invention relates to a project management support apparatus, a project management support method, and a program.

  In the past, for projects such as system development, the person in charge performed the task of estimating the required man-hours and determining the required quality, and the accuracy often depended on the experience of the person in charge. Therefore, it was difficult for inexperienced personnel to determine appropriate estimates and required quality.

  In addition, using a database that collectively manages product information and case information of computer systems and providing system planning and construction with high quality / reliability by simply entering system requirements into the database management server There is also a proposed technique (for example, Patent Document 1). In this technology, product information and case information that match the input system requirements are searched from the system construction support information database, and a summary of the system plan and estimate is created based on the searched product information and case information. .

  In addition, a technique for searching for a structure of past orders received or estimated results has also been proposed (for example, Patent Document 2). In this technology, the input information on the customer's specified requirements is compared with the information on the past orders received in the database or the requirements in the composition of the estimated results, and if these requirements match, for each requirement A predetermined number of points is added, and past orders received or estimated results are displayed on the display terminal in descending order of the total value of the added points.

JP 2003-331069 A JP 2008-204331 A

  Conventionally, when creating a plan or estimate using the input system requirements and past results, the input system requirements themselves could not be corrected. In addition, when displaying past orders received based on a match with the specified requirement, a similar item can be extracted, but the specified requirement itself cannot be corrected.

  As described above, there has been proposed a technique for extracting past results or creating estimates using the past results, but there is a possibility that the output may not be appropriate depending on the accuracy of the requirement definition. Therefore, an object of the present invention is to correct the man-hour estimation and required quality created by the user for system development or the like based on past results.

The project management support apparatus according to the present invention inputs information indicating the degree of user input regarding the scale of a project for system development or the quality required for the system to be developed, and information indicating the type of work to which the system is applied. Characteristics of information indicating the scale of the project or the quality required for the system, a self-encoder using a neural network, and information indicating the type of work to which the system is applied And an estimation unit that corrects information representing the degree input by the user using the learned learned model, and generates information representing an estimated value of the quality of the project or the system.

  In this way, using the learned model in which the features of the information representing the performance are machine-learned, the information representing the degree input by the user is corrected, and the estimated value of the quality required for the project scale or the system is obtained. be able to. That is, the man-hour estimate and required quality created by the user for system development or the like can be corrected based on past results. Since the estimate and the required quality are corrected, even an inexperienced person in charge can easily estimate the man-hours and set the required quality target. In the present invention, the accuracy of the learned model can be improved by further using information indicating the type of work to which the system is applied.

  In addition, the project management support apparatus may further include a conversion unit that converts information representing the degree input by the user into a predetermined level of detail. In this way, information representing the degree input by the user can be processed with a predetermined level of detail regardless of the level of skill of the user.

  Further, the information indicating the degree input by the user and the information indicating the actual result may be information in a predetermined image format. In this way, information can be handled in an image format suitable for processing by the self-encoder.

  The information of the predetermined image format is image data of a graph drawn at a predetermined scale. In other words, when the graph image is used as the input information, the numerical scale represented by the vertical axis, the horizontal axis, or the like may be changed to a predetermined scale. By unifying the size of the image to be handled, it becomes easier to extract the characteristics of the results.

  The contents described in the means for solving the problems can be combined as much as possible without departing from the problems and technical ideas of the present invention. The contents of the means for solving the problems can be provided as a device such as a computer or a system including a plurality of devices, a method executed by the computer, or a program executed by the computer. A recording medium that holds the program may be provided.

  The estimated man-hours and required quality created by the user for system development and the like can be corrected based on past results.

It is a figure for demonstrating the outline | summary of operation | movement of a project management assistance apparatus. It is a figure which represents typically the structure of a self-encoder. It is a block diagram which shows an example of the apparatus structure and functional structure of a project management assistance apparatus. It is a figure for demonstrating the rough flow until it learns the characteristic of the results of a project, and uses it. It is a processing flowchart which shows an example of the learning process performed in the learning stage and evaluation stage of this embodiment. It is a figure which shows an example of the information showing the performance of a project. It is a figure which shows an example of the conversion table used for a format conversion. It is a processing flowchart which shows an example of the operation process performed in an operation stage. It is a figure which shows an example of the information which shows the scale of a project. It is a figure which shows an example of the information which shows the scale of a project. It is a figure which shows an example of the information which shows the quality requested | required of a system. It is a figure which shows an example of the information which shows the quality requested | required of a system. It is a figure which shows the example of an output of quality including an achievement level. It is a figure which shows an example of the input information prepared in the format of the image. It is a figure which shows an example of the input information prepared in the format of the image.

