Data analysis method, data analysis system, and server for data analysis system

Abstract

To provide a data analysis method allowing even a user with a low level of skill to easily generate a learned model, and allowing a reduction in time required for analysis of measurement data using the learned model.SOLUTION: This data analysis method includes: a storage step of storing in advance a plurality of algorithm groups for learning 40 in which a type 33a of measurement data 30, a type 33b of analysis of the measurement data 30, and an algorithm for learning 40a are associated with each other; an algorithm for learning selection step of selecting the algorithm for learning 40a used for learning based on information on input information 33 including the type 33a of the measurement data 30 and information on the type 33b of analysis of the measurement data 30; a learned model generation step of generating a learned model 34 based on teacher data 32 and the selected algorithm for learning 40a; and an analysis result acquisition step of analyzing the measurement data 30 based on the learned model 34 to acquire an analysis result 31.SELECTED DRAWING: Figure 15

Description

Technical Field

[0001] The present invention relates to a data analysis method, a data analysis system, and a server for a data analysis system, and particularly to a data analysis method, a data analysis system, and a server for a data analysis system that generate a learned model and analyze measurement data using the generated learned model.

Background Art

[0002] Conventionally, a data analysis method that generates a learned model and analyzes measurement data using the generated learned model is known (for example, see Patent Document 1).

[0003] Patent Document 1 discloses a cell image analysis method having a learning model generation step of generating a learning model by performing machine learning, and a region estimation step of using the learning model, inputting a phase image of a cell to be analyzed as an input image, and outputting a cell region estimation image indicating a cell region.

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0005] Although not disclosed in Patent Document 1, generating a learning model (a pre-trained model) for analyzing measurement data such as cell region estimation images requires specialized knowledge, including the selection of a learning algorithm. Therefore, it is difficult for less experienced users to generate a pre-trained model. Thus, less experienced users may consider outsourcing the generation of pre-trained models. However, outsourcing the generation of pre-trained models has the disadvantage that it takes time to generate the pre-trained model, and therefore, it takes time to analyze the measurement data using the pre-trained model. Therefore, there is a need for a data analysis method that allows even less experienced users to easily generate a pre-trained model and that can shorten the time required to analyze measurement data using the pre-trained model.

[0006] This invention was made to solve the above-mentioned problems, and one of its objectives is to provide a data analysis method, a data analysis system, and a server for a data analysis system that enable even inexperienced users to easily generate trained models and shorten the time required for analyzing measurement data using the trained models. [Means for solving the problem]

[0007] The data analysis method in the first aspect of this invention is a data analysis method performed between a system management company and a customer who desires to obtain the results of analysis of measurement data acquired from a measuring device, and includes a storage step of pre-storing a group of learning algorithms that associate the type of measurement data with the type of analysis of the measurement data and the learning algorithm; a training data reception step of receiving training data input; and information regarding the type of measurement data and information regarding the type of analysis of the measurement data. For learning Accepts input information. For learning Input information reception process, For learningA learning algorithm selection step involves selecting a learning algorithm from a group of learning algorithms based on input information, and a trained model generation step involves generating a trained model based on training data and the selected learning algorithm. A trained model storage step that stores the trained model generated in the trained model generation step, the type of measurement data and the type of analysis of the measurement data received in the training input information receiving step, in association with each other; an analysis receiving step that receives input of analysis input information including information on the type of measurement data and information on the type of analysis of the measurement data, and input of measurement data acquired by a measurement device; and a trained model selection step that selects a trained model to be used for the analysis of the measurement data based on the analysis input information received in the analysis receiving step. The system includes an analysis result acquisition step that analyzes measurement data based on a trained model and obtains the analysis results. In the analysis result acquisition process, measurement data is input to the trained model selected in the trained model selection process, and the analysis results are obtained. ru.

[0008] The data analysis system in the second aspect of this invention is a data analysis system for performing data analysis between a system management company and a customer who desires to obtain the results of analysis of measurement data acquired from a measuring device, comprising: a server that generates a trained model for analyzing measurement data; and a data processing device that requests the server to analyze the measurement data, wherein the server includes a storage unit that pre-stores a plurality of training algorithms that associate the type of measurement data, the type of analysis of the measurement data, and the training algorithm; and includes information regarding the type of measurement data and information regarding the type of analysis of the measurement data. For learning Inputting input information and inputting training data Input of analysis input information including information on the type of measurement data and information on the type of analysis of the measurement data, and input of measurement data acquired by the measurement device, An input reception unit that accepts inputs, For learning A learning algorithm selection unit that determines which learning algorithm to use for training from a group of learning algorithms based on input information, and a trained model generation unit that generates a trained model based on training data and the determined learning algorithm. A trained model selection unit selects a trained model to be used for analyzing measurement data based on the analysis input information received via the input reception unit. The system includes an analysis result acquisition unit that analyzes measurement data based on a trained model generated by a trained model generation unit and acquires the analysis results. The memory unit stores the generated trained model in association with the type of measurement data received as training input information and the type of analysis of the measurement data, and the analysis result acquisition unit inputs the measurement data into the selected trained model and acquires the analysis result. ru.

[0009] A data analysis system server in the third aspect of this invention is a data analysis system server for performing data analysis between a system management company and a customer who desires to obtain the results of analysis of measurement data acquired from a measuring device, and includes a storage unit that pre-stores a plurality of learning algorithms that associate the type of measurement data, the type of analysis of the measurement data, and the learning algorithm, and includes information regarding the type of measurement data and information regarding the type of analysis of the measurement data. For learning Inputting input information and inputting training data Input of analysis input information including information on the type of measurement data and information on the type of analysis of the measurement data, and input of measurement data acquired by the measurement device, An input reception unit that accepts inputs, For learning A learning algorithm selection unit that determines which learning algorithm to use for training from a group of learning algorithms based on input information, and a trained model generation unit that generates a trained model based on training data and the determined learning algorithm. A trained model selection unit selects a trained model to be used for analyzing measurement data based on the analysis input information received via the input reception unit. The system includes an analysis result acquisition unit that analyzes measurement data based on a trained model generated by a trained model generation unit and acquires the analysis results. The memory unit stores the generated trained model in association with the type of measurement data received as training input information and the type of analysis of the measurement data, and the analysis result acquisition unit inputs the measurement data into the selected trained model and acquires the analysis result. ru. [Effects of the Invention]

[0010] A data analysis method in the first aspect of the present invention includes a storage step of pre-storing a group of learning algorithms that associate the type of measurement data with the type of analysis of the measurement data and the learning algorithm, and information relating to the type of measurement data and information relating to the type of analysis of the measurement data. For learning Accepts input information. For learning Input information reception process, For learning A learning algorithm selection process that selects a learning algorithm to be used for learning from a group of learning algorithms based on the input information. A trained model generation step that generates a trained model based on training data and a determined learning algorithm; a trained model storage step that stores the trained model generated in the trained model generation step in association with the type of measurement data and the type of analysis of the measurement data received in the learning input information receiving step; an analysis receiving step that accepts input of analysis information including information on the type of measurement data and information on the type of analysis of the measurement data, and input of measurement data acquired by a measurement device; a trained model selection step that selects a trained model to be used for analyzing the measurement data based on the analysis input information received in the analysis receiving step; and an analysis result acquisition step that analyzes the measurement data based on the trained model and obtains the analysis results. Equipped with In the analysis result acquisition process, measurement data is input to the trained model selected in the trained model selection process, and the analysis results are obtained.Therefore, by simply inputting information about the type of measurement data and the type of analysis of the measurement data, the system selects a training algorithm from a group of training algorithms that is suitable for the trained model used to analyze the measurement data. Consequently, even if the user's skill level is low, a trained model can be easily generated without outsourcing. Furthermore, since it is possible to generate a trained model without outsourcing, the time required to generate the trained model can be reduced. As a result, it is possible to provide a data analysis method that allows even users with low skill levels to easily generate a trained model and reduces the time required to analyze measurement data using the trained model.

