Inventory prediction system, inventory calculation device, inventory prediction method, and program

Abstract

An inventory prediction system (100) for predicting demand by using the hierarchical Bayesian method and calculating a product inventory amount is equipped with: a demand prediction device (1) for generating a prior distribution for each unit of each class into which products are classified, calculating a demand prediction probability distribution, which is the posterior distribution for each unit of each class of products, on the basis of the prior distribution, and generating demand prediction data which includes the demand prediction probability distribution for the smallest unit of the product; a supply LT prediction device (2) for calculating the supply lead time probability distribution for the smallest unit of the product, and generating supply LT prediction data which includes the supply lead time probability distribution for the smallest unit of the product; and a component inventory calculation device (4) for estimating the probability distribution of the inventory amount of the smallest unit of the product on the basis of the demand prediction data and the supply LT prediction data, and generating inventory amount prediction data, which includes the probability distribution of the inventory amount for the smallest unit of the product.

Description

Technical Field

[0001] The present disclosure relates to an inventory forecasting system, an inventory calculation device, an inventory forecasting method, and a program.

Background Art

[0002] In business activities, it is important to control the supply-demand balance so as not to hold excessive inventory while shortening the demand lead time and coping with demand fluctuations. Demand forecasting is performed as one of the means. In the conventional demand forecasting method, various calculations were performed using inferential statistics (hereinafter referred to as conventional statistics) that are generally used. In conventional statistics, a statistic (such as an average or a standard deviation) of a certain fixed parameter is calculated as a unique value based on data extracted from a sample.

[0003] In addition, for products and parts subject to inventory management, in order to prepare for fluctuations in the required quantity during the period (supply lead time) from the request for inventory replenishment to the actual replenishment, inventory replenishment operations are performed after holding a safety stock. A method of calculating the safety stock based on the variation in the required quantity over several years, the supply lead time, and a safety factor is generally used. Hereinafter, the supply lead time is abbreviated as supply LT. For example, a general formula for calculating the safety stock is: Safety stock = Safety factor × Standard deviation of required quantity × √(Supply LT + Order interval).

[0004] Patent Document 1 discloses an inventory management device that probabilistically forecasts future demand for each of a plurality of businesses constituting a business group with a high correlation between demand and supply. In the technology described in Patent Document 1, the trend variation of performance is modeled from past performance data by regression analysis using conventional statistics. Further, the business cycle variation is calculated by subtracting the trend variation from the total variation and modeled by a stochastic differential equation. A demand forecasting probability distribution is calculated using these models.

[0005] Patent Document 2 discloses a program for estimating the optimal order quantity of an intermittent demand product, which estimates the order quantity of a necessary product in order to suppress the probability of stockouts to a predetermined value or less. This program calculates the demand cumulative value distribution by performing a Monte Carlo simulation using a demand interval frequency distribution, which probabilistically represents the frequency of demand, and actual demand. From the demand cumulative distribution, which represents the demand forecast value, the order quantity of a necessary product is estimated, taking into account the lead time, in order to suppress the occurrence of stockouts. [Prior art documents] [Patent documents]

[0006] [Patent Document 1] JP 2004-326346 A [Patent Document 2] JP 2003-316938 A Summary of the Invention [Problem to be solved by the invention]

[0007] Inventory replenishment operations with a general safety stock have the following two problems. The first is that because safety stock needs to be calculated based on long-term performance information, practical safety stock cannot be calculated when there is a lack of sample data or when market conditions are changing rapidly. The second is that the supply LT is a fixed value regardless of individual supply conditions, so there are cases where it does not match the actual situation. These problems cannot be solved even if the demand forecasting technology described in Patent Document 1 is used and the order amount is estimated using the appropriate order amount estimation program described in Patent Document 2 to perform inventory replenishment operations.