  Hereinafter, embodiments for carrying out the present invention will be described with reference to the drawings. In addition, the structure of embodiment is an illustration and this invention is not limited to the structure shown in embodiment.

  FIG. 1 is a diagram for explaining an outline of the operation of the project management support apparatus according to the present embodiment. The project management support apparatus supports the work of estimating the scale of a project and the quality required for the system to be developed for a project such as system development.

  The scale of the project can represent the number of man-hours required for system development and the like in units such as “person month”. The man-hours may be expressed in detail for each predetermined period from the start to the end of the project, or may be roughly expressed as a total man-hour for the entire period. Further, the man-hour may be expressed in detail for each role such as a project manager, a system engineer, or a programmer, or may be expressed roughly as a total man-hour of all persons in charge.

  Further, the quality may represent the degree of importance for each quality characteristic based on a predetermined standard such as ISO / IEC25010, for example, and may represent the degree of importance for each detailed sub characteristic. May be. Information used as a quality standard required for the system is not limited to ISO / IEC 25010, and various information that can be used as a quality measurement standard can be used.

  In the example of FIG. 1, the project management support apparatus first reads out information representing past performances as a learning stage, for example, the scale of a project expressed in man-hours or quality based on a predetermined standard, and characterizes it by machine learning. Extract ((1) in FIG. 1). Machine learning can be performed using, for example, a self-encoder (auto encoder) using a neural network. Thereafter, in the operation stage, information indicating the man-hours and required quality estimated by the user for a newly started project is received, and correction is performed based on features learned in advance ((2) in FIG. 1).

FIG. 2 is a diagram schematically illustrating the configuration of the self-encoder. In the example of FIG. 2, an input layer x (x ₁ , x ₂ ,..., X _N ) including a plurality of units to which input values are input, and an intermediate layer (hidden layer) that performs predetermined conversion on the input values. ) Y (y ₁ ,..., Y _M ) and an output layer z (z ₁ , z ₂ ,..., Z _N ) that performs further predetermined conversion on the values of the intermediate layer. A plurality of intermediate layers may be provided. In addition, the number of units included in the intermediate layer y is smaller than the number of units included in the input layer x. That is, the learned model that extracted the features of the input information by compressing the information input to the input layer into an intermediate layer with a smaller amount of information and then adjusting the parameters so that the original input value can be restored Can be generated. A plurality of intermediate layers y may be provided. Information including the network configuration of the self-encoder and parameters such as weights and bias terms is referred to as a “learned model”. Each unit included in the input layer x is connected to each unit included in the intermediate layer y. On the other hand, the number of units included in the output layer z is the same as the number of units included in the input layer x. Each unit included in the intermediate layer y is connected to each unit included in the output layer z.

Conversion of the value from the input layer x to the intermediate layer y is performed using a predetermined function defined using the weight W and the bias term b. In the example of FIG. 2, the value of the intermediate layer is obtained as the sum of y obtained by the following equation (1).
y = Wx + b (1)
Note that the value of the intermediate layer may be a value further converted using a predetermined activation function. The conversion of the value from the input layer x to the intermediate layer y is called “encode”.

Further, the conversion of the value from the intermediate layer y to the output layer z is also performed using a predetermined function defined using the weight W and the bias term b. In the example of FIG. 2, the value of the output layer is obtained as the total sum of z obtained by the following equation (2).
z = Wy + b (2)
Note that the value of the intermediate layer may be a value further converted using a predetermined activation function. The conversion of the value from the intermediate layer y to the output layer z is called “decode”.

  Adjustment of parameters such as the weight W and the bias term b in the learning stage can use existing techniques such as a so-called error back propagation method (back propagation). That is, using the input value itself as a teacher value, the decoding parameter and the encoding parameter are updated in order so that the difference between the output of the output layer z and the teacher value becomes small. Then, by repeating the calculation of the output value for various input values and the update of the parameter, a parameter in which the feature of the input value is extracted is generated.