[0011] Furthermore, the data analysis system in the second aspect of the present invention and the server for the data analysis system in the third aspect include a storage unit that pre-stores a plurality of learning algorithms that associate the types of measurement data, the types of analysis of the measurement data, and the learning algorithms, and includes information regarding the types of measurement data and information regarding the types of analysis of the measurement data. For learning Inputting input information and inputting training data Input of analysis input information including information on the type of measurement data and information on the type of analysis of the measurement data, and input of measurement data acquired by the measurement device, An input reception unit that accepts inputs, For learning A learning algorithm selection unit that determines which learning algorithm to use for learning from among a group of learning algorithms based on the input information. A trained model generation unit generates a trained model based on training data and a determined learning algorithm, and a trained model selection unit selects a trained model to be used for analyzing measurement data based on analysis input information received via an input reception unit. Equipped with The memory unit stores the generated trained model in association with the type of measurement data received as training input information and the type of analysis of the measurement data, and the analysis result acquisition unit inputs the measurement data into the selected trained model and acquires the analysis result. This enables a data analysis system and a server for the data analysis system that, similar to the data analysis method in the first phase, allows even inexperienced users to easily generate trained models and reduces the time required for analyzing measurement data using the trained models. [Brief explanation of the drawing]

[0012] [Figure 1] This is a block diagram showing the overall configuration of a data analysis system according to one embodiment. [Figure 2]It is a block diagram showing the configuration of a server for data analysis according to an embodiment. [Figure 3] It is a block diagram showing the configuration of a data processing device according to an embodiment. [Figure 4] It is a schematic diagram for explaining a configuration for generating a learned model in a server and a data processing device according to an embodiment. [Figure 5] It is a schematic diagram for explaining an example of a screen when generating a learned model in a data processing device according to an embodiment. [Figure 6] It is a block diagram for explaining a group of learning algorithms. [Figure 7] [[ID=1​​​​​​​​​​​​​​​​​​​​​​​​​​​​​This is a flowchart to explain the process of obtaining analysis results. [Figure 18] This is a flowchart to explain the additional processing of the learning algorithm. [Figure 19] This is a flowchart illustrating the additional processing steps in a data processing program. [Figure 20] This is a block diagram illustrating a modified data processing system. [Modes for carrying out the invention]

[0013] The following describes embodiments of the present invention based on the drawings.

[0014] Referring to Figures 1 to 14, the configuration of the data analysis system 100 (see Figure 1) that executes the data analysis method according to this embodiment, and the server 1 for data analysis (see Figure 1) will be described. The data analysis method according to this embodiment is a data analysis method performed between a system management company and a customer who desires to obtain the analysis results 31 (see Figure 9) of measurement data 30 (see Figure 1) acquired from a measurement device 3 (see Figure 1). The data analysis system 100 according to this embodiment is a data analysis system for performing data analysis between a system management company and a customer who desires to obtain the analysis results 31 of measurement data 30 acquired from a measurement device 3. The system management company may be a manufacturer of the data analysis system 100, or a company that performs maintenance and inspection of the data analysis system 100 on a contract basis.

[0015] (Configuration of the data analysis system) As shown in Figure 1, the data analysis system 100 comprises a server 1 and a data processing device 2. The data processing device 2 is connected to a measuring device 3.

[0016] Server 1 is a server for a data analysis system that performs data analysis between a system management company and a customer who desires to obtain analysis results 31 of measurement data 30 acquired from a measuring device 3. Server 1 is also configured to generate a trained model 34 (see Figure 2) for analyzing the measurement data 30. In this embodiment, Server 1 is connected to the data processing device 2 via a network. The detailed configuration of Server 1 will be described later. Server 1 may be installed in the same facility as the data processing device 2; that is, the data analysis system 100 may be a cloud-based system. Alternatively, it may be installed in a different facility from the data processing device 2; that is, the data analysis system 100 may be an on-premise system. In this embodiment, the data analysis system 100 is an on-premise system.

[0017] The data processing device 2 is configured to request server 1 to analyze the measurement data 30. Specifically, the data processing device 2 is configured to send a control signal to server 1 to start the analysis of the measurement data 30. The data processing device 2 is connected to server 1 via a network. The data processing device 2 is also connected to measurement device 3 and is configured to acquire the measurement data 30 (described later) from measurement device 3 and send it to server 1. The data processing device 2 is installed within the customer's facility. The detailed configuration of the data processing device 2 will be described later.

[0018] The measuring device 3 is configured to acquire measurement data 30. The measuring device 3 is also connected to the data processing device 2. The measuring device 3 is configured to transmit the acquired measurement data 30 to the data processing device 2. The measuring device 3 includes a measuring device for acquiring the measurement data 30. The measuring device 3 may include, for example, an X-ray imaging device, a CT (Computed Tomography) device, a device for measuring particle size distribution, a surface measurement device, a microscope, an LC (Liquid Chromatography) device, and a GC (Gas Chromatography) device. The measuring device 3 is installed within the customer's facility.

[0019] Measurement data 30 is data acquired by the measuring device 3. For example, if the measuring device 3 is an X-ray imaging device, the measurement data 30 is an X-ray image. Alternatively, if the measuring device 3 is a CT scanner, the measurement data 30 is a tomographic image. Furthermore, if the measuring device 3 is a device that measures particle size distribution, the measurement data 30 is data on the particle size distribution. Also, if the measuring device 3 is a surface measuring device, the measurement data 30 is data showing the surface shape of the object being measured. Furthermore, if the measuring device 3 is a microscope, the measurement data 30 is an image (e.g., a cell image) taken by the microscope. Finally, if the measuring device 3 is an LC or GC, the measurement data 30 is a chromatogram.

[0020] As shown in Figure 2, the server 1 comprises a first storage unit 10, an input receiving unit 11, and a first processor 12.

[0021] Server 1 is configured to perform the process of generating a trained model 34 (the process of performing machine learning). Server 1 also has the capability to store (storage) various data used in generating the trained model 34. Furthermore, Server 1 has the capability to analyze the measurement data 30 (see Figure 1) using the trained model 34.

[0022] The first storage unit 10 is configured to store the trained model 34. The first storage unit 10 is also configured to store the training algorithm group 40. Furthermore, the first storage unit 10 is configured to store the data processing program group 41. Furthermore, the first storage unit 10 is configured to store the analysis conditions 42. Furthermore, the first storage unit 10 is configured to store various programs executed by the first processor 12. In this embodiment, the first storage unit 10 pre-stores the training algorithm group 40, the data processing program group 41, and the various programs executed by the first processor 12 when the manufacturing of the data analysis system 100 is completed. The first storage unit 10 also retrospectively stores the trained model 34 and the analysis conditions 42. The first storage unit 10 is, for example, a non-volatile storage device such as an HDD (Hard Disk Drive) or an SSD (Solid State Drive). Note that the first storage unit 10 is an example of the "storage unit" in the claims. Furthermore, details of the trained model 34, the training algorithms 40, the data processing programs 41, and the analysis conditions 42 will be described later.

[0023] The input receiving unit 11 is configured to receive input information 33 (see Figure 4), which will be described later, from the data processing device 2. The input receiving unit 11 is, for example, an input / output interface.

[0024] The first processor 12 is configured to perform various controls on the server 1. The first processor 12 is also configured to generate a trained model 34. Furthermore, the first processor 12 is configured to analyze measurement data 30 (see Figure 1) and obtain analysis results 31 (see Figure 9). The first processor 12 includes a CPU (Central Processing Unit), ROM (Read Only Memory), RAM (Random Access Memory), GPU (Graphics Processing Unit), or an FPGA (Field-Programmable Gate Array) configured for image processing. Note that the first processor 12 may include circuitry instead of a CPU.

[0025] Furthermore, the first processor 12 includes, as functional blocks of software (programs), a learning algorithm selection unit 12a, a trained model generation unit 12b, and an analysis result acquisition unit 12c. The first processor 12 also includes a trained model storage control unit 12d and a trained model selection unit 12e. The first processor 12 functions as the learning algorithm selection unit 12a, the trained model generation unit 12b, the analysis result acquisition unit 12c, the trained model storage control unit 12d, and the trained model selection unit 12e by executing various programs stored in the first storage unit 10. The learning algorithm selection unit 12a, the trained model generation unit 12b, the analysis result acquisition unit 12c, the trained model storage control unit 12d, and the trained model selection unit 12e may be individually configured as hardware by dedicated processors (processing circuits).

[0026] Details of the functions of the learning algorithm selection unit 12a, the trained model generation unit 12b, the analysis result acquisition unit 12c, the trained model storage control unit 12d, and the trained model selection unit 12e will be described later.

[0027] As shown in Figure 3, the data processing device 2 comprises a second storage unit 20 and a second processor 21. The data processing device 2 also comprises an input device 22 and a display unit 23. The data processing device 2 is configured to send a control signal to the server 1 (see Figure 2) to start generating a trained model 34 (see Figure 2) based on the operator's input. The data processing device 2 is also configured to send a control signal to the server 1 to start analyzing the measurement data 30 (see Figure 1) based on the operator's input.

[0028] The second storage unit 20 stores various programs executed by the second processor 21. The second storage unit 20 is, for example, a non-volatile storage device such as an HDD or an SSD.

[0029] The second processor 21 is configured to function as a control unit that performs various controls on the data processing device 2 by executing various programs stored in the second memory unit 20. The second processor 21 includes a CPU, ROM, RAM, GPU, or FPGA configured for image processing. The second processor 21 may also include circuitry instead of a CPU.

[0030] Input device 22 is an input device used by the operator to perform operations. Input device 22 includes, for example, a mouse and a keyboard.

