Method for creating prediction model, method for setting conditions of operation process

Abstract

To provide a prediction model forming method for forming a prediction model that can be fed back to an explanatory variable of an operating process.SOLUTION: In a method for forming a prediction model that predicts a predetermined objective variable related to an operating process, the prediction model forming method is provided which comprises: a visualization step of visualizing a relationship between a plurality of explanatory variables for objective variables using a correlation coefficient between the plurality of explanatory variables; a division step of dividing the plurality of explanatory variables into a plurality of clusters using the relationship between the explanatory variables visualized in the visualization step; a selection step of selecting, for each cluster, a representative explanatory variable, which is the explanatory variable representing a cluster, from among the explanatory variables included in the cluster in one or more of the clusters selected from the plurality of clusters; and a prediction model forming step of forming a prediction model that predicts the objective variable using the representative explanatory variable selected in the selection step.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to a method for creating a prediction model and a method for setting conditions of an operation process.

Background Art

[0002] Taking into account changes in product manufacturing conditions and manufacturing environments, accurately predicting product quality has been increasing in importance year by year in combination with the IoTization of measuring instruments and the like.

[0003] For example, Patent Document 1 describes a quality prediction device that creates a quality prediction model using data of a group of products manufactured in a manufacturing process in the past.

[0004] Also, Patent Document 2 discloses a learning method for a blast furnace charging rate prediction model that predicts the blast furnace charging rate, which is an important parameter in manufacturing.

Prior Art Documents

Patent Documents

[0005]

Patent Document 1

Patent Document 2

Summary of the Invention

Problems to be Solved by the Invention

[0006] By the way, in an operation process for manufacturing industrial products and the like, data such as raw material data, operation process data, and evaluation values of intermediate products are usually collected as data. When creating a prediction model with the final characteristics or important characteristics as target variables, the prediction model is constructed using the above data as parameters, that is, explanatory variables.

[0007] From the perspective of manufacturing a product of the desired quality with high yield, it is required that feedback can be provided from the created prediction model, and the above-described parameters in the operation process can be calculated and set.

[0008] However, no method for creating a prediction model that can perform feedback and calculate parameters in the operation process has been proposed.

[0009] Therefore, in view of the problems of the above prior art, an aspect of the present invention aims to provide a method for creating a prediction model that can create a prediction model capable of providing feedback to the explanatory variables of the operation process.

Means for Solving the Problems

[0010] According to one aspect of the present invention for solving the above problems, A method for creating a prediction model for predicting a predetermined target variable related to an operation process, A visualization step of visualizing the relationship between a plurality of explanatory variables for the target variable using the correlation coefficients between the plurality of explanatory variables; A dividing step of dividing the plurality of explanatory variables into a plurality of clusters using the relationship between the explanatory variables visualized in the visualization step; A selection step of selecting, for each of one or more of the clusters selected from the plurality of clusters, a representative explanatory variable that is the explanatory variable representing the cluster from among the explanatory variables included in the cluster; And a prediction model creation step of creating a prediction model for predicting the target variable using the representative explanatory variable selected in the selection step. and in the selection step, a predetermined number of the representative explanatory variables are selected according to predetermined conditions the conditions include selecting the explanatory variable having a high correlation coefficient with the target variable, selecting the explanatory variable related to the upstream process of the operation process, selecting the explanatory variable that is easy to control in the operation process, and when creating a prediction model for the target variable using the explanatory variables included in the cluster, selecting the explanatory variable having a high influence on the target variable, and are one or more selected from these A method for creating a prediction model is provided.

Effects of the Invention

[0011] According to one aspect of the present invention, it is possible to provide a method for creating a prediction model that can feedback to the explanatory variables of an operation process.

Brief Description of the Drawings

[0012]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Modes for Carrying Out the Invention

[0013] Hereinafter, embodiments for carrying out the present invention will be described. However, the present invention is not limited to the following embodiments, and various modifications and substitutions can be made to the following embodiments without departing from the scope of the present invention. [Method for Creating Prediction Model] The method for creating a prediction model of the present embodiment will be described.