  FIG. 3 is a block diagram illustrating an example of a device configuration and a functional configuration of the project management support device 1. The project management support device 1 is a general computer, and includes a communication I / F 11, a storage device 12, an input / output device 13, a processor 14, and a bus 15.

  The communication I / F 11 is a network card, for example, and communicates with other computers based on a predetermined protocol. The project management support apparatus 1 may be a so-called cloud system that receives a processing request from another computer, performs processing according to the present embodiment, and transmits the result to the other computer. The project management support device 1 may be a device that operates stand-alone, and in this case, the communication I / F 11 may not be provided.

  The storage device 12 includes a main storage device such as a RAM (Random Access Memory) and a ROM (Read Only Memory), and an auxiliary storage device (secondary storage) such as an HDD (Hard-disk Drive), an SSD (Solid State Drive), and a flash memory. Device). The main storage device temporarily stores programs and data read by the processor and secures a work area for the processor. The auxiliary storage device stores a program executed by the processor, information to be learned in the present embodiment, a learned model, and the like.

  The input / output device 13 is, for example, an input device such as a keyboard and a mouse, an output device such as a monitor, and a touch panel. For example, the user may input information representing the degree estimated by the user with respect to the scale of the project for system development or the quality required for the system to be developed via the input / output device 13.

The processor 14 is an arithmetic processing unit such as a CPU (Central Processing Unit), and performs each process according to the present embodiment by executing a program according to the present embodiment. In the example of FIG. 3, functional blocks are shown in the processor 14. 3 includes an input unit 141, a conversion unit 142, a learning unit 143, and an estimation unit 144.

The input unit 141 acquires information to be input to the self encoder and inputs the information to the self encoder. The conversion unit 142 performs conversion processing to unify the format of input information. The learning unit 143 performs a predetermined calculation by the self-encoder and adjusts the parameters so that the output of the self-encoder approaches the input information. The estimation unit 144 uses the learned model generated by the learning unit 143 to correct the man-hours required to execute the project input by the user and the target value of the quality required for the system. Calculate an estimate.

  The components as described above are connected via the bus 15.

<Process flow>
FIG. 4 is a diagram for explaining a rough flow from learning the characteristics of the project performance to operation. The processing performed by the system according to the present embodiment includes a learning stage (FIG. 4: S1), an evaluation stage (S2), and an operation stage (S3). In the learning stage, information (also referred to as “training data”) representing the scale of the project or the quality required for the system stored in advance in the storage device 12 is read out, and the information is recorded using the self-encoder described above Machine learning of features. In the evaluation stage, the performance of the learned model generated in the learning stage is verified. Specifically, out of the information representing the scale of the project or the quality performance required for the system, information different from the training data used in the learning stage is used, or the input and output when the training data is processed is switched. By using it, it is tested whether an output satisfying a predetermined evaluation criterion can be obtained even for data different from the training data. In the evaluation stage, it is used in parallel with the ongoing project, and it is determined whether a high-accuracy output can be obtained compared with the scale of the project created by the user or the quality estimate required for the system. May be. In the operation stage, for an unknown project, a process for correcting the scale of the project created by the user or the quality estimate required for the system is performed.

<Learning process>
FIG. 5 is a process flow diagram illustrating an example of a learning process executed in the learning stage and the evaluation stage of the present embodiment. The input unit 141 of the project management support apparatus 1 acquires past project performance information stored in the storage device 12 in advance (FIG. 5: S11). In this step, for example, the number of man-hours for each type indicating the role of personnel stored in association with the time of the project, and information indicating the quality of the system evaluated according to a predetermined evaluation standard are acquired.