[0008] The present disclosure has been made to solve the problems described above, and aims to realize an inventory replenishment operation that is suitable for actual conditions and can be put into practical use even when there is a shortage of sample data or when market conditions are changing rapidly. [Means for solving the problem]

[0009] In order to achieve the above object, an inventory prediction system according to the present disclosure is an inventory prediction system that performs demand prediction using the hierarchical Bayesian method and calculates the inventory level of an item. The inventory prediction system includes a demand prediction device, a supply LT prediction device, and an inventory calculation device. The demand prediction device generates a prior distribution for each unit of each hierarchy obtained by classifying items, calculates a demand prediction probability distribution that is a posterior distribution for each unit of each hierarchy of the item based on the prior distribution, and generates demand prediction data including the demand prediction probability distribution of the item at the minimum unit. The supply LT prediction device calculates a probability distribution of the supply lead time of the item at the minimum unit and generates supply LT prediction data including the probability distribution of the supply lead time of the item at the minimum unit. The inventory calculation device estimates a probability distribution of the inventory level of the item at the minimum unit based on the demand prediction data and the supply LT prediction data, and generates inventory level prediction data including the probability distribution of the inventory level of the item at the minimum unit.

Advantages of the Invention

[0010] According to the present disclosure, by using the hierarchical Bayesian method as a method for predicting the demand of an item, demand prediction can be performed even when the data serving as samples in the lower hierarchy is insufficient, and since the supply LT is handled as a probability distribution based on actual results, it is possible to predict the inventory level of an item closer to the actual situation. When the data serving as samples is insufficient or when the market situation changes rapidly, etc., it can be put into practical operation, and an inventory replenishment operation that suits the actual situation can be realized.

Brief Description of the Drawings

[0011]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

Figure 12

Figure 13

Figure 14

Figure 15

Figure 16

Figure 17

Figure 18

Embodiments for Carrying Out the Invention

[0012] Hereinafter, the inventory prediction system, inventory calculation device, inventory prediction method, and program according to the present embodiment will be described in detail with reference to the drawings. In the drawings, the same or corresponding parts are denoted by the same reference numerals. In the present embodiment, the hierarchical Bayesian method is used as a method for predicting the demand for products for each unit of each hierarchy obtained by classifying products. Also, demand prediction is performed on a daily basis, and the inventory levels of products and parts for each day are calculated.

[0013] The configuration of the inventory prediction system 100 according to the embodiment will be described with reference to FIG. 1. The inventory prediction system 100 includes a demand prediction device 1 that calculates a demand prediction probability distribution, which is a posterior distribution, based on a prior distribution, a supply LT prediction device 2 that calculates a probability distribution of the supply LT of products and parts, a product inventory calculation device 3 that calculates a probability distribution of the inventory level of a product based on the demand prediction probability distribution and the probability distribution of the supply LT of the product, and a component inventory calculation device 4 that calculates a probability distribution of the inventory level of a component based on the probability distribution of the inventory level of the product and the probability distribution of the supply LT of the component.

[0014] The functional configuration of the demand prediction device 1 will be described with reference to FIG. 2. The demand prediction device 1 includes a shipping plan data storage unit 11 that stores shipping plan data indicating the shipping plan for each unit of each layer of the product, an error data storage unit 12 that stores error data indicating the error between the shipping plan and the actual shipping results for each unit of each layer of the product, a statistical model storage unit 13 that stores a statistical model representing the relationship between the shipping plan data and the error data, and a prior distribution generation unit 14 that generates a prior distribution for each unit of each layer of the product using the shipping plan data, the error data, and the statistical model. Further, the demand prediction device 1 includes a demand model storage unit 15 that stores a demand model representing the demand trend of the product for each unit of each layer of the product, a demand actual result data storage unit 16 that stores demand actual result data indicating the actual demand of the product for each unit of each layer of the product, a product group master storage unit 17 that stores a product group master indicating the layers into which the product is classified, and a demand prediction data generation unit 18 that calculates a demand prediction probability distribution, which is a posterior distribution for each unit of each layer of the product, using the prior distribution data, the demand model, the demand actual result data, and the product group master, and generates demand prediction data including the demand prediction probability distribution, and a demand prediction data output unit 19 that outputs the demand prediction data.

[0015] The layers into which the product is classified are, for example, such that the lowest unit (minimum unit) is an SKU (Stock Keeping Unit), and the highest unit (maximum unit) is a group of ABC analysis results with features related to the product such as delivery date, compliance rate, production lot number, unit price, etc. as evaluation axes. The product group master is an example of an article group master.