[0031] The display unit 23 is configured to display the analysis results 31 (see Figure 9) of the measurement data 30 (see Figure 1), etc. The display unit 23 is a display device such as a liquid crystal monitor or an organic EL (Electro Luminescence) monitor.

[0032] Next, with reference to Figures 4 to 14, the data analysis method performed by the data analysis system 100 (see Figure 1) according to this embodiment will be described. The data analysis method performed by the data analysis system 100 is broadly divided into two parts: a method for generating a trained model 34 (see Figure 2), and a method for analyzing measurement data 30 (see Figure 1) and obtaining analysis results 31 (see Figure 9).

[0033] (Generating a pre-trained model) First, referring to Figures 4 to 8, we will explain how the data analysis system 100 (see Figure 1) generates the trained model 34 (see Figure 2).

[0034] A method for generating a trained model 34 includes at least a training data reception step for receiving training data 32, an input information reception step for receiving input information 33, and a step for receiving training data 32 (see Figure 8).

[0035] The training data reception process receives the input of training data 32. In the process of receiving the input of training data 32, the server 1 (see Figure 4) receives the input of training data 32 from the data processing device 2 (see Figure 4). The server 1 also stores the input training data 32 in the first storage unit 10 (see Figure 2). The training data reception process may be performed before the input information reception process, which receives the input of input information 33, or it may be performed together with the input information reception process. That is, the training data 32 may be input to the server 1 before the input information 33 is input, or it may be input to the server 1 together with the input information 33. In this embodiment, the training data reception process is performed before the input information reception process. That is, the training data 32 and the input information 33 are input to the server 1 at different timings. The training data 32 is the data used when training the learning algorithm 40a and generating the trained model 34. The learning algorithm 40a is the program used when generating the trained model 34. Learning algorithms 40a include, for example, linear regression, random forests, neural networks, and support vector machines.

[0036] The input information receiving process receives input information 33, which includes information regarding the type 33a of the measurement data 30 and information regarding the type 33b of the analysis of the measurement data 30. As shown in Figure 4, the server 1 receives input information 33 from the data processing device 2. The information regarding the type 33a of the measurement data 30 indicates what kind of data the measurement data 30 is, for example, "cell image" or "X-ray image". The information regarding the type 33b of the analysis of the measurement data 30 indicates the type of analysis to be performed on the measurement data 30, for example, "shape analysis" or "quality judgment".

[0037] Furthermore, in the example shown in Figure 4, the information 32a of the training data 32 is input to the server 1 along with the input information 33. The information 32a of the training data 32 is the information selected by the operator to determine which training data 32 to use in training the trained model 34.

[0038] The example shown in Figure 5 is screen 50, where the operator performs the process of generating a trained model 34 (see Figure 2) in the data processing device 2 (see Figure 4).

[0039] The screen 50 for the process of generating the trained model 34 is displayed on the display unit 23 (see Figure 3). The screen 50 for the process of generating the trained model 34 displays a first selection field 50a, a second selection field 50b, a training data selection field 50c, a generate button 50d, and a cancel button 50e.

[0040] The first selection field 50a is a selection field in which the type 33a of measurement data 30 (see Figure 4) is selected. The first selection field 50a is, for example, a dropdown selection field. The options displayed in the first selection field 50a are stored in the first storage unit 10 (see Figure 2). The first processor 12 (see Figure 2) retrieves the type 33a of measurement data 30 stored in the first storage unit 10 and displays it as an option in the first selection field 50a.

[0041] The second selection field 50b is a selection field in which the type of analysis 33b (see Figure 4) of the measurement data 30 is selected. The second selection field 50b is, for example, a dropdown selection field. The options displayed in the second selection field 50b are stored in the first storage unit 10. The first processor 12 retrieves the type of analysis 33b of the measurement data 30 stored in the first storage unit 10 and displays it as an option in the second selection field 50b.

[0042] The training data selection field 50c is a selection field for selecting training data 32 (see Figure 8) stored in the first storage unit 10. The training data selection field 50c is, for example, a dropdown selection field. The options displayed in the training data selection field 50c are stored in the first storage unit 10. The first processor 12 obtains the name of the training data 32 stored in the first storage unit 10 and displays it as an option in the training data selection field 50c.

[0043] The generate button 50d is a push button on the GUI (Graphical User Interface) displayed on screen 50, which performs the process of generating the trained model 34. When the generate button 50d is pressed, the input information 33 (see Figure 4) and training data 32 are sent from the data processing device 2 to the server 1 (see Figure 4), and the process of generating the trained model 34 is started on the server 1.

[0044] The cancel button 50e is a GUI push button displayed on screen 50, which is used to perform the process of generating the trained model 34. If the cancel button 50e is pressed, the trained model 34 will not be generated, and screen 50, which is used to perform the process of generating the trained model 34, will be closed.

[0045] Here, the trained model 34 is generated by training a training algorithm 40a (see Figure 6), which is used to generate a trained model 34 suitable for analyzing the measurement data 30, using the training data 32. Selecting the training algorithm 40a requires specialized knowledge. Therefore, it is difficult for less experienced users to generate a trained model 34 suitable for analyzing the measurement data 30.

[0046] Therefore, in this embodiment, when generating a trained model 34, the server 1 selects an appropriate training algorithm 40a from the training algorithm group 40 (see Figure 6) according to the measurement data 30 to be analyzed.

[0047] The learning algorithm group 40 includes multiple learning algorithms 40a. In this embodiment, as shown in Figure 6, the learning algorithm group 40 is stored in the first storage unit 10 (see Figure 2) in association with the type 33a of measurement data 30 (see Figure 1), the type 33b of analysis of measurement data 30, and the learning algorithm 40a. The learning algorithm group 40 is pre-stored in the first storage unit 10 in association with the type 33a of measurement data 30, the type 33b of analysis of measurement data 30, and the learning algorithm 40a. In this embodiment, the first storage unit 10 pre-stores multiple learning algorithm groups 40 for each type 33a of measurement data 30 and type 33b of analysis of measurement data 30. In this embodiment, the first storage unit 10 stores multiple learning algorithm groups 40 as presets. In other words, the first memory unit 10 pre-stores a group of learning algorithms 40 when the manufacturing of the data analysis system 100 is completed.

[0048] As shown in the matrix diagram 60 of Figure 7, the learning algorithm group 40 is stored in a state where learning algorithms 40a are combined in any combination of multiple types 33a of measurement data 30 and multiple types 33b of analysis of measurement data 30. For example, for a combination where the type 33a of measurement data 30 is "b" and the type 33b of analysis of measurement data 30 is "α", learning algorithm 40a is associated with "B". Therefore, by selecting the type 33a of measurement data 30 and the type 33b of analysis of measurement data 30, the operator can select an appropriate learning algorithm 40a from among the multiple learning algorithm groups 40. Note that in the example shown in Figure 7, in the matrix diagram 60, no learning algorithm 40a is associated with any combination of the type 33a of measurement data 30 and the type 33b of analysis of measurement data 30. However, the first storage unit 10 may store a learning algorithm 40a associated with each of the all combinations of the type 33a of the measurement data 30 and the type 33b of the analysis of the measurement data 30.

[0049] Furthermore, in this embodiment, in addition to the pre-stored (pre-set) learning algorithms 40a, other learning algorithms 40a can be added. Specifically, the first processor 12 (see Figure 2) is configured to accept the addition of other learning algorithms 40a associated with the types 33a of measurement data 30 and the types 33b of analysis of measurement data 30 to the multiple pre-stored learning algorithm groups 40. In other words, the server 1 is configured to allow the addition (plugging in) of other learning algorithms 40a to the multiple pre-stored learning algorithm groups 40.

[0050] Next, referring to Figure 8, we will describe the process by which the first processor 12 generates the trained model 34.

[0051] As shown in Figure 8, the input receiving unit 11 receives input information 33 from the data processing device 2, which includes information regarding the type 33a of the measurement data 30 (see Figure 1) and information regarding the type 33b of the analysis of the measurement data 30. The input receiving unit 11 outputs the input information 33 to the learning algorithm selection unit 12a and the trained model storage control unit 12d.

[0052] Furthermore, the input receiving unit 11 receives information 32a from the training data 32. The input receiving unit 11 outputs information 32a from the training data 32 to the trained model generation unit 12b.

[0053] The learning algorithm selection unit 12a selects a learning algorithm 40a to be used for learning from among a group of learning algorithms 40 based on the input information 33. Specifically, the learning algorithm selection unit 12a selects a learning algorithm 40a from among the group of learning algorithms 40 that matches the information regarding the type 33a of the measurement data 30 and the information regarding the type 33b of the analysis of the measurement data 30. In this embodiment, the learning algorithm selection unit 12a selects one learning algorithm 40a that matches the information regarding the type 33a of the measurement data 30 and the information regarding the type 33b of the analysis of the measurement data 30. The learning algorithm selection unit 12a outputs the selected learning algorithm 40a to the trained model generation unit 12b.

