WO2017094207A1 - 情報処理システム、情報処理方法および情報処理用プログラム - Google Patents
情報処理システム、情報処理方法および情報処理用プログラム Download PDFInfo
- Publication number
- WO2017094207A1 WO2017094207A1 PCT/JP2016/003927 JP2016003927W WO2017094207A1 WO 2017094207 A1 WO2017094207 A1 WO 2017094207A1 JP 2016003927 W JP2016003927 W JP 2016003927W WO 2017094207 A1 WO2017094207 A1 WO 2017094207A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- prediction model
- variable
- unit
- learning
- optimization
- Prior art date
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/02—Knowledge representation; Symbolic representation
- G06N5/022—Knowledge engineering; Knowledge acquisition
- G06N5/025—Extracting rules from data
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
- G06N20/10—Machine learning using kernel methods, e.g. support vector machines [SVM]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/01—Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0283—Price estimation or determination
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/04—Inference or reasoning models
- G06N5/045—Explanation of inference; Explainable artificial intelligence [XAI]; Interpretable artificial intelligence
Definitions
- the present invention relates to an information processing system, an information processing method, and an information processing program that perform optimization based on a learned prediction model.
- Patent Document 1 also describes a method for generating a prediction model based on past results.
- Patent Document 2 describes a system that presents a recommended price that is optimized to maximize the expected profit based on expected sales and price sensitivity.
- the price sensitivity model models the price sensitivity of a specific product through a function that fluctuates with the price, and models the change in sales as a function of price change. Has been.
- the system described in Patent Document 2 predicts sales under a given assumption, uses the prediction and price sensitivity conclusions to predict product sales, and under given constraints. To generate a set of optimal prices that maximizes gross profit.
- the system also displays the price sensitivity model type and the price sensitivity variable value.
- Patent Document 3 describes a system for assessing the effectiveness of communication content and optimizing content distribution.
- the system described in Patent Document 3 uses reinforcement learning.
- Patent Document 3 describes generating a content distribution schedule that is predicted to maximize the effectiveness evaluation scale (or objective function). Further, Patent Document 3 describes using regression analysis on historical data to predict the best “mixed” content, which maximizes results.
- the input data to the mathematical programming method is the amount of material, cost, manufacturing time, etc. necessary for making a product that is each line.
- These input data are all data that can be observed by the analyst when the analyst executes the mathematical programming.
- an object of the present invention is to provide an information processing system, an information processing method, and an information processing program capable of appropriately performing optimization even in a situation where there is input data that is not observed by mathematical optimization.
- An information processing system includes a learning unit that learns a prediction model that represents a relationship between an explained variable and an explanatory variable based on the explained variable and the explanatory variable, and is represented by a function of the explanatory variable, and a prediction model
- the objective variable that optimizes the objective function under the constraint condition for the objective function that takes the prediction model visualized by the visualization part as an argument in response to accepting an operation from the user
- an optimizing unit for calculating.
- the information processing method learns a prediction model that represents a relationship between the explained variable and the explanatory variable based on the explained variable and the explanatory variable, and is expressed by a function of the explanatory variable.
- the information processing program learns, based on the explained variable and the explanatory variable, to the computer, a prediction model that indicates the relationship between the explained variable and the explanatory variable and is expressed by a function of the explanatory variable.
- Learning function, visualization process that visualizes the prediction model, and the objective function that takes the prediction model visualized by the visualization process as an argument in response to accepting an operation from the user, under the constraint condition An optimization process for calculating an objective variable for optimizing the process is executed.
- optimization can be performed appropriately even in a situation where there is input data that is not observed by mathematical optimization.
- the information processing system of the present invention learns a prediction model for predicting unobserved data from past data, automatically generates an objective function of mathematical programming based on the prediction model, and executes optimization.
- the information processing system of the present invention visualizes the learned prediction model (in other words, displays it on the display device), and performs the optimization in response to a user operation instructing the execution of the optimization. To do.
- FIG. 1 is a block diagram showing an example of an information processing system of the present invention.
- the information processing system 1 includes a training data storage unit 2, a learning unit 3, a display control unit 4, a display device 5, a storage unit 6, an external information input unit 7, and a problem storage unit 8.
- the objective function generation unit 9 and the optimization unit 10 are provided.
- the display control unit 4 visualizes the prediction model learned by the learning unit 3 (in other words, displays information on the prediction model on the display device 5).
- the display control unit 4 not only displays information related to the prediction model, but also a user interface for the user to input parameters used for learning, a user interface for the user to input optimization constraints, and optimization results Are also displayed on the display device 5.
- FIG. 2 is an explanatory diagram showing an example of a screen displayed on the display device 5 by the display control unit 4.
- a screen 21 displayed by the display control unit 4 includes a prediction model display field 22, an input field 23 for parameters used for learning (hereinafter referred to as a first input field 23), an optimization result display field 25, It includes an input column 26 (hereinafter referred to as a second input column 26) for constraint conditions in optimization.
- the screen 21 has a button 24 (hereinafter referred to as a first instruction button 24) for the user to instruct execution of learning, and a button 27 (hereinafter referred to as a second button) for the user to instruct execution of optimization. Designated button 27)).
- FIG. 2 is an illustration of a screen displayed by the display control unit 4, and the mode of the screen is not limited to the example shown in FIG. 2.
- FIG. 2 shows an example in which the user instructs execution of learning or optimization with a button, but a screen in which these instructions are performed in other modes may be used.
- the prediction model display column 22 is a column for displaying information regarding the learned prediction model.
- the prediction model indicates the relationship between the explained variable and the explanatory variable, and is represented by a function of the explanatory variable.
- the learning unit 3 learns a plurality of types of prediction models using a plurality of types of learning algorithms for each explained variable.
- the learning unit 3 learns a prediction model for each learning algorithm using three kinds of learning algorithms (here, regression analysis, neural network, and support vector machine).
- the learning unit 3 the number of sales items 1 (and S 1.) With respect to the prediction model by regression analysis, to learn the prediction model using a neural network, and Support Vector Machines predictive models, respectively.
- the learning unit 3 similarly learns three types of prediction models for other explained variables.
- the prediction model which the learning part 3 learns for every to-be-explained variable may not be three types.
- the prediction model display field 22 includes a tab 22a for selecting the explained variable. There is a one-to-one correspondence between each tab and each explained variable.
- the tab corresponding to the explained variable “number of sales of product 1” is selected by the user, and the display control unit 4 displays the three types of prediction models learned regarding “number of sales of product 1” as a prediction model. The case where it displays in the column 22 is illustrated.
- f a shown in FIG. 2 is a prediction model based on regression analysis
- f b is a prediction model based on a neural network
- f c is a prediction model based on a support vector machine.
- a typical function is displayed in the prediction model display field 22.