[0016] For the shipping plan data, it is advisable to use the shipping plan of the product formulated by an expert. Thereby, the know-how of the expert that has not been digitized can be utilized for demand prediction, and an improvement in the accuracy of demand prediction can be expected. The shipping plan and the actual shipping results are examples of the demand plan and the actual demand, respectively. The shipping plan data is an example of the demand plan data. The shipping plan data and the error data are generated for each unit of each layer of the product.

[0017] The prior distribution generation unit 14 generates a prior distribution for each unit of each layer of the product using the shipment plan data, error data, and statistical model for each unit of each layer of the product. The prior distribution generation unit 14 sends prior distribution data indicating the generated prior distribution for each unit of each layer of the product to the demand prediction data generation unit 18. A method for generating the prior distribution executed by the prior distribution generation unit 14 will be specifically described. For example, assume that the daily planned quantity of the shipment plan for a product with a model name of M1 is 100 units. Assume that the statistical model representing the relationship between the shipment plan data and the error data for the product with the model name of M1 is a normal distribution with a standard deviation of 20. The prior distribution generation unit 14 generates a normal distribution with a mean of 100 (units) and a standard deviation of 20 as the prior distribution for the product with the model name of M1.

[0018] When the demand prediction data generation unit 18 receives the prior distribution data for each unit of each layer of the product from the prior distribution generation unit 14, it calculates the likelihood based on the demand actual performance data and the demand model for each unit of each layer of the product up to the previous day. The demand prediction data generation unit 18 multiplies the likelihood by the prior distribution indicated by the prior distribution data and executes MCMC (Markov chain Monte Carlo methods) sampling to calculate the demand prediction probability distribution, which is the posterior distribution for each unit of each layer of the product. The demand prediction data generation unit 18 generates demand prediction data including the demand prediction probability distribution for the product at the minimum unit.

[0019] A method for calculating the demand prediction probability distribution executed by the demand prediction data generation unit 18 will be specifically described. For example, assume that the prior distribution for a product with a model name of M1 is a probability distribution of the average λ of the number of units shipped per day with an error of ±20% centered around 100 units. Assume that the average number of units shipped per day up to the previous day indicated by the demand actual performance data is 90 units / day. The demand prediction data generation unit 18 calculates the likelihood based on this demand actual performance data and the demand model. If the demand model is a Poisson distribution and the likelihood is a probability distribution of the average λ centered around 90 units, the demand prediction data generation unit 18 multiplies the prior distribution and the likelihood to calculate the demand prediction probability distribution, which is the posterior distribution. At this time, MCMC sampling is performed to facilitate numerical calculation.

[0020] The demand prediction data generation unit 18 refers to the product group master and uses the hierarchical Bayesian method to estimate the posterior distribution of the lower layer based on the posterior distribution of the upper layer. Therefore, even when the demand performance data of the lower layer is insufficient, by using the posterior distribution of the upper layer and the prior distribution of the lower layer, a practical demand prediction probability distribution with a not-too-wide tail of the probability distribution can be obtained.

[0021] Here, the demand prediction probability distribution for each unit of each layer of the product will be described with reference to FIG. 3. FIG. 3 shows the demand prediction probability distributions of models M1 to M10. Model M1 includes products with model names T1 to T10, and model M10 includes products with model names T91 to T100. As shown in FIG. 3, letting T be the model name, M be the model name, and S[T,t] be the number of units sold by model name on the t-th day, S[T,t] follows a Poisson distribution with μ[T] as the mean. μ[T] follows a normal distribution with ν[M] as the mean and σ[M]2 as the variance. ν[M] follows a normal distribution with ξ as the mean and δ2 as the variance.

[0022] Returning to FIG. 2, the demand prediction data output unit 19 outputs the demand prediction data generated by the demand prediction data generation unit 18 to the product inventory calculation device 3. Note that the demand prediction data output unit 19 may be configured to generate, for example, demand prediction data for browsing from the demand prediction data, display it on the screen, or transmit it to the terminal used by the user. In this case, the demand prediction data output unit 19 performs MCMC sampling on the demand prediction probability distribution indicated by the demand prediction data, generates and outputs the demand prediction data for browsing. For the output method of the demand prediction data for browsing, for example, a spreadsheet application or a BI (Business Intelligence) tool may be used, or a new user interface system may be used.