[0054] The trained model generation unit 12b generates a trained model 34 based on the training data 32 and the selected training algorithm 40a. In this embodiment, the trained model generation unit 12b reads the training data 32 from the first storage unit 10 based on the information 32a of the training data 32. Then, the trained model generation unit 12b generates a trained model 34 based on the training algorithm 40a input from the training algorithm selection unit 12a and the training data 32 read from the first storage unit 10. The trained model generation unit 12b also outputs the generated trained model 34 to the trained model storage control unit 12d.

[0055] The trained model storage control unit 12d stores the generated trained model 34 in association with the type 33a of the measurement data 30 and the type 33b of the analysis of the measurement data 30. In this embodiment, the trained model storage control unit 12d stores the trained model 34 in association with the type 33a of the measurement data 30 and the type 33b of the analysis of the measurement data 30.

[0056] (Analysis of measurement data) Next, referring to Figures 9 to 14, we will explain how the data analysis system 100 (see Figure 1) analyzes the measurement data 30 (see Figure 9).

[0057] As shown in Figure 9, the method for analyzing measurement data 30 includes a step in which the server 1 acquires the measurement data 30 and input information 33, which includes the type of measurement data 30 33a and the type of analysis of the measurement data 30 33b, from the data processing device 2. The method for analyzing measurement data 30 also includes an analysis result acquisition step in which the measurement data 30 is analyzed based on a trained model 34 and an analysis result 31 is obtained. The analysis result 31 may be data in various formats depending on the type of measurement data 30 33a and the type of analysis of the measurement data 30 33b. For example, the analysis result 31 may be image data. Alternatively, the analysis result 31 may be numerical data.

[0058] The example shown in Figure 10 is screen 51, where the operator requests the data processing device 2 (see Figure 9) to perform analysis processing on the measurement data 30 (see Figure 9).

[0059] The screen 51 for requesting analysis processing of the measurement data 30 is displayed on the display unit 23 (see Figure 3). The screen 51 for requesting analysis processing of the measurement data 30 displays the first selection field 51a, the second selection field 51b, the measurement data selection field 51c, the analysis button 51d, and the cancel button 51e.

[0060] The first selection field 51a is a selection field in which the type 33a of measurement data 30 (see Figure 9) is selected. The first selection field 51a is, for example, a dropdown selection field. The options displayed in the first selection field 51a are stored in the first storage unit 10. The first processor 12 (see Figure 2) retrieves the type 33a of measurement data 30 stored in the first storage unit 10 and displays it as an option in the first selection field 51a.

[0061] The second selection field 51b is a selection field in which the type of analysis 33b (see Figure 9) of the measurement data 30 is selected. The second selection field 51b is, for example, a dropdown selection field. The options displayed in the second selection field 51b are stored in the first storage unit 10. The first processor 12 retrieves the type of analysis 33b of the measurement data 30 stored in the first storage unit 10 and displays it as an option in the second selection field 51b.

[0062] The measurement data selection field 51c is a selection field for selecting measurement data 30 (see Figure 9).

[0063] The analysis button 51d is a GUI push button displayed on screen 51, which is used to request the analysis of the measurement data 30. When the analysis button 51d is pressed, the data processing device 2 sends the input information 33 (see Figure 9) and the measurement data 30 to the server 1, and the server 1 starts the analysis of the measurement data 30 using the trained model 34.

[0064] The cancel button 51e is a GUI push button displayed on screen 51, which is used to request the analysis of the measurement data 30. When the cancel button 51e is pressed, the measurement data 30 is not sent to server 1, nor is the analysis of the measurement data 30 performed on server 1, and screen 51, which is used to request the analysis of the measurement data 30, is closed.

[0065] In this embodiment, when analyzing measurement data 30 using the trained model 34 (see Figure 2), a data processing program 41a (see Figure 11) can also be used. The data processing program 41a is a program that performs data processing different from that of the trained model 34. Specifically, the data processing program 41a includes at least one of the following: a program that performs preprocessing on the measurement data 30 before analysis by the trained model 34, and a program that performs postprocessing on the data after analysis by the trained model 34.

[0066] Therefore, for example, if the data processing program 41a includes a program for preprocessing, selecting the analysis condition 42 will cause the data processing program 41a to perform preprocessing on the measurement data 30. The preprocessed measurement data 30 will then be input into the trained model 34, and the analysis result 31 (see Figure 8) will be obtained. Also, for example, if the data processing program 41a includes a program for postprocessing, selecting the analysis condition 42 will cause the measurement data 30 to be input into the trained model 34, and the analysis result 31 (see Figure 8) will be obtained. The data processing program 41a will then perform postprocessing on the analysis result 31. If the data processing program 41a includes both a program for preprocessing and a program for postprocessing, both preprocessing on the measurement data 30 and postprocessing on the analysis result 31 will be performed.

[0067] As shown in Figure 11, in this embodiment, the data processing program 41a is pre-stored in the first storage unit 10 (see Figure 2) as a data processing program group 41, associated with the type 33a of the measurement data 30 (see Figure 9) and the type 33b of the analysis of the measurement data 30. In this embodiment, the first storage unit 10 pre-stores a plurality of data processing program groups 41. That is, the first storage unit 10 stores a plurality of data processing program groups 41 as presets. In other words, the first storage unit 10 pre-stores a plurality of data processing program groups 41 when the manufacturing of the data analysis system 100 is completed.

[0068] As shown in the matrix diagram 61 of Figure 12, the data processing program group 41 stores data processing programs 41a in any combination of multiple types 33a of measurement data 30 and multiple types 33b of analysis of measurement data 30. For example, for a combination where the type 33a of measurement data 30 is "b" and the type 33b of analysis of measurement data 30 is "α", the data processing program group 41 is associated with "K". Therefore, by selecting the type 33a of measurement data 30 and the type 33b of analysis of measurement data 30, the operator can select the appropriate data processing program 41a from among the multiple data processing program groups 41. Note that in the example shown in Figure 12, the data processing program 41a is not associated with any combination of the type 33a of measurement data 30 and the type 33b of analysis of measurement data 30 in the matrix diagram 61. However, the first storage unit 10 may store a data processing program 41a associated with each of the all combinations of the type 33a of the measurement data 30 and the type 33b of the analysis of the measurement data 30.

[0069] Furthermore, in this embodiment, in addition to the pre-stored (pre-set) data processing programs 41a, other data processing programs 41a can be added. Specifically, the first processor 12 (see Figure 2) is configured to accept the addition of other data processing programs 41a associated with the types 33a of measurement data 30 and the types 33b of analysis of measurement data 30 to the pre-stored data processing program group 41. In other words, the server 1 is configured to allow the addition (plugging in) of other data processing programs 41a to a plurality of pre-stored data processing program groups 41.

[0070] Furthermore, in this embodiment, the analysis result acquisition unit 12c (see Figure 2) is configured to analyze the measurement data 30 based on the analysis conditions 42 (see Figure 13) and acquire the analysis results 31 (see Figure 9).

[0071] As shown in Figure 13, the analysis conditions 42 include a trained model 34 and a data processing program 41a. Specifically, the analysis conditions 42 are stored in the first storage unit 10 (see Figure 2) in a manner that associates the trained model 34, the data processing program 41a, the type of measurement data 30 (see Figure 9) 33a, and the type of analysis 33b of the measurement data 30.

[0072] Next, referring to Figure 14, we will describe the configuration in which the first processor 12 acquires the analysis result 31.

[0073] As shown in Figure 14, the input receiving unit 11 receives input information 33 and measurement data 30 from the data processing device 2. The input receiving unit 11 outputs the input information 33 to the trained model selection unit 12e. The input receiving unit 11 also outputs the measurement data 30 to the analysis result acquisition unit 12c.

[0074] The trained model selection unit 12e selects a trained model 34 based on the input information 33. Specifically, the trained model selection unit 12e selects a trained model 34 to be used for analyzing the measurement data 30 based on the information of the type 33a of the measurement data 30 and the information of the type 33b of the analysis of the measurement data 30. In this embodiment, the trained model selection unit 12e selects one trained model 34 based on the information of the type 33a of the measurement data 30 and the information of the type 33b of the analysis of the measurement data 30.

[0075] Furthermore, in this embodiment, when the trained model selection unit 12e selects the trained model 34, it also selects the data processing program 41a. That is, the trained model selection unit 12e selects the trained model 34 and the data processing program 41a, along with the analysis conditions 42. The trained model selection unit 12e outputs the acquired trained model 34 and data processing program 41a (analysis conditions 42) to the analysis result acquisition unit 12c.