- the display control unit 4 displays information on the three types of prediction models learned for the explained variable corresponding to the tab.
- the prediction model display column 22 includes a user interface for the user to select one prediction model from among three types of prediction models for each explained variable (in other words, for each tab).
- FIG. 2 illustrates a case where this user interface is a radio button 22b.
- the user determines the most appropriate prediction model from the three types of prediction models displayed, and operates the radio button 22b to select the one prediction model.
- the user may select a prediction model independently for each explained variable. For example, the user selects a prediction model based on a neural network regarding “number of sales of product 1”, selects a prediction model based on regression analysis regarding “number of sales of product 2”, and supports vector machines regarding “number of sales of product 3”. You may select the prediction model by.
- the first input field 23 is a field for the user to input parameters used for learning.
- a case where “5” is input by the user as the number of explanatory variables used in the prediction model is illustrated.
- the learning unit 3 learns each prediction model so that the number of explanatory variables used in the prediction model is five.
- the number of explanatory variables used in the prediction model is illustrated as an example of the parameters used for learning, but the parameters used for learning are not limited to this example. For example, “in which period of past training is to be performed based on training data” may be designated in the first input field 23.
- the optimization result display field 25 is a field for displaying the value of the optimized objective variable. In this example, it is assumed that “price of product 1”, “price of product 2”, and “price of product 3” are objective variables.
- the display control unit 4 displays these optimized results in the optimization result display field 25.
- the second input column 26 is a column for the user to input optimization constraint conditions.
- the content of the constraint condition is arbitrary.
- a business constraint or the like is input as the constraint condition.
- P 3 is a variable corresponding to “price of commodity 3”
- P 2 is a variable corresponding to “price of commodity 2”. Therefore, “P 3 > P 2 ” illustrated in the second input field 26 indicates a constraint condition that “the price of the product 3” is higher than the “price of the product 2”.
- the optimization unit 10 optimizes the value of each objective variable so as to satisfy this constraint condition.
- the constraint condition is not limited to the above example.
- a constraint condition “S 1 ⁇ q” may be input to the second input field 26 when the number of sales of the product 1 is S 1 and the norm is q.
- parameters used for optimization may be input to the second input field 26.
- the first instruction button 24 is a button for the user to instruct execution of learning. When the first instruction button 24 is clicked, the learning unit 3 learns the prediction model.
- the second instruction button 27 is a button for the user to instruct the execution of optimization. When the second instruction button 27 is clicked, the objective function generation unit 9 and the optimization unit 10 sequentially execute processing.
- P 1 , P 2 , P 3 , and x 1 to x 6 are explanatory variables.
- the horizontal bars represent the coefficients corresponding to the explanatory variables. Specifically, the sign of the coefficient is expressed by whether the bar is on the right side or the left side of the center line, and the absolute value of the coefficient is indicated by the length of the bar. When the bar is on the right side of the center line, the coefficient is positive. When the bar is on the left side of the center line, the coefficient is negative.
- FIG. 3 illustrates an example in which a symbol representing an explanatory variable is shown in the vicinity of the bar, but the explanatory variable corresponding to each bar may be indicated in another manner.
- the user can easily check whether important explanatory variables are missing, and whether the value of the coefficient of the important explanatory variables is an inappropriate value. This makes it easier to accurately determine the most appropriate prediction model.
- the value of the coefficient of an important explanatory variable is inappropriate, for example, the coefficient that should be positive is negative, the coefficient that should be negative is positive, This is the case when the absolute value of the coefficient is extremely large or extremely small.
- the display control unit 4 uses the test data including the past explained variable value and the explanation variable value for each prediction model.
- the value of the explanatory variable may be calculated, and the difference between the value of the explained variable and the value of the past explained variable may be visualized. Further, the display control unit 4 may visualize the difference using a scatter diagram.
- FIG. 4 is an explanatory diagram showing an example of the prediction model display field 22 including such a scatter diagram display field 22c. In the scatter diagram display field 22c, the value of the explained variable calculated from the test data is described as a predicted value, and the value of the past explained variable is described as the actual value.
- triangular markers indicate the relationship between the predicted value and the actual value calculated by the first predicted model being displayed. It is assumed that the circle marker indicates the relationship between the predicted value and the actual value calculated by the displayed second prediction model. It is assumed that the rectangular marker indicates the relationship between the predicted value and the actual value calculated by the displayed third prediction model. The closer the marker is to the broken line shown in the scatter diagram, the smaller the difference between the predicted value and the actual value. Therefore, for example, when the scatter diagram illustrated in FIG. 4 is displayed, the circular marker is close to the broken line, and thus the user can determine that the displayed second prediction model is most appropriate. When the tab selection is switched, the display control unit 4 redisplays the scatter diagram.
- the display control unit 4 may display the display field 22c of the scatter diagram illustrated in FIG. 4 in the display mode shown in FIG.
- the display control unit 4 may perform cross-validation.
- the training data storage unit 2 stores various types of training data used by the learning unit 3 for learning the prediction model.
- the training data storage unit 2 stores performance data acquired in the past for variables (object variables) that the optimization unit 10 outputs as optimization results. For example, when the optimization unit 10 intends to optimize the prices of a plurality of products, the training data storage unit 2 uses the price of each product corresponding to the explanatory variable or the explained variable as the actual data acquired in the past. The number of sales of each product corresponding to is stored.
- the training data storage unit 2 also stores external information other than the above (for example, weather and calendar information). These external information can also be explanatory variables.
- the training data storage unit 2 is realized by, for example, a magnetic disk device.
- the learning unit 3 When an operation to instruct execution of learning is performed (in this example, when the first instruction button 24 is clicked), the learning unit 3 performs a machine operation based on various types of training data stored in the training data storage unit 2. By learning, a prediction model is learned for each set explained variable. At this time, the learning unit 3 learns a plurality of types of prediction models using a plurality of types of learning algorithms for each explained variable. The learning unit 3 learns each prediction model using the parameters input in the first input field 23. For example, as described above, when “5” is designated as the number of explanatory variables used in the prediction model, the learning unit 3 sets each prediction model so that the number of explanatory variables used in the prediction model is 5. To learn.
- the prediction model learned in the present embodiment is represented by a function including a variable (object variable) output as an optimization result by the optimization unit 10 as an explanatory variable. That is, the objective variable is an explanatory variable of the prediction model.
- the objective variable is an explanatory variable of the prediction model.
- the learning unit 3 uses P 1 , P 2 , and P 3 as explanatory variables in each prediction model.
- the learning unit 3 automatically determines other explanatory variables.
- the learning unit 3 may determine an explanatory variable from various items (for example, weather) included in the external information stored in the training data storage unit 2.
- the learning unit 3 determines an explanatory variable used in the prediction model according to the number.
- the explanatory variables other than P 1 , P 2 , and P 3 may be different between the prediction models.