[0014] The inventor of the present invention has studied a method for creating a prediction model that can feedback to the explanatory variables of an operation process.

[0015] In recent years, the operation process has become a multi-step process, and the parameters of the operation process have also increased. In such a multi-step operation process, usually, parameters of raw materials, various parameters of the operation process, evaluation values of intermediates, etc. can be explanatory variables of the prediction model. However, in the case of multi-steps, inevitably, the types of parameters become very numerous. When creating a prediction model with these parameters as explanatory variables, it becomes very difficult to feedback and obtain appropriate values for each explanatory variable.

[0016] For example, it is also conceivable to select only the parameters related to the target variable, which are the final characteristics or important characteristics, as explanatory variables and create a prediction model. However, it is rare that the parameters of raw materials and the evaluation values of intermediates are uncorrelated. In addition, the parameters of the operation process often change in conjunction with various parameters and are quantities that move with correlation to each other. Therefore, it has been difficult to select the explanatory variables used when creating a prediction model.

[0017] Therefore, the inventor of the present invention further studied and completed the method for creating the prediction model of the present invention.

[0018] The method for creating a prediction model of the present embodiment is a method for creating a prediction model for predicting a predetermined target variable related to an operation process. As shown in the flow 10 of FIG. 1, it can have the following visualization step S11, division step S12, selection step S13, and prediction model creation step S14.

[0019] Note that the type of the target variable of the prediction model created in the prediction model creation step of the present embodiment is not particularly limited, and examples include the final characteristics required for the product to be manufactured, particularly important characteristics, etc.

[0020] The visualization step can visualize the relationship between a plurality of explanatory variables for the target variable using the correlation coefficients between the plurality of explanatory variables.

[0021] The splitting process can split a plurality of explanatory variables into a plurality of clusters using the relationships between the explanatory variables visualized in the visualization process.

[0022] In the selection process, in one or more clusters selected from a plurality of clusters, for each cluster, a representative explanatory variable, which is an explanatory variable representing the cluster, can be selected from among the explanatory variables included in the cluster.

[0023] The prediction model creation process can create a prediction model for predicting the target variable using the representative explanatory variables selected in the selection process. (1) Regarding each process Hereinafter, each process will be described. (1-1) Visualization process (S11) In the visualization process, the relationships between a plurality of explanatory variables regarding the target variable can be visualized using the correlation coefficients between the plurality of explanatory variables.

[0024] As a method for reducing redundancy between explanatory variables, principal component analysis etc. are generally used, but in order to obtain a feedback-enabled prediction model, it is preferable to handle the original explanatory variables as they are.

[0025] Therefore, in the method for creating a prediction model of the present embodiment, based on the correlation coefficients between the explanatory variables, the relationship of the correlation coefficients can be illustrated, that is, visualized.

[0026] In the visualization process, first, the explanatory variables during the operation process regarding the target variable of the prediction model to be created can be enumerated. And the correlation coefficients between all the enumerated explanatory variables can be obtained. At this time, the correlation coefficient between the target variable and the explanatory variables can also be obtained together.

[0027] For visualization, it is also possible to illustrate using the correlation coefficients as they are, but for example, a method of defining a distance based on the correlation coefficients between the explanatory variables and performing hierarchical clustering is preferable. As the distance, 1-|r| etc. can be used. Note that r means the correlation coefficient between each pair of explanatory variables.

[0028] The figure to be created is not limited, and it is preferable to create a figure that can show the explanatory variables and the relationship between the correlation coefficients and distances among the explanatory variables. For example, a dendrogram (tree diagram) can be created. (1-2) Division step (S12) In the division step, using the relationship between the explanatory variables visualized in the visualization step, a plurality of explanatory variables can be divided into a plurality of clusters, that is, groups.

[0029] At this time, the number of clusters to be divided and the division conditions are not particularly limited. For example, division can be performed based on a predetermined threshold value for the above-mentioned correlation coefficient and distance. Also, while confirming the physical meaning, it may be divided into a plurality of clusters.