[0023] A method for generating demand prediction data for browsing will be described with reference to FIGS. 4 to 6. FIG. 4 shows the demand prediction probability distributions of model names T1 to T3. In the example of FIG. 4, the sales volumes of the products with model names T1 and T2 belonging to product group G-1 follow a normal distribution. For example, the average of the normal distribution of the sales volume of model name T1 follows a normal distribution with an average of 90 and a standard deviation of 20, and the standard deviation follows a normal distribution with an average of 15 and a standard deviation of 5. In the example of FIG. 5, the sales volumes of the products with model names T1 to T3 are sampled 1000 times each by MCMC sampling. In the example of FIG. 6, the values of each sampling NO. (No. in the figure) 1 to 1000 are summed up for each product group. When the demand prediction data output unit 19 generates demand prediction data for browsing in terms of model names, for example, a table as shown in FIG. 5 is generated. When the demand prediction data output unit 19 generates demand prediction data for browsing in terms of product groups, for example, a table as shown in FIG. 6 is generated. Alternatively, the demand prediction data output unit 19 may output a graph representing the values of sampling NO. 1 to 1000 in FIGS. 5 and 6 in a distribution group. Thereby, the user can visually grasp the demand prediction data.

[0024] In the examples of FIGS. 4 to 6, the method for generating demand prediction data for browsing in the case of a two - layer structure of model names and product groups has been described. However, for example, the same applies to cases with three or more layers such as model names, model types, and product groups.

[0025] Here, the functional configuration of the supply LT prediction device 2 will be described with reference to FIG. 7. The supply LT prediction device 2 includes a product supply performance data storage unit 21 that stores product supply performance data indicating the supply performance including the supply LT of the product in the minimum unit, a component supply performance data storage unit 22 that stores component supply performance data indicating the supply performance including the supply LT of the component, a supply LT prediction data generation unit 23 that generates product supply LT prediction data including the probability distribution of the supply LT of the product in the minimum unit based on the product supply performance data, and generates component supply LT prediction data including the probability distribution of the supply LT of the component based on the component supply performance data, and a supply LT prediction data output unit 24 that outputs the product supply LT prediction data and the component supply LT prediction data. The supply LT of the product is the period from when the production of the product is determined until it is stocked. The supply LT of the component is the period from when the component is ordered until it is stocked. The product supply performance data and the component supply performance data are examples of supply performance data.

[0026] The supply LT prediction data generation unit 23 calculates the probability distribution of the supply LT of the product in the minimum unit based on the past actual values of the supply LT of the product in the minimum unit included in the product supply performance data, and generates product supply LT prediction data including the probability distribution of the supply LT of the product in the minimum unit. The supply LT prediction data generation unit 23 calculates the probability distribution of the supply LT of the component based on the past actual values of the supply LT of the component included in the component supply performance data, and generates component supply LT prediction data including the probability distribution of the supply LT of the component. For the probability distribution of the supply LT of the product and the probability distribution of the supply LT of the component, an interval such as from the 5% point to the 95% point can be defined. The supply LT prediction data output unit 24 outputs the product supply LT prediction data including the probability distribution of the supply LT of the product to the product inventory calculation device 3. The supply LT prediction data output unit 24 outputs the component supply LT prediction data including the probability distribution of the supply LT of the component to the component inventory calculation device 4.

[0027] Note that the supply LT prediction data output unit 24 may be configured to generate, for example, product supply LT prediction data for viewing from the product supply LT prediction data, generate component supply LT prediction data for viewing from the component supply LT prediction data, display it on the screen, or transmit it to the terminal used by the user. In this case, the supply LT prediction data output unit 24 performs MCMC sampling on the probability distribution indicated by the product supply LT prediction data to generate and output the product supply LT prediction data for viewing. The supply LT prediction data output unit 24 performs MCMC sampling on the probability distribution indicated by the component supply LT prediction data to generate and output the component supply LT prediction data for viewing. As the output method of the product supply LT prediction data for viewing and the component supply LT prediction data for viewing, for example, a spreadsheet application or a BI (Business Intelligence) tool may be used, or a new user interface system may be used.