- the above plural types of learning algorithms are not particularly limited.
- the regression analysis, the neural network, and the support vector machine are illustrated, but the learning unit 3 may adopt the method described in Patent Literature 1 as one of the prediction model learning methods. Further, the number of types of learning algorithms is not limited.
- a set of indexes to be optimized is denoted as ⁇ m
- m 1,..., M ⁇ .
- the optimization target is the price of each product
- M corresponds to the number of products.
- the content to be predicted for each optimization target m is denoted as S m .
- S m corresponds to the number of sales of the product m.
- the contents optimized for each optimized m i.e., objective variables of optimization
- P m corresponds to the price of the product m.
- P m corresponds to the price of the product m.
- a prediction model for predicting S m is, for example, It is represented by Formula 1 illustrated.
- f d is a feature generation function and represents a transformation for P ′ m .
- D is the number of feature generation function, the number of conversion to be performed on P'm.
- the content of f d is arbitrary, and may be, for example, a function that performs linear transformation or a function that performs nonlinear transformation such as logarithm or polynomial.
- f d represents, for example, a sales response regarding the price.
- the sales response includes, for example, that the sales response improves when the price is reduced to some extent, or the response becomes worse, or the number of sales is squared in response to the price reduction.
- g d is extrinsic features (for example above, weather, etc.), D'is the number of external features. Note that the external features may be converted in advance.
- ⁇ , ⁇ , and ⁇ in Equation 1 are constant terms and coefficients of a regression equation obtained as a result of machine learning by the learning unit 3, respectively.
- the prediction model is learned based on the explained variable (S m ) and the explanatory variable (P m , various external features, etc.), and between the explained variable and the explanatory variable. It is expressed by a function of explanatory variables.
- the storage unit 6 stores each prediction model selected for each explained variable by the user, the constraint condition in the optimization input to the second input field 26, and parameters used in the optimization.
- the display control unit 4 selects each prediction model selected for each explained variable by the user,
- the storage unit 6 stores the constraint conditions and parameters (parameters used for optimization) input in the second input field 26.
- the storage unit 6 also stores external information input by the external information input unit 7.
- the storage unit 6 is realized by a magnetic disk device, for example.
- the external information input unit 7 inputs external information used for optimization other than the prediction model selected for each explained variable by the user and the constraint conditions and parameters for optimization. For example, in the above-described example, when trying to optimize the price of a certain day, the external information input unit 7 may input information regarding the weather (predicted weather) of the day. In addition, for example, when the number of visitors to the day can be predicted, the external information input unit 7 may input information related to the number of visitors to the day. As in this example of the number of customers visiting the store, the external information may be generated by a prediction model based on machine learning. The information input here is applied to explanatory variables of the prediction model, for example.
- Information input by the external information input unit 7 may be prepared in advance by the user, for example.
- the problem storage unit 8 stores a mathematical programming problem to be solved by optimization.
- the mathematical programming problem is stored in the problem storage unit 8 in advance by a user or the like. Note that when the mathematical programming problem is input to the second input field 26 and the second instruction button 27 is clicked, the display control unit 4 stores the mathematical programming problem in the problem storage unit 8. Also good.
- the problem storage unit 8 stores mathematical programming problems in advance will be described as an example.
- the problem storage unit 8 is realized by, for example, a magnetic disk device.
- the “shape” of the objective function described in the mathematical programming problem is defined so that the prediction model becomes a parameter.
- the problem storage unit 8 stores a mathematical programming problem for maximizing the total sales amount of a plurality of products.
- the optimization unit 10 optimizes the price of each product so as to maximize the total sales amount of the plurality of products.
- the sales amount of each product can be defined by the product of the price of the product and the number of sales predicted by the prediction model. Therefore, the “shape” of the objective function that represents the total sales amount of each product is expressed by the following equation 2.
- Equation 2 is the sum of the product of the number of sales that are predicted by the predictive model and price of the item illustrates as an expression, the prediction model representing a sales number S m has not been assigned. Therefore, Equation 2 is referred to as the “shape” of the objective function.
- the mathematical programming problem shown in the following formula 3 may be stored in the problem storage unit 8 in advance.
- Equation 3 represents the mathematical programming problem of maximizing the total sales of each product. Equation 3 describes the “form” of the objective function shown in Equation 2.
- the objective function generator 9 generates an objective function for the mathematical programming problem. Specifically, the objective function generation unit 9 generates an objective function of a mathematical programming problem using the prediction model as a parameter. The objective function generation unit 9 predicts the “function” (formula 2 in the above example) of the objective function described in the mathematical programming problem stored in the problem storage unit 8 for each explained variable by the user. Generate an objective function by substituting the model.
- M in Equation 2 is 3 (that is, the number of products is 3).
- the user specifically represents the explained variable for each of the explained variables “sales number S 1 of the product 1 ”, “sales number S 2 of the product 2 ”, and “sales number S 3 of the product 3 ”.
- a predictive model (function) corresponding to S 1 , S 2 , and S 3 is stored in the storage unit 6.
- the objective function generation unit 9 generates an objective variable by substituting each of these prediction models into the “form” of the objective function shown in Equation 2.
- the optimization unit 10 uses various information stored in the storage unit 6 (optimization constraints and parameters input via the second input field 26 and external information input by the external information input unit 7). Based on this, the target content is optimized. At this time, the optimization unit 10 optimizes the value of the objective variable so that the value of the objective function becomes optimal. Since the constraint condition is defined for the objective variable, etc., the optimization unit 10 optimizes the value of the objective variable so that the value of the objective function is optimized (for example, maximum, minimum, etc.) while satisfying the constraint condition. Turn into.
- the optimization unit 10 satisfies the constraints and sets the objective function value so as to maximize. Optimize variable values. Optimizing unit 10, by solving the mathematical programming problem that is specified by formula 3, to optimize price of individual products P 1, P 2 corresponding to the target variable, ..., and P M. That is, the optimization unit 10 derives the price of each product that maximizes the total sales amount of each product.
- the optimization unit 10 When the optimization unit 10 optimizes the value of each objective variable, the optimization unit 10 stores the value of each objective variable in the storage unit 6. Then, the display control unit 4 displays the optimum value of each objective variable in the optimization result display field 25 in the screen 21 displayed on the display device 5.
- the learning unit 3, the display control unit 4, the external information input unit 7, the objective function generation unit 9, and the optimization unit 10 are realized by a CPU of a computer that operates according to a program (information processing program), for example.
- the CPU reads a program from a program recording medium such as a computer program storage device (not shown in FIG. 1), and in accordance with the program, the learning unit 3, the display control unit 4, the external information input unit 7, What is necessary is just to operate
- a program information processing program
- the learning unit 3, the display control unit 4, the external information input unit 7, the objective function generation unit 9, and the optimization unit 10 may each be realized by dedicated hardware.