[0030] Note that the number of explanatory variables included in one cluster is not particularly limited, and each cluster can include any number of explanatory variables. Also, the number of explanatory variables included in the clusters may be different. (1-3) Selection step (S13) In the selection step, in one or more clusters selected from a plurality of clusters, for each cluster, a representative explanatory variable, which is an explanatory variable representing the cluster, can be selected from among the explanatory variables included in the cluster.

[0031] As described above, in recent years, since the operation process has become multi-step, when creating a prediction model for the operation process, there is a risk that the number of explanatory variables will become very large. Therefore, it has sometimes been difficult to perform feedback on the prediction model and calculate the explanatory variables for obtaining a desired target variable.

[0032] Therefore, in the method for creating a prediction model of the present embodiment, in the above visualization step and division step, highly relevant explanatory variables are classified into clusters, and in one or more selected clusters, a representative explanatory variable is selected from each cluster as the representative explanatory variable, and a prediction model can be created. That is, a prediction model can be obtained that represents the variations of similar explanatory variables by the representative explanatory variable.

[0033] By clustering highly relevant explanatory variables in this way, selecting a representative explanatory variable, which is a representative value, from the cluster, and reducing the number of explanatory variables, it becomes possible to easily perform feedback from the target variable and obtain the explanatory variables. Also, since highly relevant explanatory variables are clustered and a representative explanatory variable is selected from the cluster, the explanatory variables can be appropriately selected. (1-3-1) Regarding the selection conditions for representative explanatory variables In the selection process, the criteria for selecting the representative explanatory variable are not particularly limited. In the selection process, a predetermined number of representative explanatory variables can be selected according to predetermined conditions.

[0034] As the conditions, it is preferable to select one or more from, for example, the following Conditions 1 to 4. (Condition 1) Select an explanatory variable with a high correlation coefficient with the target variable. (Condition 2) Select an explanatory variable related to the upstream process in the operation process. (Condition 3) Select an explanatory variable that is easy to control in the operation process. (Condition 4) When creating a prediction model for the target variable using the explanatory variables included in the cluster, select an explanatory variable that has a high influence on the target variable.

[0035] Note that the above Conditions 1 to 4 can also be used in combination. Also, for example, the conditions can be changed according to the cluster. (Regarding Condition 1) Selecting an explanatory variable with a high correlation coefficient with the target variable means obtaining the correlation coefficient between the explanatory variables constituting each cluster and the target variable, and sequentially selecting from the explanatory variables with a high correlation coefficient according to the number of explanatory variables to be selected from the cluster. Since an explanatory variable with a high correlation coefficient with the target variable is a variable that has a large influence on the target variable, by selecting according to the above conditions, a particularly appropriate prediction model can be obtained. (Regarding Condition 2) Selecting explanatory variables related to the upstream process in the operation process means arranging the explanatory variables that make up the cluster in order from the upstream of the related operation process, and selecting them in order from the upstream according to the number of explanatory variables to be selected from the cluster. Usually, when there is a correlation between variables, it is often more appropriate to control the upstream side of the operation process. Therefore, by selecting according to the above conditions, more appropriate representative explanatory variables can be selected. (Regarding Condition 3) Selecting explanatory variables that are easy to control in the operation process means, for example, numerically quantifying the ease of control in advance for the explanatory variables, and selecting the explanatory variables based on the numerical value according to the number of explanatory variables to be selected from the cluster. When feedback is performed from the prediction model and explanatory variables for obtaining a desired target are obtained, if the explanatory variables are parameters that are easy to control, the operation process can be easily adjusted, especially increasing the yield. (Regarding Condition 4) When creating a prediction model for the target variable using the explanatory variables included in the cluster, when selecting explanatory variables that have a high influence on the target variable, first create a prediction model for the target variable using the explanatory variables included in the cluster for which the representative explanatory variables are selected. Then, explanatory variables with a high degree of influence on the target variable in the prediction model can be selected. Specifically, for example, when the same fluctuation range is set for the explanatory variables, the explanatory variables with a high degree of influence can be set in descending order of the fluctuation range of the target variable. (1-3-2) Regarding the number of representative explanatory variables to be selected from each cluster In the selection process, the number of representative explanatory variables to be selected from each cluster is not particularly limited. For example, the number of representative explanatory variables to be selected from each cluster may be one or multiple. Also, the number of representative explanatory variables selected for each cluster may be different. For example, from a cluster considered to have a low importance, no representative explanatory variable may be selected. It is also possible to select three or more representative explanatory variables from each cluster, but since the effort for feedback increases, it is preferable that the number of representative explanatory variables selected from each cluster is two or less, and more preferably one.