[0076] The analysis result acquisition unit 12c analyzes the measurement data 30 based on the trained model 34 and acquires the analysis result 31. In this embodiment, the analysis result acquisition unit 12c analyzes the measurement data 30 based on the trained model 34 generated by the trained model generation unit 12b and acquires the analysis result 31. Specifically, the analysis result acquisition unit 12c inputs the measurement data 30 to the trained model 34 selected by the trained model selection unit 12e and acquires the analysis result 31. In other words, the analysis result acquisition unit 12c acquires the analysis result 31 based on the selected analysis conditions 42. Furthermore, when acquiring the analysis result 31, the analysis result acquisition unit 12c performs data processing by the data processing program 41a according to the trained model 34. The analysis result acquisition unit 12c outputs the acquired analysis result 31 to the data processing device 2.

[0077] The data processing device 2 displays the acquired analysis results 31 on the display unit 23 (see Figure 3).

[0078] (Data analysis processing) Next, with reference to Figure 15, the data analysis process performed by the data analysis system 100 (see Figure 1) according to this embodiment will be described.

[0079] In step 101, server 1 (see Figure 1) pre-stores a group of learning algorithms 40 (see Figure 6) which associates the type 33a (see Figure 6) of measurement data 30 (see Figure 1) with the type 33b (see Figure 6) of analysis of the measurement data 30 with the learning algorithm 40a (see Figure 6).

[0080] In step 102, server 1 pre-stores a group of data processing programs 41 (see Figure 11) that associate data processing programs 41a (see Figure 11) that perform data processing different from that of the trained model 34 (see Figure 2) with the types of measurement data 30 33a and the types of analysis 33b of the measurement data 30.

[0081] In step 103, the data analysis system 100 generates a trained model 34. Details of the process in step 103 will be described later.

[0082] In step 104, the data analysis system 100 analyzes the measurement data 30 based on the trained model 34 and obtains the analysis result 31 (see Figure 9). After that, the process ends. Details of the process in step 104 will be described later.

[0083] In this embodiment, the processes in step 101 and step 102 are performed when the system management company manufactures the data analysis system 100. The processes in step 103 and step 104 are performed when a customer operator uses the data analysis system 100.

[0084] (Trained model generation process) Next, referring to Figure 16, the process by which the data analysis system 100 (see Figure 1) generates the trained model 34 (see Figure 2) (the process in step 103 in Figure 15) will be explained. Note that the process shown in Figure 16 starts when the generate button 50d (see Figure 5) is pressed on screen 50 (see Figure 5), which is the screen for generating the trained model 34.

[0085] In step 103a, the data processing device 2 transmits training data 32 (see Figure 8) to the server 1.

[0086] In step 103b, Server 1 accepts the input of training data 32. Server 1 stores the received training data 32 in the first storage unit 10 (see Figure 2).

[0087] In step 103c, the data processing device 2 (see Figure 4) sends input information 33 (see Figure 4) to the server 1 (see Figure 4).

[0088] In step 103d, server 1 accepts input information 33. Specifically, server 1 accepts input information 33 which includes information about the type 33a of measurement data 30 (see Figure 4) and information about the type 33b of analysis of measurement data 30 (see Figure 4).

[0089] In step 103e, the learning algorithm selection unit 12a (see Figure 8) selects a learning algorithm 40a (see Figure 8) to be used for learning from among a group of learning algorithms 40 (see Figure 8) based on the input information 33. In other words, by the operator inputting the input information 33, a learning algorithm 40a suitable for learning is automatically selected.

[0090] In step 103f, the data processing device 2 sends a control signal to the server 1 to start generating the trained model 34. After that, processing in the data processing device 2 ends. When the processing in step 103f is performed, the second processor 21 (see Figure 3) displays the training algorithm 40a selected in step 103e on the display unit 23 (see Figure 3). This allows the user to confirm the selected training algorithm 40a and then start the operation to generate the trained model 34.

[0091] In step 103g, Server 1 determines whether or not it has received a control signal to start generating the trained model 34. If it has not received a control signal to start generating the trained model 34, it repeats the process in step 103g. If it has received a signal to start generating the trained model 34, the process proceeds to step 103h.

[0092] In step 103h, the trained model generation unit 12b (see Figure 8) generates a trained model 34 based on the training data 32 and the selected training algorithm 40a.

[0093] In step 103i, the trained model storage control unit 12d associates the type 33a of the measurement data 30, the type 33b of the analysis of the measurement data 30, and the trained model 34, and stores them in the first storage unit 10 (see Figure 8). After that, the process proceeds to step 104 (see Figure 15), and the generation process of the trained model 34 is completed.

[0094] (Measurement data analysis processing) Next, referring to Figure 17, we will explain the process by which the data analysis system 100 (see Figure 1) analyzes the measurement data 30 (see Figure 14) and obtains the analysis result 31 (see Figure 14) (the process of step 104 shown in Figure 15). Note that the process shown in Figure 17 starts when the analysis button 51d (see Figure 10) is pressed on the screen 51 (see Figure 10) where the operation to request the analysis of the measurement data 30 is performed.

[0095] In step 104a, the data processing device 2 (see Figure 9) transmits input information 33 (see Figure 9) to the server 1.

[0096] In step 104b, the data processing device 2 transmits the measurement data 30 to the server 1 (see Figure 9). Note that the processing in step 104a and the processing in step 104b may be performed in either order.

[0097] In step 104c, server 1 receives the measurement data 30. That is, server 1 receives the measurement data 30 acquired by the measuring device 3. Note that the processing in step 104c may be performed before the processing in step 104b, provided that the processing in step 104a has been performed.

[0098] In step 104d, server 1 receives input information 33. That is, server 1 receives information regarding the type 33a (see Figure 9) of the measurement data 30 and information regarding the type 33b (see Figure 9) of the analysis of the measurement data 30. In other words, server 1 receives an input for selecting the analysis conditions 42 (see Figure 13). Note that the processing in step 104d may be performed before the processing in step 104c, provided that the processing in step 104b has been performed.

[0099] In step 104e, the trained model selection unit 12e (see Figure 14) selects a trained model 34 (see Figure 14) based on the input information 33. In this embodiment, in step 104e, the trained model selection unit 12e selects a data processing program 41a (see Figure 14) along with the trained model 34 based on the input information 33. That is, the trained model selection unit 12e selects the analysis conditions 42 based on the input information 33. Therefore, by inputting the input information 33, the analysis conditions 42 suitable for analyzing the measurement data 30 are automatically selected.

[0100] In step 104f, the data processing device 2 sends a control signal to the server 1 to start analyzing the measurement data 30. When step 104f is performed, the second processor 21 (see Figure 3) displays the analysis conditions 42 selected in step 104e on the display unit 23 (see Figure 3). This allows the user to confirm the selected analysis conditions 42 and then start the analysis of the measurement data 30.

[0101] In step 104g, the first processor 12 (see Figure 14) determines whether or not it has received a control signal to start analyzing the measurement data 30. If it has not received a control signal to start analyzing the measurement data 30, it repeats the process in step 104g. If it has received a control signal to start analyzing the measurement data 30, the process proceeds to step 104h.

[0102] In step 104h, the analysis result acquisition unit 12c (see Figure 14) analyzes the measurement data 30 based on the trained model 34 and acquires the analysis result 31. In this embodiment, the analysis result acquisition unit 12c inputs the measurement data 30 to the trained model 34 selected by the trained model selection unit 12e and acquires the analysis result 31. That is, the analysis result acquisition unit 12c acquires the analysis result 31 based on the analysis conditions 42 selected by the trained model selection unit 12e.

[0103] In step 104i, the first processor 12 sends the analysis result 31 to the data processing device 2. After that, the process of server 1 acquiring the analysis result 31 is completed.

[0104] In step 104j, the data processing device 2 obtains the analysis result 31.

[0105] In step 104k, the data processing device 2 (second processor 21) displays the analysis result 31 on the display unit 23 (see Figure 3). After that, the process ends.

[0106] (Additional processing for learning algorithm) Next, referring to Figure 18, we will explain the process by which Server 1 (see Figure 1) adds the learning algorithm 40a (see Figure 6).

[0107] In step 110, the first processor 12 (see Figure 2) accepts the addition of other learning algorithms 40a to the pre-stored group of learning algorithms 40 (see Figure 6), which are associated with the type 33a (see Figure 6) of measurement data 30 (see Figure 1) and the type 33b (see Figure 6) of analysis of the measurement data 30.

[0108] In step 111, the first processor 12 stores other learning algorithms 40a associated with the type 33a of measurement data 30 and the type 33b of analysis of measurement data 30 in the first storage unit 10 (see Figure 2). After that, the process ends. The learning algorithms 40a stored in step 111 are learning algorithms 40a that have a different association between the type 33a of measurement data 30 and the type 33b of analysis of measurement data 30 from the learning algorithms 40a already stored as part of the learning algorithm group 40. Even if the type of learning algorithm 40a is the same, it can be added to the learning algorithm group 40 if at least one of the type 33a of measurement data 30 and the type 33b of analysis of measurement data 30 is different.