FIG. 6 is a diagram illustrating an example of information indicating the performance of a project. The table in FIG. 6 includes “case number”, “case name”, “function”, “application”, “characteristic”, “sub-characteristic”, “measurement item”, “level”, “start date”, “ It has the attributes of “end date”, “total man-hour”, “personnel”, and “man-hour”. In the “item number” field, identification information for identifying each item is registered in the form of a serial number. The name of each item is registered in the “Item name” field. In the “Function” field, a value for specifying whether each record is information used for the function of estimating the “quality” or the function of creating the “estimation” among the functions provided in the system. Is registered. In the example of FIG. 6, a value obtained by subdividing whether each information is simple information or detailed information is registered. In the “application work” field, information indicating the type of work to which the system related to each case is applied is registered. Examples of business types include customer management, supply chain management (SCM), EC (Electronic Commerce) and We.
b. Businesses to which the system is typically applied in system development, such as IT consulting, personnel / salary, accounting, and information infrastructure, are registered. In the “characteristic” field, a characteristic serving as a quality evaluation standard is registered. In the “sub-characteristic” field, information obtained by subdividing characteristics that are quality evaluation criteria is registered. In the “measurement item” field, measurable items that serve as objective criteria for evaluating the degree of the above-described characteristics and sub-characteristics are registered. That is, the information indicating the quality required for the system may be obtained from quantitative data that can be measured, for example. Note that the contents of the measurement item are determined in advance in association with at least one of the characteristic and the sub-characteristic. In the “level” field, information indicating the degree of each evaluation criterion is registered with a predetermined level of detail. In the “start date” and “end date” fields, the start and end of the period when the man-hours applied to the project are recorded for each predetermined period are registered. In the “total man-hour” field, the total man-hour required for the period is registered. In the “personnel” field, information indicating the type when man-hours are recorded for each type representing the role of the personnel is registered. In the “man-hour” field, the man-hour required for the period is registered.

  The table shown in FIG. 6 is an example of a configuration in which information is stored in a database table. The illustrated information or similar information is normalized to an arbitrary data structure in the storage device 12. Can be retained.

  Further, the conversion unit 142 of the project management support apparatus 1 determines whether to convert the format of the acquired information (FIG. 5: S12). In the present embodiment, when machine learning is performed using information representing past project performance, the format of input information is used in a unified manner with predetermined items and details. In this step, for example, simple information and detailed information included in the table of FIG. 6 are converted into any degree of detail.

  FIG. 7 is a diagram illustrating an example of a conversion table used for format conversion. In the example of FIG. 7, information indicating a correspondence relationship before and after conversion for converting detailed information into simple information is registered. Specifically, a detailed level represented by a ratio of 0 to 100% is divided at a predetermined position, and converted into any one of three levels of low, medium, and high previously associated with each range. Note that ratio values corresponding to low, medium, and high may be determined in advance so that simple information can be converted into detailed information.

  As for the man-hours, for example, information aggregated for each month and for each occupation type of personnel may be converted into the total man-hours of all personnel for all periods. Conversely, monthly information obtained by averaging the total man-hour information over the project period may be calculated.

  In S12 of FIG. 5, if the input information is not unified to a predetermined level of detail, it is determined to convert the format (S12: YES), and conversion is performed using, for example, a conversion table (S13). If it is determined in S12 that the format is not to be converted (S12: NO), or after S13, the learning unit 143 of the project management support apparatus 1 performs a compression process using a self-encoder (S14).

In this step, the above-described input information is input to the neural network having the configuration shown in FIG. 2 to learn its features. That is, the input information on the quality shown in FIG. 6 is assigned, for example, a unit of the input layer of the self-encoder for each characteristic or sub-characteristic, and information indicating the level for each characteristic or sub-characteristic is input to each unit Is done. Note that the input information represented by order scales such as high, medium, and low may be expressed as a bit string in the one-hot format, for example, and input to a plurality of units corresponding to each bit. Further, as the input information for the estimation shown in FIG. 6, information indicating the man-hour is input to the unit of the input layer.

Further, the compression processing may be learned for each application of the case. For example, a learned model may be generated for each type registered in the application business field of FIG. 6, or information represented by a nominal scale such as a type registered in the application business field may be For example, it may be expressed as a bit string in the One-hot format and input to a plurality of units corresponding to each bit. By learning the characteristics for each application, it is possible to extract the man-hours and the trend of required quality that are characteristic for the development of the system used for each operation, and to improve the output accuracy.