- the learning unit 3, the display control unit 4, the external information input unit 7, the objective function generation unit 9, and the optimization unit 10 may each be realized by an electric circuit configuration (circuitry).
- the electric circuit configuration (circuitry) is a term that conceptually includes a single device (single device), a plurality of devices (multiple devices), a chipset (chipset), or a cloud (cloud).
- the information processing system 1 of the present invention may have a configuration in which two or more physically separated devices are connected by wire or wirelessly.
- 5 and 6 are flowcharts showing an example of processing progress of the present invention.
- the display control unit 4 displays the screen 21 on the display device 5. However, in the initial state, the prediction model and the optimization result are not displayed in the prediction model display field 22 and the optimization result display field 25. The first input field 23 and the second input field 26 are blank.
- the user inputs parameters used for learning in the first input field 23.
- the display control unit 4 accepts input of parameters used for learning via the first input field 23 (step S11).
- the display control unit 4 displays the input parameters in the first input field 23.
- the information processing system 1 waits until the first instruction button 24 is clicked. During this time, the parameters input in the first input field 23 may be corrected by the user.
- the learning unit 3 When the user clicks the first instruction button 24 (Yes in step S12), the learning unit 3 is based on the parameters input in the first input field 23 and various types of training data stored in the training data storage unit 2. Then, the prediction model is learned for each preset variable to be explained (step S13). At this time, the learning unit 3 learns a plurality of types of prediction models using a plurality of types of learning algorithms for each explained variable.
- the display control unit 4 displays information on the prediction model learned for each explained variable (step S14). Since a plurality of types of prediction models are learned for each explained variable, the display control unit 4 displays a plurality of prediction models for each explained variable.
- the display control unit 4 may display information about the prediction model in the prediction model display column 22 including the tab 22a and the radio button 22b, for example, as illustrated in FIG. There is a one-to-one correspondence between each tab and each explained variable. For example, when one tab is selected by the user, the display control unit 4 displays a plurality of prediction models learned for the explained variable corresponding to the tab in the prediction model display field 22. When the selection of the tab is switched, the display control unit 4 may display a plurality of prediction models learned for the explained variable corresponding to the newly selected tab in the prediction model display field 22.
- the display control unit 4 may display information on the prediction model in the manner illustrated in FIG. Moreover, the display control part 4 may display the above-mentioned scatter diagram with a prediction model (refer FIG. 4).
- the user determines the most appropriate prediction model from the displayed prediction models, and operates the radio button 22b to select the one prediction model. Select a predictive model.
- the display control unit 4 accepts selection of one prediction model for each variable to be explained (step S15).
- the case where the user interface for the user to select one prediction model from the prediction models is a radio button, but the format of this user interface may be other than the radio button. .
- the user inputs a constraint condition for optimization in the second input field 26.
- the display control unit 4 accepts an input of constraint conditions for optimization via the second input field 26 (step S16).
- the display control unit 4 displays the input constraint conditions in the second input field 26.
- the user may input parameters used for optimization in the second input field 26.
- the display control unit 4 may accept this parameter input in the same manner, and display the parameter in the second input field 26.
- the external information input unit 7 inputs external information and stores it in the storage unit 6 (step S17).
- step S17 when the user does not click the second instruction button 27 (No in step S18), the information processing system 1 waits until the second instruction button 27 is clicked. During this time, the prediction model may be selected again by the user, or the constraint condition or the like input in the second input field 26 may be corrected by the user.
- the display control unit 4 stores the prediction model selected for each explained variable and the constraint condition input in the second input field 26. 6 (step S19).
- the display control unit 4 also stores the parameter in the storage unit 6.
- the objective function generation unit 9 reads the prediction model selected for each explained variable from the storage unit 6 and reads the mathematical programming problem stored in the problem storage unit 8.
- This mathematical programming problem is represented, for example, as shown in Equation 3, and the “form” of the objective function exemplified in Equation 2 is described in the mathematical programming problem.
- the objective function generation unit 9 generates an objective function by substituting a specific prediction model selected for each explained variable into the “form” of the objective function exemplified in Equation 2 (step S20).
- the optimization unit 10 reads the constraint conditions and external information from the storage unit 6, and derives the optimum value of each objective variable value by solving the logic planning problem that satisfies the constraint conditions (step S21). .
- the optimization unit 10 is generated in step S20 under the constraint condition.
- objective variable objective function is such that the maximum P 1, P 2, ⁇ , to derive a value of P M.
- the optimization unit 10 stores the derived optimum values of the objective variables in the storage unit 6.
- the display control unit 4 reads the optimum value of each objective variable from the storage unit 6 and displays it in the optimization result display field 25. As a result, the user can grasp the optimum value of each objective variable (for example, the price of each product that maximizes the total sales amount).
- step S14 when the user determines that there is no appropriate prediction model in the prediction model displayed in step S14, the user reviews the parameters used for learning and inputs new parameters to the first input field 23. May be. In that case, the information processing system 1 may re-execute processing from step S11. That is, the information processing system 1 can repeat learning of the prediction model until the user determines that an appropriate prediction model has been obtained for each explained variable.
- the user may review the parameters used for learning and input new parameters into the first input field 23. Good.
- the information processing system 1 may re-execute processing from step S11.
- the user may redo the operation of selecting one prediction model for each explained variable in the prediction model display field 22.
- the information processing system 1 may execute the process again from step S15.
- the user may review the constraint conditions for optimization and input new constraint conditions into the second input field 26.
- the information processing system 1 may execute the process again from step S16. That is, the information processing system 1 can repeat learning or optimization until the user determines that an appropriate value is obtained as the optimum value of each objective variable.
- the objective function generation unit 9 generates the objective function by substituting the prediction model into a predetermined “function” of the objective function. Then, the optimization unit 10 optimizes the value of the objective variable so that the value of the objective function is optimized. That is, the optimization unit 10 optimizes the value of the objective variable so that the value of the objective function is optimized for the objective function having the prediction model as an argument. Therefore, according to the present invention, optimization can be performed appropriately even in a situation where there is input data that is not observed by mathematical optimization.
- the display control unit 4 displays the prediction model learned for each explained variable in the prediction model display field 22. Therefore, the user can confirm whether or not the learned prediction model is appropriate. Therefore, for example, the user can confirm whether or not a phenomenon known empirically is reflected in the prediction model. Further, the user can examine why the obtained optimal solution is optimal after optimization.
- the learning unit 3 learns a plurality of types of prediction models for each explained variable
- the display control unit 4 displays a plurality of types of prediction models for each explained variable in the prediction model display field 22.
- the user can select a prediction model determined to be most appropriate from a plurality of prediction models for each explained variable.
- the accuracy of the optimum value of the objective variable can be improved. For example, even if it is determined that not all prediction models are appropriate for any of the explained variables, the parameters are reviewed as described above, and the learning of the prediction model is performed again by the learning unit 3. Can be made.