[0109] (Additional processing for data processing program) Next, referring to Figure 19, we will explain the process by which Server 1 (see Figure 1) adds the data processing program 41a (see Figure 11).

[0110] In step 120, the first processor 12 (see Figure 2) accepts the addition of other data processing programs 41a to the pre-stored data processing program group 41 (see Figure 11), which are associated with the type 33a (see Figure 11) of the measurement data 30 (see Figure 1) and the type 33b (see Figure 11) of the analysis of the measurement data 30.

[0111] In step 121, the first processor 12 stores other data processing programs 41a associated with the type 33a of measurement data 30 and the type 33b of analysis of measurement data 30 in the first storage unit 10 (see Figure 2). After that, the processing ends. Note that the data processing programs 41a stored in step 121 are data processing programs 41a that have a different association between the type 33a of measurement data 30 and the type 33b of analysis of measurement data 30 from the data processing programs 41a already stored as part of the data processing program group 41. Note that even if the type of data processing program 41a is the same, it can be added to the data processing program group 41 if at least one of the type 33a of measurement data 30 and the type 33b of analysis of measurement data 30 is different.

[0112] (Effects of this embodiment) In this embodiment, the following effects can be obtained.

[0113] In the above embodiment, the data analysis method is a data analysis method performed between a system management company and a customer who desires to obtain analysis results 31 of measurement data 30 acquired from a measurement device 3, and comprises: a storage step of pre-storing a group of learning algorithms 40 that associates a type 33a of measurement data 30, a type 33b of analysis of the measurement data 30, and a learning algorithm 40a; a teacher data reception step of receiving teacher data 32 as input; an input information reception step of receiving input information 33 that includes information about the type 33a of measurement data 30 and information about the type 33b of analysis of the measurement data 30; a learning algorithm selection step of selecting a learning algorithm 40a to be used for learning from among the group of learning algorithms 40 based on the input information 33; a trained model generation step of generating a trained model 34 based on the teacher data 32 and the selected learning algorithm 40a; and an analysis result acquisition step of analyzing the measurement data 30 based on the trained model 34 and obtaining analysis results 31.

[0114] As a result, by simply inputting information about the type 33a of measurement data 30 and the type 33b of analysis of the measurement data 30, a training algorithm 40a suitable for the trained model 34 used to analyze the measurement data 30 is selected from among a group of training algorithms 40. Therefore, even if the user's skill level is low, the trained model 34 can be easily generated without outsourcing. Furthermore, since the trained model 34 can be generated without outsourcing, the time required to generate the trained model 34 can be reduced. As a result, it is possible to provide a data analysis method that enables even users with low skill levels to easily generate a trained model 34 and reduces the time required to analyze the measurement data 30 using the trained model 34.

[0115] Furthermore, in the above embodiment, the data analysis system 100 is a data analysis system for performing data analysis between a system management company and a customer who desires to obtain analysis results 31 of measurement data 30 acquired from a measurement device 3, and comprises a server 1 that generates a trained model 34 for analyzing the measurement data 30, and a data processing device 2 that requests the server 1 to analyze the measurement data 30, and the server 1 has a first storage unit 10 that pre-stores a plurality of training algorithm groups 40 that associate the types of measurement data 30 33a, the types of analysis 33b of the measurement data 30, and a training algorithm 40a, and the types of measurement data 30 33a The system includes an input receiving unit 11 that receives input information 33, which includes information about the measurement data 30 and information about the type of analysis 33b of the measurement data 30, and input of training data 32; a learning algorithm selection unit 12a that determines a learning algorithm 40a to be used for learning from among a group of learning algorithms 40 based on the input information 33; a trained model generation unit 12b that generates a trained model 34 based on the training data 32 and the selected learning algorithm 40a; and an analysis result acquisition unit 12c that analyzes the measurement data 30 based on the trained model 34 generated by the trained model generation unit 12b and obtains an analysis result 31.

[0116] This makes it possible to provide a data analysis system 100 that, similar to the data analysis method described above, allows even users with low skill levels to easily generate a trained model 34, and also reduces the time required to analyze measurement data 30 using the trained model 34.

[0117] Furthermore, in the above embodiment, the data analysis system server 1 is a data analysis system server for performing data analysis between a system management company and a customer who wishes to obtain the analysis results 31 of measurement data 30 acquired from the measuring device 3, and includes a first storage unit 10 that pre-stores a plurality of learning algorithm groups 40 which associate the types 33a of measurement data 30, the types 33b of analysis of measurement data 30, and the learning algorithm 40a, and input information including information on the types 33a of measurement data 30 and information on the types 33b of analysis of measurement data 30. The system includes an input receiving unit 11 that receives the input of report 33 and the input of training data 32; a training algorithm selection unit 12a that determines a training algorithm 40a to be used for training from among a group of training algorithms 40 based on the input information 33; a trained model generation unit 12b that generates a trained model 34 based on the training data 32 and the selected training algorithm 40a; and an analysis result acquisition unit 12c that analyzes the measurement data 30 based on the trained model 34 generated by the trained model generation unit 12b and obtains the analysis result 31.

[0118] This makes it possible to provide a data analysis system server 1 that, similar to the data analysis method described above, allows even users with low skill levels to easily generate a trained model 34, and also reduces the time required to analyze measurement data 30 using the trained model 34.

[0119] Furthermore, in the above embodiment, the following additional effects can be obtained by configuring it as follows.

[0120] In other words, in this embodiment, as described above, in the learning algorithm selection step, a learning algorithm 40a that matches the information regarding the type 33a of the measurement data 30 and the information regarding the type 33b of the analysis of the measurement data 30 is selected from among a group of learning algorithms 40. As a result, a learning algorithm 40a that matches the type 33a of the measurement data 30 and the type 33b of the analysis of the measurement data 30 is selected from among the group of learning algorithms 40. Therefore, even if the user has little knowledge of learning algorithms 40a, they can easily create a trained model 34 using a learning algorithm 40a suitable for creating a trained model 34 to be used for analyzing the measurement data 30. As a result, even a user with low skill level can easily generate a trained model 34 suitable for analyzing the measurement data 30.

[0121] Furthermore, in this embodiment, as described above, the system further includes a trained model storage step that stores the trained model 34 generated in the trained model generation step in association with the type 33a and type 33b of analysis of the measurement data 30 received in the input information reception step, an analysis reception step that receives information on the type 33a of the measurement data 30, information on the type 33b of analysis of the measurement data 30, and the measurement data 30 acquired by the measurement device 3, and a trained model selection step that selects a trained model 34 to be used for analyzing the measurement data 30 based on the information on the type 33a of the measurement data 30 and the type 33b of analysis of the measurement data 30 received in the analysis reception step, and an analysis result acquisition step that inputs the measurement data 30 to the trained model 34 selected in the trained model selection step and acquires the analysis result 31. As a result, the generated pre-trained models 34 are stored in association with the types 33a of the measurement data 30 and the types 33b of the analysis of the measurement data 30. Therefore, when analyzing the measurement data 30 using the generated pre-trained models 34, it is possible to easily select a pre-trained model 34 suitable for the analysis of the measurement data 30 based on the information of the types 33a of the measurement data 30 and the information of the types 33b of the analysis of the measurement data 30. Consequently, even users with low skill levels can easily analyze the measurement data 30 using the generated pre-trained models 34, thereby improving user convenience (usability).

[0122] Furthermore, in this embodiment, as described above, a data processing program storage step is provided in which a group of data processing programs 41, each associated with a type of measurement data 30 33a and a type of analysis 33b of the measurement data 30, is stored in advance. In the analysis acceptance step, an operation input is further accepted to select an analysis condition 42 that includes the learned model 34 and the data processing program 41a. In the analysis result acquisition step, the analysis result 31 is acquired based on the selected analysis condition 42. As a result, since the analysis condition 42 includes the data processing program 41a along with the learned model 34, the measurement data 30 can be analyzed by, for example, the data processing program 41a that performs the processing necessary for analysis by the learned model 34. As a result, the accuracy of the analysis result 31 of the measurement data 30 can be improved compared to a configuration in which the measurement data 30 is analyzed using only the learned model 34.

[0123] Furthermore, in this embodiment, as described above, the system further includes a data processing program addition acceptance step that accepts the addition of other data processing programs 41a associated with the types 33a of measurement data 30 and the types 33b of analysis of measurement data 30 to the pre-stored data processing program group 41. This makes it easier to add data processing programs 41a when a user desires to perform analysis using data processing programs 41a other than those pre-stored, compared to, for example, a configuration that updates the entire system. As a result, it becomes possible to easily add data processing programs 41a desired by the user, and thus the functionality of data processing programs 41a in the analysis of measurement data 30 can be easily expanded.