  Then, the parameters of the weights and bias terms are updated so that the value output from each unit of the output layer approaches the value input to the corresponding unit of the input layer, and the process is repeated with a plurality of records in FIG. 6 as input. .

  Thereafter, the learning unit 143 performs an evaluation process on the created learned model (S15). In this step, it is verified whether the learned model generated by the compression process has learned the characteristics of the quality of the project scale or the system. Also in this step, encoding and decoding are performed on the input information of the evaluation process using a self-encoder similar to the compression process. As the input information for the evaluation process, information that is not used in the compression process among the information indicating the project scale or the quality results required for the system, or the output information obtained as a result of processing the input information in the compression process Is used. Thus, if the value output from the output layer is equal to or greater than the predetermined reference and close to the input information, it can be determined that the created learned model has learned the characteristics of the quality of the project or the system required. Further, evaluation processing may be performed in parallel with the ongoing project, and information indicating the scale of the project created by the user or the quality required for the system may be used as input information. In this case, if the value output by the output layer is close to a predetermined standard or higher than the number of man-hours required to perform the actual project and the quality that the system should have, the created learned model will be stored in the project scale or system. It can be judged that the required quality characteristics have been learned. If it is determined that the performance of the learned model is insufficient, the process of S13 may be performed again by changing the input information.

  Then, the learning unit 143 stores the created learned model in the storage device 12 and ends the learning process.

<Operation processing>
FIG. 8 is a process flow diagram showing an example of the operation process executed in the operation stage. The input unit 141 of the project management support apparatus 1 acquires the function selected by the user based on the user's operation (FIG. 8: S21). In this step, the user is allowed to select either the function for outputting the estimated man-hour of the project or the function for outputting the quality required for the system.

  Next, the input unit 141 acquires the application selected by the user based on the user's operation (S22). In this step, the user is allowed to select a task to which the system to be developed in the project is applied from predetermined options. Options include typical operations such as customer management, supply chain management, EC / Web, IT consulting, personnel / salary, accounting, information infrastructure, and the like.

  Thereafter, the input unit 141 acquires information indicating the man-hour estimated by the user according to the function selected in S21 or the degree of quality estimated that the user is required for the system based on the user's operation (S23). . Also in this step, simple information or detailed information is input according to the user's selection.

9 and 10 are diagrams illustrating an example of information indicating the scale of the project input in S23. The table in FIG. 9 has attributes of “requirement”, “difficulty”, “project (PJ) period”, and “total man-hour”. In the “requirement” field, “supply chain management (SCM)”, which is an example of information indicating work to which the system is applied, is registered. In the “difficulty level” field, information indicating the difficulty level evaluated by the user based on the specific contents of the project is registered in stages such as high, medium, and low. In the “PJ period” field, “5 months”, which is an example of a period required to complete the project, is registered. In the “total man-hours” field, “28 man-months”, which is an example of man-hours required for executing the project, is registered.

  The table in FIG. 10 represents the scale of the project in more detail than the table shown in FIG. In the table of FIG. 10, the number of personnel assigned to the project for each job type is registered for the year and month when the project is executed. As job types, FIG. 5 illustrates a project manager (PM), a system engineer (SE), and a programmer (PG).

  When the man-hour is selected in S21, the input information as described above is received from the user via a predetermined input screen and stored in the storage device 12. The input information may be stored for each project, for example, so that it can be read out or further modified.

  11 and 12 are diagrams illustrating an example of information indicating quality required for the system, which is input in S23. 11 and 12 is based on a quality model defined in ISO / IEC25010. The table in FIG. 11 has attributes of “characteristic” and “level”. In the “characteristic” field, quality characteristics such as “functional compatibility”, “performance efficiency”, “compatibility”, and “usability” are registered. In the “level” field, information indicating the level of quality required for each quality characteristic, such as “high”, “medium”, and “low”, is registered.

  The example of FIG. 12 has attributes of “characteristic”, “subcharacteristic”, and “target”. In the “characteristic” field, quality characteristics such as “functional compatibility”, “performance efficiency”, and “compatibility” are registered. The “sub-characteristic” fields include “functional integrity”, “accuracy”, “appropriateness”, “time efficiency”, “resource availability”, “capacity”, “coexistence”, etc. Sub-characteristics that detail each characteristic are registered. In the “level” field, information indicating the degree of quality required for each quality characteristic in the form of a ratio is registered.