[0036] The representative explanatory variables can be selected for all clusters. However, as described above, depending on the cluster, it is also possible not to select the representative explanatory variables. Therefore, one or more clusters selected from a plurality of clusters can be used to select the representative explanatory variables. However, when creating a prediction model, it is preferable to have a plurality of representative explanatory variables. Therefore, it is more preferable to select the representative explanatory variables with two or more clusters. (1-4) Prediction model creation process (S14) In the prediction model creation process, a prediction model for predicting the target variable can be created using the representative explanatory variables selected in the selection process.

[0037] The prediction model can be created using linear regression, logistic regression, random forest, boosting, neural network, etc. (2) Regarding the operation process The operation process to which the method for creating the prediction model of the present embodiment is applied is not particularly limited, and can be applied to any operation process.

[0038] The prediction model obtained by the method for creating the prediction model of the present embodiment described above has representative explanatory variables used in the prediction model as important feature quantities representing each cluster, and the number thereof is also limited. Therefore, it can be easily executed to feedback to the feature quantities based on the prediction model.

[0039] That is, according to the method for creating the prediction model of the present embodiment, a prediction model that can be fed back to the explanatory variables of the operation process can be created. [Method for setting conditions of operation process] The method for setting conditions of the operation process of the present embodiment can set a plurality of explanatory variables, for example, using the prediction model obtained by the method for creating the prediction model described above.

[0040] Specifically, the aforementioned prediction model has a target variable and representative explanatory variables as parameters for predicting the target variable. Therefore, by determining the target variable to be achieved, calculating the prediction model inversely, and obtaining each representative explanatory variable.

[0041] Then, based on the obtained representative explanatory variables, the conditions of the operation process can be set. [Simulation Device] The simulation device of this embodiment is a simulation device for creating a prediction model. Therefore, it can also be said to be a prediction model creation device and has the following visualization unit, division unit, selection unit, and prediction model creation unit.

[0042] According to the simulation device of this embodiment, the method for creating the aforementioned prediction model can be implemented. Therefore, the description of some matters already explained will be partially omitted.

[0043] The visualization unit can visualize the relationship between a plurality of explanatory variables for the target variable using the correlation coefficients between the plurality of explanatory variables.

[0044] The division unit can divide a plurality of explanatory variables into a plurality of clusters using the relationship between the explanatory variables visualized by the visualization unit.

[0045] The selection unit can select, for each of one or more clusters selected from a plurality of clusters, a representative explanatory variable, which is an explanatory variable representing the cluster, from among the explanatory variables included in the cluster.

[0046] The prediction model creation unit can create a prediction model for predicting the target variable using the representative explanatory variables selected by the selection unit.

[0047] As shown in the hardware configuration diagram shown in FIG. 2, the simulation device 20 of the present embodiment is configured by, for example, an information processing device (computer). Physically, it can be configured as a computer system including a CPU (Central Processing Unit) 21 which is an arithmetic processing unit, a RAM (Random Access Memory) 22 and a ROM (Read Only Memory) 23 which are main storage devices, an auxiliary storage device 24, an input / output interface 25, a display device 26 which is an output device, and the like. These are interconnected by a bus 27. Note that the auxiliary storage device 24 and the display device 26 may be provided externally.

[0048] The CPU 21 controls the overall operation of the simulation device 20 and performs various information processes. The CPU 21 can execute a program (simulation program), which will be described later, stored in the ROM 23 or the auxiliary storage device 24 to calculate the reflectance and the like.

[0049] The RAM 22 is used as a work area for the CPU 21 and may include a non-volatile RAM that stores main parameters and information.

[0050] The ROM 23 can store programs (simulation programs) and the like.