- the user reviews the parameters used for learning or selects one prediction model for each explained variable. It is possible to cause the information processing system 1 to execute the optimization again by redoing or reviewing the constraint conditions in the optimization.
- the learning unit 3 learns a plurality of types of prediction models using a plurality of types of learning algorithms for each explained variable.
- the learning algorithm is limited to one, and the learning unit 3 may learn one prediction model for each explained variable.
- the display control part 4 should just display the information regarding one prediction model for every to-be-explained variable (for every tab) in the prediction model display column 22.
- the display control unit 4 does not have to display the radio button 22b in the prediction model display field 22, and does not have to execute the process of step S15.
- the display control unit 4 may store the prediction model created for each explained variable in the storage unit 6.
- the learning unit 3 may learn a plurality of prediction models using the same learning algorithm. For example, the learning unit 3 learns a prediction model by a certain learning algorithm using the training data of November 2015, and uses another training model of the training data of December 2015 to acquire another prediction model by the same learning algorithm. You may learn. Further, for example, the learning unit 3 uses the same learning algorithm and the same training data to learn the prediction model with priority given to the small residual in the learning interval, and emphasizes the generalization performance for another prediction. You may learn the model.
- the display control unit 4 may receive a user editing operation on the displayed prediction model in the prediction model display field 22.
- the user can edit the prediction model displayed in the prediction model display field 22. Accordingly, the user can adjust the coefficients and constant terms of the explanatory variables of the displayed prediction model without reexamining the parameters and causing the learning unit 3 to perform learning again.
- the display control unit 4 may store the edited prediction model in the storage unit 6.
- the user interface for instructing execution of learning or optimization may be in a form other than the first instruction button 24 and the second instruction button 27.
- the display control unit 4 may display a command line user interface together with the screen 21.
- the information processing system 1 executes the processing from step S13 onward and performs a predetermined character string indicating an optimization execution instruction. Is input, the process of step S19 may be executed.
- the number of objective variables is three (“product 1 price”, “product 2 price”, and “product 3 price”), the values are to be optimized.
- the number of objective variables is not particularly limited.
- the prediction target is not limited to the product, and may be a service, for example.
- the price of a plurality of products is optimized so as to maximize the sum of the sales of the plurality of products based on the prediction of the number of sales of the plurality of products
- the types of products are distinguished by symbols A, B, C, and D.
- the sandwich group includes four types of sandwiches A, B, C, and D.
- the information processing system 1 according to the present invention is configured so that the total sales of the sandwich group, that is, the total sales of the four types of sandwiches A, B, C, and D is maximized. , B, C and D will be solved.
- the training data storage unit 2 stores the number of past sales of each sandwich and the price of each past sandwich.
- the training data storage unit 2 also stores external information such as past weather and calendar information.
- the learning unit 3 uses a plurality of types of learning algorithms for each sandwich type based on various types of training data stored in the training data storage unit 2, and uses a plurality of types of prediction models (in this example, the number of sales). Learning prediction model).
- the number of sandwiches sold is considered to be affected by the price of sandwich A itself.
- the number of sales of sandwich A is also considered to be affected by the prices of sandwiches displayed on the product shelf together with sandwich A, that is, sandwiches B, C and D. This is because the customer who visits the retail store is considered to selectively purchase a preferred sandwich from among the sandwiches A, B, C, and D displayed on the merchandise shelf at the same time.
- cannibalization is a relationship where lowering the price of a product increases the number of sales of that product, while decreasing the number of sales of other competing products (multiple products with similar properties and characteristics). It is.
- each of the prediction models has, as explanatory variables, the price P A of the sandwich A , the price P B of the sandwich B , and the price of the sandwich C. learning as a function including a price P D of P C and sandwich D.
- the learning unit 3 automatically selects appropriately from various items (for example, weather) included in the external information for each prediction model.
- the prediction model includes the explained variable (in this example, the number of sandwich sales) and the explanatory variable (in this example, the price of the sandwich of interest and the price of the competing sandwich, etc.). Based on the above, the relationship between the explained variable and the explanatory variable is shown, and is expressed by a function of the explanatory variable.
- the information processing system 1 accepts selection of one prediction model from the user for each of the sandwiches A, B, C, and D via the screen illustrated in FIG. Furthermore, the information processing system 1 accepts an input of optimization constraint conditions via the screen illustrated in FIG. 2, and the second instruction button 27 is clicked. Then, the objective function generation unit 9 uses the prediction model selected for each sandwich type in the “form” (formula 2 in this example) of the objective function described in the mathematical programming problem (formula 3 in this example). The objective function of the optimization problem is generated by substituting.
- the optimizing unit 10 sets the value of the objective variable that optimizes the objective function under the constraint condition (that is, P A , P B , to calculate the value) of the P C and P D.
- the present invention can also be applied when optimizing the prices of a plurality of products. For example, when optimizing the price of each product to maximize profit, the objective function “form” representing profit is described and a mathematical programming problem indicating that profit is maximized is stored as a problem. What is necessary is just to memorize
- the learning unit 3 generates a plurality of types of prediction models of the sales number S m of the product m of interest for each product. At this time, the learning unit 3 generates a plurality of types of prediction models including each explanatory variable indicating the position of the shelf of each product. The learning unit 3 automatically selects other explanatory variables as appropriate.
- Equation 3 The mathematical programming problem in this example is expressed by Equation 3, for example.
- the objective function generation unit 9 When the selection of one prediction model for each product is received from the user, further, the input of the optimization constraint condition is received, and the second instruction button 27 is clicked, the objective function generation unit 9 generates the objective function.
- the optimization unit 10 may obtain a value of the objective variable (the position of each product shelf) that maximizes the objective function.
- the purpose is to maximize sales or profit, so the objective function is represented by a function for calculating sales or profit.
- the objective variable for example, a setting fee for a use plan of each room of a hotel can be mentioned.
- the “sandwich” shown in the retail example corresponds to, for example, “single room breakfast plan” in this application.
- the external information includes, for example, weather, seasons, events held around the hotel, and the like.
- the objective function is represented by a function for calculating the sales or profit.
- contents considering price and stock are selected.
- a variable that indicates how much the room used in each plan will be sold at what time and as a second objective variable, at what time the room used in each plan Variables that indicate whether to sell rooms, etc.
- the external information includes, for example, weather, seasons, events performed in the vicinity of the hotel, and the like.
- each air ticket represents the route to the destination and the type (class) of the seat.
- a variable indicating how much each ticket is sold at what time and as a second objective variable, how many tickets each ticket is sold at.
- the variable to represent, etc. are mentioned.
- examples of the external information include seasons and events to be held.
- the objective function is represented by a function for calculating the sales or profit.
- the objective variable include a parking fee for each time zone and place.