  When quality is selected in S21, the above input information is received from the user via a predetermined input screen and stored in the storage device 12.

  Further, the conversion unit 142 of the project management support apparatus 1 determines whether to convert the format of the acquired information (FIG. 8: S24). In this step, as in S12 of FIG. 5, if the input information is not unified with a predetermined level of detail, it is determined that the format is to be converted (S24: YES), and conversion is performed using, for example, a conversion table (S25). If it is determined in S24 that the format is not to be converted (S24: NO), or after S25, the learning unit 143 of the project management support apparatus 1 performs an expansion process using the learned model (S26). In the present embodiment, the following description will be given assuming that the input information is converted into the simple information shown in FIGS. 9 and 11 in S25.

In this step, the input information acquired in S23 is input to the neural network having the configuration shown in FIG. 2, and an output corrected based on the features extracted from past results is obtained. That is, in the input information for the estimation shown in FIG. 9, information indicating the difficulty level, the project period, and the man-hour is input to the unit of the input layer. Further, the processing of this step may be performed for each application. That is, similarly to the learning process, the process may be performed using the learned model generated for each application, or information indicating the application may be input to the unit of the input layer. Note that the “difficulty level” is not subject to learning or correction, and the man-hours finally output may be simply multiplied by a predetermined coefficient corresponding to the difficulty level in the operation stage. For example, for a project with a difficulty level of “high”, the man-hour output by the system is multiplied by 1.3. For projects with a medium difficulty level, the man-hour output by the system is multiplied by 1.0. For projects with a low difficulty level, the man-hour output by the system is multiplied by 0.8. Further, the “difficulty level” may also be a learning target, and the man-hour may be multiplied by a predetermined coefficient corresponding to the above-described difficulty level.

In addition, the input information on the quality shown in FIG. 11 is assigned, for example, a unit of the input layer of the self-encoder for each characteristic or sub-characteristic, and information indicating the level for each characteristic or sub-characteristic is input to each unit Is done. Note that level values represented by order scales such as high, medium, and low may be expressed by, for example, a one-hot format bit string and input to a plurality of units corresponding to each bit. Further, the processing of this step may be performed for each application. That is, similarly to the learning process, the process may be performed using the learned model generated for each application, or information indicating the application may be input to the unit of the input layer.

  Information output from the output layer of the self-encoder is calculated using the input information input as described above and the learned model stored in the storage device 12.

  Then, the estimation unit 144 of the project management support apparatus 1 is a value obtained by correcting by the self-encoder the information indicating the man-hour estimated by the user or the quality level estimated by the user to be required for the system, calculated in S26. Is output (S27). For example, corrected values are output for the same items as in the tables of FIGS. In this step, the values may be output in a table format as shown in FIGS. 9 to 12, or a graph such as a bar graph or a line graph may be drawn and output. In the output of quality, the ratio of the actual value to the level (target value) may be output as the achievement level based on the measurement value (actual value) of a predetermined measurement item.

  FIG. 13 is a diagram illustrating an output example of quality including the achievement level. The table of FIG. 13 has attributes of “characteristic”, “sub-characteristic”, “measurement item”, “level (target value)”, “actual value”, and “achievement level”. “Characteristics”, “Sub-characteristics”, and “Measurement items” are the same as those in FIG. In the “level (target value)” field, the corrected value calculated in S26 is registered. In the “actual value” field, a value measured for the measurement item is registered. That is, the system to be developed is operated, the turnaround time satisfaction rate, which is the rate at which the turnaround time satisfies a predetermined standard, the request processing rate when a predetermined load is applied, and the like are measured, and the result is registered. In the “achievement level” field, information indicating the degree of satisfaction of the required quality determined based on the ratio of the actual value to the target value is registered.

<Effect>
According to the system according to the present embodiment, it is possible to obtain a correction value based on the characteristics of a project performed in the past with respect to the value estimated by the user for the man-hour indicating the scale of system development and the quality required for the system to be developed. it can. As for the value input by the user, even if it is simple information, the system side converts it and unifies the format, so that the user can obtain a result with little effort. In other words, the user can receive corrections based on past performance for the estimated man-hours and target values for required quality, and even if the user does not have sufficient project management experience, a highly accurate estimate or target value can be obtained. Can be obtained.