[0051] The auxiliary storage device 24 is a storage device such as an SSD (Solid State Drive) or an HDD (Hard Disk Drive), and can store various data, files, etc. necessary for the operation of the simulation device.

[0052] The input / output interface 25 includes both a user interface such as a touch panel, a keyboard, a display screen, and operation buttons, and a communication interface that captures information from an external data recording server or the like and outputs analysis information to other electronic devices.

[0053] The display device 26 is a monitor display or the like. On the display device 26, an analysis screen is displayed, and the screen is updated according to input / output operations via the input / output interface 25.

[0054] Each function of the simulation device 20 shown in FIG. 2 is realized by, for example, reading a program (simulation program) or the like from a main storage device or an auxiliary storage device 24 such as a RAM 22 or a ROM 23, and executing it by the CPU 21, thereby reading and writing data in the RAM 22 or the like, and operating the input / output interface 25 and the display device 26.

[0055] FIG. 3 shows a functional block diagram of the simulation device 20 of the present embodiment.

[0056] As shown in FIG. 3, the simulation device 20 can include a reception unit 31, a processing device 32, and an output unit 33. Each of these units is realized by the cooperation of software and hardware by the CPU executing a program (for example, a program described later) stored in advance in an information processing device such as a personal computer equipped with a CPU, a storage device, various interfaces, etc. that the simulation device 20 has.

[0057] The configuration of each unit will be described below. (1) Reception unit The reception unit 31 receives input of commands and data from a user related to the processing executed by the processing device 32. Examples of the reception unit 31 include a keyboard and a mouse on which the user operates to input commands, a communication device that inputs via a network, and a reading device that inputs from various storage media such as a CD-ROM and a DVD-ROM.

[0058] The reception unit 31 can receive, for example, objective variables regarding an operation process, a list of explanatory variables, and data of each explanatory variable for obtaining a correlation coefficient. (2) Processing device The processing device 32 can include a visualization unit 321, a division unit 322, a selection unit 323, and a prediction model creation unit 324. (2-1) Visualization unit The visualization unit 321 can visualize the relationship between a plurality of explanatory variables for a target variable using the correlation coefficients between the plurality of explanatory variables. Specifically, the correlation coefficients between the input explanatory variables are obtained, and a diagram related to the explanatory variables, such as a dendrogram, is created and visualized using the correlation coefficients and the aforementioned distances. (2-2) Division unit In the division unit 322, a plurality of explanatory variables can be divided into a plurality of clusters using the relationship between the explanatory variables visualized by the visualization unit. As described above, the explanatory variables can be divided based on the correlation coefficients and the predetermined threshold values for the distances. Also, while confirming the physical meaning, the division into a plurality of clusters may be performed. (2-3) Selection unit The selection unit 323 can select, for each of one or more clusters selected from a plurality of clusters, a representative explanatory variable that is an explanatory variable representing the cluster from among the explanatory variables included in the cluster. Since the selection conditions and the like of the representative explanatory variable have already been described, the description is omitted here. (2-4) Prediction model creation unit In the prediction model creation unit 324, a prediction model for predicting the target variable can be created using the representative explanatory variables selected by the selection unit.

[0059] The method for creating the prediction model is not particularly limited, and for example, it can be created using linear regression, logistic regression, random forest, boosting, neural network, etc. (3) Output unit The output unit 33 can include a display or the like. The prediction model obtained by the prediction model creation unit 324 can be output to the output unit 33.

[0060] According to the simulation device of the present embodiment described above, a prediction model that can be fed back to the explanatory variables of the operation process can be created. [Program] Next, the program of this embodiment will be described.

[0061] The program of this embodiment relates to a program for creating a prediction model, and can cause a computer to function as the following visualization unit, division unit, selection unit, and prediction model creation unit.

[0062] The visualization unit can visualize the relationship between a plurality of explanatory variables regarding the target variable using the correlation coefficients between the plurality of explanatory variables.

[0063] The division unit can divide a plurality of explanatory variables into a plurality of clusters using the relationship between the explanatory variables visualized by the visualization unit.

[0064] The selection unit can select, for each of one or more clusters selected from a plurality of clusters, a representative explanatory variable that is an explanatory variable representing the cluster from among the explanatory variables included in the cluster.