[0124] Furthermore, in this embodiment, as described above, the system further includes a learning algorithm addition acceptance step that accepts the addition of other learning algorithms 40a associated with the type 33a of measurement data 30 and the type 33b of analysis of the measurement data 30 to a group of learning algorithms 40 that have been stored in advance. This makes it easier to add learning algorithms 40a when a user desires to generate a trained model 34 using a learning algorithm 40a other than those stored in advance, compared to a configuration that updates the entire system. As a result, it becomes possible to easily add the learning algorithm 40a desired by the user, and thus the function for generating a trained model 34 suitable for analyzing the measurement data 30 can be easily expanded.

[0125] [Differentiation] It should be noted that the embodiments disclosed herein are illustrative and not restrictive in all respects. The scope of the present invention is indicated by the claims rather than by the description of the embodiments above, and further includes all modifications (exceptions) within the meaning and scope equivalent to the claims.

[0126] For example, the above embodiment shows an example in which the data analysis system 100 performs analysis of the measurement data 30 using the server 1, but the present invention is not limited thereto. For example, the analysis of the measurement data 30 may be performed by a device other than the server 1, as shown in the modified data analysis system 200 in Figure 20. The modified data analysis system 200 shown in Figure 20 differs from the data analysis system 100 in that it includes a measurement device data processing device 4.

[0127] The measurement device data processing unit 4 is configured to control the measurement device 3. The measurement device data processing unit 4 is also connected to the server 1. Furthermore, the measurement device data processing unit 4 is configured to analyze the measurement data 30.

[0128] As shown in Figure 20, the modified measurement device data processing device 4 is configured to transmit input information 33 to the server 1. Upon receiving the input information 33 from the measurement device data processing device 4, the server 1 selects the analysis conditions 42. The configuration for the server 1 to select the analysis conditions 42 is the same as in the above embodiment, so a detailed explanation is omitted. The server 1 also transmits the selected analysis conditions 42 to the measurement device data processing device 4.

[0129] Upon receiving the analysis conditions 42, the measurement device data processing device 4 analyzes the measurement data 30 based on the analysis conditions 42 and obtains the analysis result 31. The measurement device data processing device 4 also transmits the obtained analysis result 31 to the data processing device 2. This allows the data processing device 2 to display the analysis result 31. The configuration in which the measurement device data processing device 4 obtains the analysis result 31 is the same as the configuration in which the analysis result acquisition unit 12c obtains the analysis result 31 in the above embodiment, so a detailed explanation is omitted.

[0130] Furthermore, while the above embodiment shows an example in which the learning algorithm selection unit 12a selects one learning algorithm 40a that matches the type 33a of the measurement data 30 and the type 33b of the analysis of the measurement data 30, the present invention is not limited to this. For example, the learning algorithm selection unit 12a may be configured to select multiple candidate learning algorithms 40a from the group of learning algorithms 40 based on the type 33a of the measurement data 30 and the type 33b of the analysis of the measurement data 30. With this configuration, multiple candidate learning algorithms 40a can be displayed on the display unit 23. As a result, the operator can select multiple learning algorithms 40a, thereby improving the degree of freedom in selecting a learning algorithm 40a.

[0131] Furthermore, while the above embodiment shows an example in which the trained model selection unit 12e selects one trained model 34 based on the type 33a of the measurement data 30 and the type 33b of the analysis of the measurement data 30, the present invention is not limited to this. For example, the trained model selection unit 12e may be configured to select a plurality of candidate trained models 34 based on the type 33a of the measurement data 30 and the type 33b of the analysis of the measurement data 30. With this configuration, it becomes possible to display a plurality of candidate trained models 34 on the display unit 23, thereby improving the degree of freedom in selecting a trained model 34.

[0132] Furthermore, although the above embodiment shows an example in which the first storage unit 10 has pre-stored a group of data processing programs 41, the present invention is not limited to this. The first storage unit 10 does not need to pre-store a group of data processing programs 41. In this case, the operator only needs to input the data processing program 41a when analyzing the measurement data 30. However, if the operator has to input the data processing program 41a every time they analyze the measurement data 30, the operation becomes complicated and the burden on the operator increases. Therefore, it is preferable for the first storage unit 10 to pre-store a group of data processing programs 41.

[0133] Furthermore, while the above embodiment shows an example in which the server 1 is configured to allow the addition of data processing program 41a to the data processing program group 41, the present invention is not limited to this. For example, the server 1 does not need to be configured to allow the addition of data processing program 41a to the data processing program group 41. However, if the server 1 is not configured to allow the addition of data processing program 41a to the data processing program group 41, it becomes difficult for the operator to easily add the desired data processing program 41a, reducing user convenience (usability). Therefore, it is preferable that the server 1 is configured to allow the addition of data processing program 41a to the data processing program group 41.

[0134] Furthermore, in the above embodiment, an example was shown in which the first processor 12 accepts the addition of other learning algorithms 40a associated with the types 33a of measurement data 30 and the types 33b of analysis of measurement data 30 to the pre-stored group of learning algorithms 40, but the present invention is not limited thereto. For example, the first processor 12 may accept the input of the types 33a of measurement data 30, the input of the types 33b of analysis of measurement data 30, and the input of other learning algorithms 40a individually. In this case, the first processor 12 only needs to be configured to associate the types 33a of measurement data 30 and the types 33b of analysis of measurement data 30 with other learning algorithms 40a and store them in the first storage unit 10.

[0135] Furthermore, while the above embodiment shows an example in which the server 1 is configured to allow the addition of a learning algorithm 40a to the learning algorithm group 40, the present invention is not limited to this. For example, the server 1 does not need to be configured to allow the addition of a learning algorithm 40a to the learning algorithm group 40. However, if the server 1 is not configured to allow the addition of a learning algorithm 40a to the learning algorithm group 40, it becomes difficult for the operator to easily add a data processing program 41a desired by the operator, and user convenience (usability) decreases. Therefore, it is preferable that the server 1 is configured to allow the addition of a learning algorithm 40a to the learning algorithm group 40.

[0136] Furthermore, in the above embodiment, an example was shown in which the first processor 12 accepts the addition of other data processing programs 41a associated with the types 33a of measurement data 30 and the types 33b of analysis of measurement data 30 to a pre-stored group of data processing programs 41, but the present invention is not limited thereto. For example, the first processor 12 may be configured to accept the input of the types 33a of measurement data 30, the input of the types 33b of analysis of measurement data 30, and the input of other data processing programs 41a separately. In this case, the first processor 12 only needs to be configured to associate the types 33a of measurement data 30 and the types 33b of analysis of measurement data 30 with other data processing programs 41a and then store them in the first storage unit 10.

[0137] Furthermore, although the above embodiment shows an example configuration in which the first processor 12 receives the training data 32 at a different timing than the input information 33 and stores it in the first storage unit 10, the present invention is not limited to this. For example, the training data 32 may be input to the server 1 together with the input information 33. That is, the first processor 12 may be configured to generate a trained model 34 based on the training data 32 input together with the input information 33.

[0138] Furthermore, in the above embodiment, for the sake of explanation, a flow-driven flowchart was used to describe the data analysis process, the process of generating a trained model 34, the process of obtaining the analysis results 31 of the measurement data 30, the process of adding a learning algorithm 40a, and the process of adding a data processing program 41a, in which these processes are carried out sequentially according to the processing flow. However, the present invention is not limited to this. In the present invention, processing operations may be carried out by event-driven processing, which executes processing on an event-by-event basis. In this case, it may be carried out as a completely event-driven system, or a combination of event-driven and flow-driven systems may be used.

[0139] [Pattern] Those skilled in the art will understand that the exemplary embodiments described above are specific examples of the following embodiments.

[0140] (Item 1) A data analysis method performed between a system management company and a customer who desires to obtain the results of the analysis of measurement data acquired from a measuring device, A storage step of pre-storing a group of learning algorithms that associate the type of measurement data, the type of analysis of the measurement data, and the learning algorithm; A training data reception process that accepts input of training data, An input information receiving step that receives input information including information regarding the type of measurement data and information regarding the type of analysis of the measurement data, A learning algorithm selection step, in which a learning algorithm to be used for learning is selected from among the plurality of learning algorithms based on the input information, A trained model generation step, which generates a trained model based on the aforementioned training data and the determined training algorithm, A data analysis method comprising: an analysis result acquisition step of analyzing the measurement data based on the trained model and obtaining the analysis result.

[0141] (Item 2) The data analysis method according to item 1, wherein in the learning algorithm selection step, a learning algorithm is selected from the plurality of learning algorithms that matches the information regarding the type of measurement data and the information regarding the type of analysis of the measurement data.