- external information for example, parking fees for surrounding parking, location information (a residential area, an office district, a distance from a station, etc.) can be cited.
- the service according to the present invention can be provided in SaaS (Software as a Service) format.
- FIG. 7 is a schematic block diagram showing a configuration example of a computer according to the embodiment of the present invention.
- the computer 1000 includes a CPU 1001, a main storage device 1002, an auxiliary storage device 1003, an interface 1004, a display device 1005, and an input device 1006.
- the information processing system of the present invention is mounted on the computer 1000.
- the operation of the information processing system of the present invention is stored in the auxiliary storage device 1003 in the form of a program (information processing program).
- the CPU 1001 reads out the program from the auxiliary storage device 1003, expands it in the main storage device 1002, and executes the above processing according to the program.
- the auxiliary storage device 1003 is an example of a tangible medium that is not temporary.
- Other examples of the non-temporary tangible medium include a magnetic disk, a magneto-optical disk, a CD-ROM, a DVD-ROM, and a semiconductor memory connected via the interface 1004.
- this program is distributed to the computer 1000 via a communication line, the computer 1000 that has received the distribution may develop the program in the main storage device 1002 and execute the above processing.
- the program may be for realizing a part of the above-described processing.
- the program may be a differential program that realizes the above-described processing in combination with another program already stored in the auxiliary storage device 1003.
- FIG. 8 is a block diagram showing an outline of the information processing system of the present invention.
- the information processing system of the present invention includes a learning unit 71, a visualization unit 72, and an optimization unit 73.
- the learning unit 71 learns a prediction model that represents the relationship between the explained variable and the explanatory variable based on the explained variable and the explanatory variable and is represented by a function of the explanatory variable.
- the visualization unit 72 (for example, the display control unit 4) visualizes the prediction model.
- the optimizing unit 73 (for example, the optimizing unit 10), in response to receiving an operation from the user, the objective function having the prediction model visualized by the visualizing unit 72 as an argument under the constraint condition. An objective variable that optimizes the objective function is calculated.
- the learning unit 71 learns a plurality of types of prediction models using a plurality of types of learning algorithms for each explained variable, and the visualization unit 72 visualizes the plurality of types of prediction models for each explained variable.
- the optimization unit 73 accepts the selection of the prediction model by the user, and the optimization unit 73 optimizes the objective function under the constraint condition for the objective function having the prediction model selected for each explained variable by the user. Variables may be calculated.
- the visualization unit 72 calculates the value of the explained variable for each prediction model using the test data including the value of the past explained variable and the value of the explained variable, and the value of the explained variable and the past explained value. The difference from the value of the explanatory variable may be visualized.
- the visualization unit 72 may accept a user editing operation on the visualized prediction model.
- the learning unit 71 may learn the prediction model in response to receiving an operation for instructing learning of the prediction model from the user.
- the present invention is preferably applied to an information processing system that performs optimization based on a learned prediction model.
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Software Systems (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- Business, Economics & Management (AREA)
- Development Economics (AREA)
- Computational Linguistics (AREA)
- Strategic Management (AREA)
- Finance (AREA)
- Accounting & Taxation (AREA)
- Medical Informatics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Biomedical Technology (AREA)
- Molecular Biology (AREA)
- Entrepreneurship & Innovation (AREA)
- General Health & Medical Sciences (AREA)
- Game Theory and Decision Science (AREA)
- Biophysics (AREA)
- Economics (AREA)
- Marketing (AREA)
- General Business, Economics & Management (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
Description
上記の実施形態では、学習部3が、被説明変数毎に、複数種類の学習アルゴリズムを用いて、複数種類の予測モデルを学習する。学習アルゴリズムが1つに限定され、学習部3が、被説明変数毎に1つの予測モデルを学習してもよい。この場合、表示制御部4は、予測モデル表示欄22において、被説明変数毎に(タブ毎に)、1つの予測モデルに関する情報を表示すればよい。また、この場合、表示制御部4は、予測モデル表示欄22内にラジオボタン22bを表示しなくてよく、ステップS15の処理も実行しなくてよい。また、第1指示ボタン24がクリックされた場合、表示制御部4は、被説明変数毎に作成された予測モデルをそれぞれ記憶部6に記憶させればよい。
2 訓練データ記憶部
3 学習部
4 表示制御部
5 ディスプレイ装置
6 記憶部
7 外的情報入力部
8 問題記憶部
9 目的関数生成部
10 最適化部
Claims (7)
- 被説明変数および説明変数に基づいて、前記被説明変数と前記説明変数との間の関係を示し前記説明変数の関数で表される予測モデルを学習する学習部と、