<Modification>
The information input to the self-encoder may be image information. In the case of estimating man-hours, the image information may be represented by a bar graph or a line graph. Further, when creating a target value of required quality, a graph such as a bar graph, a line graph, or a radar chart may be represented. In this case, in the learning process, each of the pixels included in the image information may be input to each unit included in the input layer of the self-encoder illustrated in FIG.

  14 and 15 are diagrams illustrating an example of input information prepared in an image format. FIG. 14 is a bar graph showing the estimated man-hours for the project. FIG. 15 is a bar graph showing an example of the target value of the required quality. The information input to the input layer of the self-encoder shown in FIG. 2 may be such image information.

  Even when an image is used as input information, it is preferable to perform learning processing and operation processing for each type of work to which the system is applied. In this way, it is possible to generate a learned model in which features common to business are extracted. Further, the scales such as the vertical axis and the horizontal axis may be made to coincide for a plurality of graphs input to the self-encoder. This makes it easy to extract features having an absolute size between a plurality of graphs.

  Moreover, you may make it use for projects other than system development, such as architecture and product development. Even in this case, it is possible to learn the man-hours indicating the scale of the project and the quality required for the project deliverables such as buildings, plants, and some products in advance, and to correct the values entered by the user at the operation stage. . At this time, feature learning or input value correction may be performed for each use of the deliverable.

<Others>
The present invention includes a computer program that executes the above-described processing. Furthermore, a computer-readable recording medium on which the program is recorded also belongs to the category of the present invention. With respect to the recording medium on which the program is recorded, the above-described processing can be performed by causing the computer to read and execute the program on the recording medium.

  Here, the computer-readable recording medium refers to a recording medium in which information such as data and programs is accumulated by electrical, magnetic, optical, mechanical, or chemical action and can be read from a computer. Examples of such a recording medium that can be removed from the computer include a flexible disk, a magneto-optical disk, an optical disk, a magnetic tape, and a memory card. Further, examples of the recording medium fixed to the computer include a hard disk drive and a ROM.

1: Project management support device 11: Communication I / F
12: storage device 13: input / output device 14: processor 141: input unit 142: conversion unit 143: learning unit 144: estimation unit 15: bus

Claims (6)

An input unit that receives input of information indicating the degree of input by the user and information indicating the type of use of the deliverable, regarding the scale of the project or the quality required for the deliverable of the project,
A feature of information indicating the scale of the project or the quality results required for the deliverables of the project is learned using a self-encoder using a neural network and information indicating the type of use of the deliverables. Using the learned model, correcting the information representing the degree input by the user, and generating an information representing the scale of the project or the estimated value of the quality required for the deliverable of the project,
A project management support device comprising: The project management support apparatus according to claim 1, further comprising a conversion unit that converts information representing a degree input by the user into a predetermined level of detail. The project management support apparatus according to claim 1, wherein the information indicating the degree estimated by the user and the information indicating the actual result are information in a predetermined image format. The project management support apparatus according to claim 3, wherein the information of the predetermined image format is image data of a graph drawn at a predetermined scale. Regarding the scale of the project or the quality required for the deliverables of the project, information indicating the degree input by the user and information indicating the type of use of the deliverables are received.
Learning in which features of information indicating the scale of the project or the quality results required for the deliverable are learned using a self-encoder using a neural network and information indicating the type of use of the deliverable Using a completed model, modifying the information representing the degree input by the user, and generating information representing an estimate of the quality of the project or the quality required for the deliverable of the project,
A project management support method in which processing is executed by a computer. Regarding the scale of the project or the quality required for the deliverables of the project, information indicating the degree input by the user and information indicating the type of use of the deliverables are received.
Learning in which features of information indicating the scale of the project or the quality results required for the deliverable are learned using a self-encoder using a neural network and information indicating the type of use of the deliverable Using a completed model, modifying the information representing the degree input by the user, and generating information representing an estimate of the quality of the project or the quality required for the deliverable of the project,
A program that causes a computer to execute processing.

Images