[0065] The prediction model creation unit can create a prediction model for predicting the target variable using the representative explanatory variables selected by the selection unit.

[0066] The program of this embodiment can be stored, for example, in various storage media of a main storage device or an auxiliary storage device such as a RAM or a ROM of the aforementioned simulation device. Then, by loading such a program and executing it by the CPU, data can be read and written in the RAM or the like, and the input / output interface and the display device can be operated to execute. Therefore, the description of the matters already described in the simulation device will be omitted.

[0067] According to the program of this embodiment described above, a prediction model that can be fed back to the explanatory variables of the operation process can be created.

Example

[0068] Hereinafter, the present invention will be described more specifically with reference to examples. However, the present invention is not limited to the following examples. [Example 1] A prediction model for predicting the target variable related to the operation process was created according to the following procedure. (Visualization step) In the visualization step, for the explanatory variables shown as feature quantities on the vertical axis of FIG. 4 and the target variable, the correlation coefficients between the explanatory variables and between the explanatory variable and the target variable were obtained, and the dendrogram shown in FIG. 4 was created.

[0069] When creating the dendrogram, the distance defined based on the correlation coefficient was used. The distance is defined as 1 - |r|. Here, r means the correlation coefficient between each pair of explanatory variables. (Division step) Based on the dendrogram obtained in the visualization step, the explanatory variables were divided into a plurality of clusters.

[0070] Specifically, regarding the distance, with 0.65 as a threshold criterion, it was divided into 10 clusters from cluster 41 to cluster 50. (Selection step) In the selection step, for each cluster, representative explanatory variables B1 to B8, which are the explanatory variables representing the cluster, were selected from among the explanatory variables included in the cluster. When selecting, the explanatory variables that are easy to control in the operation process were selected. Note that no representative explanatory variables were selected from cluster 46, cluster 48, and cluster 50. (Prediction model creation step) In the prediction model creation step, a prediction model for the target variable was created using the representative explanatory variables B1 to B8 selected in the selection step.

[0071] The measured values of the fitting data, which are the data used when creating the prediction model, and the predicted values calculated using the above prediction model are shown in FIG. 5.

[0072] Also, the measured values of the test data, which are the data not used when creating the prediction model, and the predicted values calculated using the above prediction model are also shown in FIG. 5.

[0073] According to the results shown in FIG. 5, it can be confirmed that the test data and the fitting data show almost the same distribution, and an appropriate prediction model could be created.

[0074] Also, in the prediction model of this embodiment, it is narrowed down to 8 representative explanatory variables from the 28 feature quantities shown in FIG. 4 and used. That is, since the number of parameters required for the prediction model is small, it can be seen that feedback from the prediction model can be easily performed by back-calculation.

Explanation of Signs

[0075] S11 Visualization process S12 Division process S13 Selection process S14 Prediction model creation process

Claims

1. A method for creating a prediction model that predicts a predetermined target variable related to an operation process, comprising: a visualization step of visualizing the relationship between a plurality of explanatory variables for the target variable using the correlation coefficients between the plurality of explanatory variables; a division step of dividing the plurality of explanatory variables into a plurality of clusters using the relationship between the explanatory variables visualized in the visualization step; a selection step of selecting, for each of one or more of the clusters selected from the plurality of clusters, a representative explanatory variable that is the explanatory variable representing the cluster from among the explanatory variables included in the cluster; a prediction model creation step of creating a prediction model for predicting the target variable using the representative explanatory variable selected in the selection step; in the selection step, a predetermined number of the representative explanatory variables are selected according to predetermined conditions; the conditions include selecting an explanatory variable having a high correlation coefficient with the target variable, selecting an explanatory variable related to an upstream process in the operation process, selecting an explanatory variable that is easy to control in the operation process, and when creating a prediction model for the target variable using the explanatory variables included in the cluster, selecting an explanatory variable having a high influence on the target variable, and the method for creating a prediction model is one or more selected from these.

2. A method for setting conditions of an operation process for setting a plurality of the explanatory variables using the prediction model obtained by the method for creating a prediction model according to Claim 1.

Images