前記予測モデルを可視化する可視化部と、
ユーザからの操作を受け付けたことに応じて、前記可視化部によって可視化された前記予測モデルを引数とする目的関数について、制約条件のもとで当該目的関数を最適化する目的変数を算出する最適化部とを備える
ことを特徴とする情報処理システム。 - 学習部は、被説明変数毎に、複数種類の学習アルゴリズムを用いて複数種類の予測モデルを学習し、
可視化部は、被説明変数毎に、前記複数種類の予測モデルを可視化し、ユーザによる予測モデルの選択を受け付け、
最適化部は、前記ユーザによって被説明変数毎に選択された予測モデルを引数とする目的関数について、制約条件のもとで当該目的関数を最適化する目的変数を算出する
請求項1に記載の情報処理システム。 - 可視化部は、過去の被説明変数の値および説明変数の値を含むテストデータを用いて、予測モデル毎に被説明変数の値を算出し、当該被説明変数の値と前記過去の被説明変数の値との差を可視化する
請求項1または請求項2に記載の情報処理システム。 - 可視化部は、可視化した予測モデルに対するユーザの編集操作を受け付ける
請求項1から請求項3のうちのいずれか1項に記載の情報処理システム。 - 学習部は、予測モデルの学習を指示する操作をユーザから受け付けたことに応じて、予測モデルを学習する
請求項1から請求項4のうちのいずれか1項に記載の情報処理システム。 - 被説明変数および説明変数に基づいて、前記被説明変数と前記説明変数との間の関係を示し前記説明変数の関数で表される予測モデルを学習し、
前記予測モデルを可視化し、
ユーザからの操作を受け付けたことに応じて、可視化された前記予測モデルを引数とする目的関数について、制約条件のもとで当該目的関数を最適化する目的変数を算出する
ことを特徴とする情報処理方法。 - コンピュータに、
被説明変数および説明変数に基づいて、前記被説明変数と前記説明変数との間の関係を示し前記説明変数の関数で表される予測モデルを学習する学習処理、
前記予測モデルを可視化する可視化処理、および、
ユーザからの操作を受け付けたことに応じて、前記可視化処理で可視化された前記予測モデルを引数とする目的関数について、制約条件のもとで当該目的関数を最適化する目的変数を算出する最適化処理
を実行させるための情報処理用プログラム。
Priority Applications (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US15/778,690 US20180336476A1 (en) | 2015-11-30 | 2016-08-29 | Information processing system, information processing method, and information processing program |
JP2017553600A JP6819607B2 (ja) | 2015-11-30 | 2016-08-29 | 情報処理システム、情報処理方法および情報処理用プログラム |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US201562260764P | 2015-11-30 | 2015-11-30 | |
US62/260,764 | 2015-11-30 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2017094207A1 true WO2017094207A1 (ja) | 2017-06-08 |
Family
ID=58796613
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/JP2016/003927 WO2017094207A1 (ja) | 2015-11-30 | 2016-08-29 | 情報処理システム、情報処理方法および情報処理用プログラム |
Country Status (3)
Country | Link |
---|---|
US (1) | US20180336476A1 (ja) |
JP (1) | JP6819607B2 (ja) |
WO (1) | WO2017094207A1 (ja) |
Cited By (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2019016279A (ja) * | 2017-07-10 | 2019-01-31 | 株式会社三菱総合研究所 | 情報処理装置及び情報処理方法 |
WO2019087526A1 (ja) * | 2017-11-02 | 2019-05-09 | 日本電気株式会社 | 評価システム、評価方法および評価用プログラム |
WO2019187289A1 (ja) * | 2018-03-30 | 2019-10-03 | 日本電気株式会社 | 評価システム、評価方法および評価用プログラム |
JP2020009502A (ja) * | 2019-10-21 | 2020-01-16 | 株式会社三菱総合研究所 | 情報処理装置及び情報処理方法 |
WO2020144820A1 (ja) * | 2019-01-10 | 2020-07-16 | 日本電気株式会社 | 属性生成装置、属性生成方法および属性生成プログラム |
JP2020113151A (ja) * | 2019-01-16 | 2020-07-27 | 株式会社東芝 | 処理水質推定装置、処理水質推定方法及びプログラム |
WO2020157913A1 (ja) * | 2019-01-31 | 2020-08-06 | 日本電気株式会社 | スケジューリング装置、スケジューリング方法、記録媒体 |
WO2021033338A1 (ja) * | 2019-08-22 | 2021-02-25 | 日本電気株式会社 | 分析システム、装置、制御方法、及びプログラム |
JPWO2021100077A1 (ja) * | 2019-11-18 | 2021-05-27 | ||
JP2021144511A (ja) * | 2020-03-12 | 2021-09-24 | 株式会社グルーヴノーツ | 情報処理装置、情報処理方法及び情報処理プログラム |
JP2021179668A (ja) * | 2020-05-11 | 2021-11-18 | Tdk株式会社 | データ解析システム、データ解析方法及びデータ解析プログラム |
WO2021260981A1 (ja) * | 2020-06-22 | 2021-12-30 | 株式会社Yamato | 情報処理装置及び情報処理方法 |
EP4421572A1 (en) | 2023-02-21 | 2024-08-28 | OMRON Corporation | Information processing apparatus, information processing method and information processing program |
Families Citing this family (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20210325343A1 (en) * | 2018-12-05 | 2021-10-21 | Revorn Co., Ltd. | Information processing apparatus, information processing method, learned model generation method, and program |
KR20210106546A (ko) | 2018-12-24 | 2021-08-30 | 디티에스, 인코포레이티드 | 딥 러닝 이미지 분석을 사용한 룸 음향 시뮬레이션 |
JP7310171B2 (ja) * | 2019-02-28 | 2023-07-19 | 富士通株式会社 | 配分方法、抽出方法、配分プログラム、抽出プログラム、配分装置及び抽出装置 |
JP6686208B1 (ja) * | 2019-04-19 | 2020-04-22 | ヤフー株式会社 | 情報処理装置、情報処理方法、及びプログラム |
JP7215324B2 (ja) * | 2019-05-17 | 2023-01-31 | 富士通株式会社 | 予測プログラム、予測方法及び予測装置 |
EP4024312A4 (en) * | 2019-08-29 | 2022-08-10 | Fujitsu Limited | SAMPLE EXTRACTION PROGRAM, DEVICE AND METHOD |
CN112700011B (zh) * | 2020-12-31 | 2024-05-31 | 第四范式(北京)技术有限公司 | 一种智能体决策信息展示方法、装置、电子设备及存储介质 |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2004336890A (ja) * | 2003-05-08 | 2004-11-25 | Hitachi Ltd | 電力売買支援システム |
JP2014180187A (ja) * | 2013-03-15 | 2014-09-25 | Toshiba Corp | 電力需要予測装置、方法及びプログラム並びに需要抑制計画策定装置 |
JP2015158812A (ja) * | 2014-02-24 | 2015-09-03 | 富士通株式会社 | 発注量決定方法、発注量決定装置および発注量決定プログラム |
-
2016
- 2016-08-29 WO PCT/JP2016/003927 patent/WO2017094207A1/ja active Application Filing
- 2016-08-29 JP JP2017553600A patent/JP6819607B2/ja active Active
- 2016-08-29 US US15/778,690 patent/US20180336476A1/en not_active Abandoned
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2004336890A (ja) * | 2003-05-08 | 2004-11-25 | Hitachi Ltd | 電力売買支援システム |
JP2014180187A (ja) * | 2013-03-15 | 2014-09-25 | Toshiba Corp | 電力需要予測装置、方法及びプログラム並びに需要抑制計画策定装置 |
JP2015158812A (ja) * | 2014-02-24 | 2015-09-03 | 富士通株式会社 | 発注量決定方法、発注量決定装置および発注量決定プログラム |
Non-Patent Citations (1)
Title |
---|
S HIGEAKI SAKURAI: "Prediction of Sales Volume of Apparel Items based on RFID Data,", THE SECOND FORUM ON DATA ENGINEERING AND INFORMATION MANAGEMENT- 2 DEIM 2010, 9 June 2010 (2010-06-09), pages 1 - 8 * |
Cited By (28)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2019016279A (ja) * | 2017-07-10 | 2019-01-31 | 株式会社三菱総合研究所 | 情報処理装置及び情報処理方法 |
US11586951B2 (en) | 2017-11-02 | 2023-02-21 | Nec Corporation | Evaluation system, evaluation method, and evaluation program for evaluating a result of optimization based on prediction |