[0142] (Item 3) A trained model storage step stores the trained model generated in the trained model generation step and the type of measurement data and the type of analysis of the measurement data received in the input information reception step in association with each other. An analysis reception process that receives information on the type of measurement data, information on the type of analysis of the measurement data, and the measurement data acquired by the measurement device. The system further comprises a trained model selection step, which selects a trained model to be used for analyzing the measurement data based on the information of the type of measurement data received in the analysis acceptance step and the information of the type of analysis of the measurement data, The data analysis method according to item 2, wherein in the analysis result acquisition step, the measurement data is input to the trained model selected in the trained model selection step, and the analysis result is acquired.

[0143] (Item 4) The system further includes a data processing program storage step, which stores in advance a group of data processing programs that perform data processing different from the aforementioned trained model, each of which is associated with the type of measurement data and the type of analysis of the measurement data. In the analysis acceptance step, the operator further accepts an input for selecting analysis conditions including the trained model and the data processing program. The data analysis method described in item 3, wherein in the step of obtaining the analysis results, the analysis results are obtained based on the selected analysis conditions.

[0144] (Item 5) The data analysis method according to item 4, further comprising a data processing program addition acceptance step for accepting the addition of other data processing programs associated with the type of measurement data and the type of analysis of the measurement data to the pre-stored group of data processing programs.

[0145] (Item 6) The data analysis method according to item 1, further comprising a learning algorithm addition acceptance step for accepting the addition of other learning algorithms associated with the type of measurement data and the type of analysis of the measurement data to the plurality of learning algorithms that have been stored in advance.

[0146] (Item 7) A data analysis system for performing data analysis between a system management company and a customer who desires to obtain the results of the analysis of measurement data acquired from a measuring device, A server that generates a trained model for analyzing the aforementioned measurement data, The system includes a data processing device that requests the server to analyze the measurement data, The aforementioned server, A storage unit that pre-stores a group of learning algorithms that associate the type of measurement data, the type of analysis of the measurement data, and the learning algorithm; An input receiving unit that accepts input information including information about the type of measurement data and information about the type of analysis of the measurement data, and input of training data. A learning algorithm selection unit that determines the learning algorithm to be used for learning from among the plurality of learning algorithms based on the input information, A trained model generation unit generates the trained model based on the aforementioned training data and the determined training algorithm, A data analysis system comprising: an analysis result acquisition unit that analyzes the measurement data based on the trained model generated by the trained model generation unit and acquires the analysis results.

[0147] (Item 8) A server for a data analysis system used to perform data analysis between a system management company and a customer who desires to obtain the results of the analysis of measurement data acquired from a measuring device, A storage unit that pre-stores a group of learning algorithms that associate the type of measurement data, the type of analysis of the measurement data, and the learning algorithm; An input receiving unit that accepts input information including information about the type of measurement data and information about the type of analysis of the measurement data, and input of training data. A learning algorithm selection unit that determines the learning algorithm to be used for learning from among the plurality of learning algorithms based on the input information, A trained model generation unit generates a trained model based on the aforementioned training data and the determined training algorithm. A server for a data analysis system, comprising: an analysis result acquisition unit that analyzes the measurement data based on the trained model generated by the trained model generation unit and acquires the analysis results. [Explanation of symbols]

[0148] 1 server (server for data analysis system) 2 Data Processing Devices 3. Measuring device 10 First storage section (storage section) 11 Input reception section 12a Learning Algorithm Selection Unit 12b Pre-trained model generation unit 12c Analysis result acquisition part 30 Measurement data 31 Analysis results 32 Training Data 33 Input Information 33a Types of measurement data 33b Types of analysis of measurement data 34 Pre-trained models 40 Learning Algorithms 40a Learning Algorithm 41 Data Processing Programs 41a Data Processing Program 42 Analysis conditions 100, 200 Data Analysis Systems

Claims

1. A data analysis method performed between a system management company and a customer who desires to obtain the results of the analysis of measurement data acquired from a measuring device, A storage step of pre-storing a group of learning algorithms that associate the type of measurement data, the type of analysis of the measurement data, and the learning algorithm; A training data reception process that accepts input of training data, A learning input information receiving step that receives learning input information including information about the type of measurement data and information about the type of analysis of the measurement data, A learning algorithm selection step, in which a learning algorithm to be used for learning is selected from among the plurality of learning algorithms based on the learning input information, A trained model generation step, which generates a trained model based on the aforementioned training data and the determined training algorithm, A trained model storage step stores the trained model generated in the trained model generation step and the type of measurement data and the type of analysis of the measurement data received in the training input information reception step in association with each other. An analysis reception step that accepts input of analysis input information including information on the type of measurement data and information on the type of analysis of the measurement data, and input of the measurement data acquired by the measurement device, A trained model selection step, in which a trained model to be used for analyzing the measurement data is selected based on the analysis input information received in the analysis acceptance step, The system includes an analysis result acquisition step which involves analyzing the measurement data based on the trained model and obtaining the analysis results, A data analysis method comprising the steps of obtaining the analysis result, inputting the measurement data into the trained model selected in the trained model selection step, and obtaining the analysis result.

2. The data analysis method according to claim 1, wherein in the learning algorithm selection step, a learning algorithm is selected from among the plurality of learning algorithms that matches the information regarding the type of measurement data and the information regarding the type of analysis of the measurement data.

3. The system further includes a data processing program storage step, which stores in advance a group of data processing programs that perform data processing different from the aforementioned trained model, each of which is associated with the type of measurement data and the type of analysis of the measurement data. In the analysis acceptance step, the operator further accepts an input for selecting analysis conditions including the trained model and the data processing program. The data analysis method according to claim 2, wherein in the step of obtaining the analysis results, the analysis results are obtained based on the selected analysis conditions.

4. The data analysis method according to claim 3, further comprising a data processing program addition acceptance step for accepting the addition of other data processing programs associated with the type of measurement data and the type of analysis of the measurement data to a pre-stored group of data processing programs.

5. The data analysis method according to claim 1, further comprising a learning algorithm addition acceptance step for accepting the addition of other learning algorithms to a plurality of learning algorithms stored in advance, which are associated with the type of measurement data and the type of analysis of the measurement data.

6. A data analysis system for performing data analysis between a system management company and a customer who desires to obtain the results of the analysis of measurement data acquired from a measuring device, A server that generates a trained model for analyzing the aforementioned measurement data, The system includes a data processing device that requests the server to analyze the measurement data, The aforementioned server, A storage unit that pre-stores a group of learning algorithms that associate the type of measurement data, the type of analysis of the measurement data, and the learning algorithm; An input receiving unit that receives input of learning input information including information about the type of measurement data and information about the type of analysis of the measurement data, input of training data, input of analysis input information including information about the type of measurement data and information about the type of analysis of the measurement data, and input of the measurement data acquired by the measurement device. A learning algorithm selection unit that determines the learning algorithm to be used for learning from among the group of learning algorithms based on the learning input information, A trained model generation unit generates the trained model based on the aforementioned training data and the determined training algorithm, A trained model selection unit selects a trained model to be used for analyzing the measurement data based on the analysis input information received via the input receiving unit, The system includes an analysis result acquisition unit that analyzes the measurement data based on the trained model generated by the trained model generation unit and acquires the analysis results, The memory unit stores the generated trained model in association with the type of measurement data and the type of analysis of the measurement data received as training input information. The aforementioned analysis result acquisition unit is a data analysis system that inputs the measurement data into the selected pre-trained model and acquires the analysis results.

7. A server for a data analysis system used to perform data analysis between a system management company and a customer who desires to obtain the results of the analysis of measurement data acquired from a measuring device, A storage unit that pre-stores a group of learning algorithms that associate the type of measurement data, the type of analysis of the measurement data, and the learning algorithm; An input receiving unit that receives input of learning input information including information about the type of measurement data and information about the type of analysis of the measurement data, input of training data, input of analysis input information including information about the type of measurement data and information about the type of analysis of the measurement data, and input of the measurement data acquired by the measurement device. A learning algorithm selection unit that determines the learning algorithm to be used for learning from among the group of learning algorithms based on the learning input information, A trained model generation unit generates a trained model based on the aforementioned training data and the determined training algorithm. A trained model selection unit selects a trained model to be used for analyzing the measurement data based on the analysis input information received via the input receiving unit, The system includes an analysis result acquisition unit that analyzes the measurement data based on the trained model generated by the trained model generation unit and acquires the analysis results, The memory unit stores the generated trained model in association with the type of measurement data and the type of analysis of the measurement data received as training input information. The aforementioned analysis result acquisition unit is a server for a data analysis system that inputs the measurement data into the selected pre-trained model and acquires the analysis results.

Images