WO2019087526A1 (ja) * | 2017-11-02 | 2019-05-09 | 日本電気株式会社 | 評価システム、評価方法および評価用プログラム |
JPWO2019087526A1 (ja) * | 2017-11-02 | 2020-11-12 | 日本電気株式会社 | 評価システム、評価方法および評価用プログラム |
US20210027109A1 (en) * | 2018-03-30 | 2021-01-28 | Nec Corporation | Evaluation system, evaluation method, and program for evaluation |
WO2019187289A1 (ja) * | 2018-03-30 | 2019-10-03 | 日本電気株式会社 | 評価システム、評価方法および評価用プログラム |
JP7044153B2 (ja) | 2018-03-30 | 2022-03-30 | 日本電気株式会社 | 評価システム、評価方法および評価用プログラム |
JPWO2019187289A1 (ja) * | 2018-03-30 | 2021-03-11 | 日本電気株式会社 | 評価システム、評価方法および評価用プログラム |
JP7163975B2 (ja) | 2019-01-10 | 2022-11-01 | 日本電気株式会社 | 属性生成装置、属性生成方法および属性生成プログラム |
JPWO2020144820A1 (ja) * | 2019-01-10 | 2021-11-18 | 日本電気株式会社 | 属性生成装置、属性生成方法および属性生成プログラム |
WO2020144820A1 (ja) * | 2019-01-10 | 2020-07-16 | 日本電気株式会社 | 属性生成装置、属性生成方法および属性生成プログラム |
JP2020113151A (ja) * | 2019-01-16 | 2020-07-27 | 株式会社東芝 | 処理水質推定装置、処理水質推定方法及びプログラム |
JP7263020B2 (ja) | 2019-01-16 | 2023-04-24 | 株式会社東芝 | 処理水質推定装置、処理水質推定方法及びプログラム |
WO2020157913A1 (ja) * | 2019-01-31 | 2020-08-06 | 日本電気株式会社 | スケジューリング装置、スケジューリング方法、記録媒体 |
JPWO2020157913A1 (ja) * | 2019-01-31 | 2021-11-11 | 日本電気株式会社 | スケジューリング装置、スケジューリング方法、プログラム |
JP7322899B2 (ja) | 2019-01-31 | 2023-08-08 | 日本電気株式会社 | スケジューリング装置、スケジューリング方法、プログラム |
JPWO2021033338A1 (ja) * | 2019-08-22 | 2021-02-25 | ||
WO2021033338A1 (ja) * | 2019-08-22 | 2021-02-25 | 日本電気株式会社 | 分析システム、装置、制御方法、及びプログラム |
JP7371690B2 (ja) | 2019-08-22 | 2023-10-31 | 日本電気株式会社 | 分析システム、装置、制御方法、及びプログラム |
JP7012696B2 (ja) | 2019-10-21 | 2022-01-28 | 株式会社三菱総合研究所 | 情報処理装置及び情報処理方法 |
JP2020009502A (ja) * | 2019-10-21 | 2020-01-16 | 株式会社三菱総合研究所 | 情報処理装置及び情報処理方法 |
JP7334796B2 (ja) | 2019-11-18 | 2023-08-29 | 日本電気株式会社 | 最適化装置、最適化方法、プログラム |
JPWO2021100077A1 (ja) * | 2019-11-18 | 2021-05-27 | ||
JP2021144511A (ja) * | 2020-03-12 | 2021-09-24 | 株式会社グルーヴノーツ | 情報処理装置、情報処理方法及び情報処理プログラム |
JP2021179668A (ja) * | 2020-05-11 | 2021-11-18 | Tdk株式会社 | データ解析システム、データ解析方法及びデータ解析プログラム |
JP7456273B2 (ja) | 2020-05-11 | 2024-03-27 | Tdk株式会社 | データ解析システム、データ解析方法及びデータ解析プログラム |
WO2021260981A1 (ja) * | 2020-06-22 | 2021-12-30 | 株式会社Yamato | 情報処理装置及び情報処理方法 |
EP4421572A1 (en) | 2023-02-21 | 2024-08-28 | OMRON Corporation | Information processing apparatus, information processing method and information processing program |
Also Published As
Publication number | Publication date |
---|---|
US20180336476A1 (en) | 2018-11-22 |
JPWO2017094207A1 (ja) | 2018-09-13 |
JP6819607B2 (ja) | 2021-01-27 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2017094207A1 (ja) | 情報処理システム、情報処理方法および情報処理用プログラム | |
Kaipia et al. | Information sharing for sales and operations planning: Contextualized solutions and mechanisms | |
JP6856023B2 (ja) | 最適化システム、最適化方法および最適化プログラム | |
JP6848884B2 (ja) | 最適化システム、最適化方法、及び、プログラム | |
Mendes | Demand driven supply chain: A structured and practical roadmap to increase profitability | |
US20150254589A1 (en) | System and Method to Provide Inventory Optimization in a Multi-Echelon Supply Chain Network | |
Xia | Competitive strategies and market segmentation for suppliers with substitutable products | |
WO2017056367A1 (ja) | 情報処理システム、情報処理方法および情報処理用プログラム | |
US20210158259A1 (en) | Orchestrated intelligent supply chain optimizer | |
US20130290072A1 (en) | Automated Planning, Value Calculation And Decision Optimization System And Method | |
US10803415B2 (en) | Method of determining optimal business metrics from a product mix constrained by at least physical shelf space and at least one business rule | |
JP4887846B2 (ja) | 出荷計画システム | |
US20200211040A1 (en) | Method of determining the cost of implementing a pricing strategy formed by changeable business rules | |
Parsa et al. | Quantifying the benefits of continuous replenishment program for partner evaluation | |
JP6791151B2 (ja) | 最適化システム、最適化方法および最適化プログラム | |
Baykasoğlu et al. | Revenue management for make-to-order manufacturing systems with a real-life application | |
Khalili-Damghani et al. | Solving a multi-objective multi-echelon supply chain logistic design and planning problem by a goal programming approach | |
Liu et al. | A methodology to assess the supply chain performance based on gap-based measures | |
JP2005056024A (ja) | 供給計画を作成するための支援システムおよび支援方法および支援のためのコンピュータプログラム | |
JP4335223B2 (ja) | 資産運用シミュレーション装置及びプログラム | |
CN114626660A (zh) | 用于激增调整预报的方法和装置 | |
JP2021039542A (ja) | 情報処理方法および情報処理装置 | |
Molamohamadi et al. | Ordering Policies of a Deteriorating Item in an EOQ Model under Upstream Partial Order‐Quantity‐Dependent Trade Credit and Downstream Full Trade Credit | |
JP6424568B2 (ja) | 発注量決定プログラム、発注量決定方法および発注量決定装置 | |
Avigad et al. | Supporting the selection of robust engineering concepts under suppliers related uncertainties |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 16870150 Country of ref document: EP Kind code of ref document: A1 |
|
ENP | Entry into the national phase |
Ref document number: 2017553600 Country of ref document: JP Kind code of ref document: A |
|
WWE | Wipo information: entry into national phase |
Ref document number: 15778690 Country of ref document: US |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
122 | Ep: pct application non-entry in european phase |
Ref document number: 16870150 Country of ref document: EP Kind code of ref document: A1 |