[0028] The supply LT prediction device 2 may be configured to output an alert regarding the delivery date of the product based on the product supply LT prediction data and the component supply LT prediction data. The functional configuration of the supply LT prediction device 2 in this case will be described with reference to FIG. 8. In the example of FIG. 8, in addition to the product supply actual data storage unit 21, the component supply actual data storage unit 22, the supply LT prediction data generation unit 23, and the supply LT prediction data output unit 24, the supply LT prediction device 2 includes a supply LT alert condition storage unit 25 that stores a supply LT alert condition, which is a condition for outputting a predetermined alert regarding the delivery date of the product, for the probability distribution of the supply LT of the product in the minimum unit indicated by the product supply LT prediction data and the probability distribution of the supply LT of the component indicated by the component supply LT prediction data, and a supply LT alert unit 26 that outputs an alert when the supply LT alert condition is satisfied. The alert may be output, for example, by being displayed on the screen or transmitted to the terminal used by the user.

[0029] For example, the supply LT alert condition storage unit 25 stores product current inventory information indicating the current inventory level of the product, component current inventory information indicating the current inventory level of the components, component configuration information indicating the component configuration of the product, delivery date information indicating the delivery date of the product, and operating day information indicating the operating days. The shortest possible product supply date is set as a supply LT alert condition when the delivery date information of the product cannot meet the delivery date of the product.

[0030] The supply LT alert unit 26 calculates the shortest possible product supply date based on the product current inventory information, component current inventory information, maximum value of the supply LT of the minimum unit of the product, maximum value of the supply LT of the components, and operating day information stored in the supply LT alert condition storage unit 25, and determines whether the shortest possible product supply date can meet the delivery date of the product indicated by the delivery date information. When the supply LT alert unit 26 determines that the shortest possible product supply date cannot meet the delivery date of the product indicated by the delivery date information, it outputs an alert. When the supply LT alert unit 26 outputs an alert, the user can know the possibility that the delivery date of the product cannot be met, and for example, can adjust the production plan of the product to respond.

[0031] Next, the functional configuration of the product inventory calculation device 3 will be described with reference to FIG. 9. The product inventory calculation device 3 includes a demand prediction data acquisition unit 31 that acquires demand prediction data from the demand prediction device 1, a product supply LT prediction data acquisition unit 32 that acquires product supply LT prediction data from the supply LT prediction device 2, a product inventory performance data storage unit 33 that stores product inventory performance data indicating the inventory performance of the product in the minimum unit, an instruction remaining data storage unit 34 that stores instruction remaining data indicating the quantity of the product in the minimum unit for which production has been determined but not yet recorded in the product inventory performance, a product inventory quantity prediction data generation unit 35 that calculates the probability distribution of the inventory quantity of the product in the minimum unit based on the product inventory performance data, the instruction remaining data, the demand prediction data, and the product supply LT prediction data, and generates product inventory quantity prediction data including the probability distribution of the inventory quantity of the product in the minimum unit, and a product inventory quantity prediction data output unit 36 that outputs the product inventory quantity prediction data. The product inventory quantity prediction data is an example of inventory quantity prediction data. The instruction remaining data is an example of inventory replenishment scheduled data. The product inventory performance data is an example of inventory performance data. The product inventory quantity prediction data generation unit 35 is an example of an inventory quantity prediction data generation unit.

[0032] The product inventory quantity prediction data generation unit 35 executes MCMC sampling on the demand prediction probability distribution indicated by the demand prediction data and the probability distribution indicated by the product supply LT prediction data, and estimates the probability distribution of the inventory quantity of the product in the minimum unit by day based on the product inventory performance data and the instruction remaining data. The product inventory quantity prediction data generation unit 35 generates product inventory quantity prediction data including the probability distribution of the inventory quantity of the product in the minimum unit by day. For the probability distribution of the inventory quantity of the product in the minimum unit by day, an interval such as from the 5% point to the 95% point can be defined.

[0033] The product inventory quantity prediction data output unit 36 outputs the product inventory quantity prediction data including the probability distribution of the inventory quantity of the product in the minimum unit by day to the component inventory calculation device 4. Further, the product inventory quantity prediction data output unit 36 may be configured to generate, for example, browsing product inventory quantity prediction data from the product inventory quantity prediction data, display it on a screen, or transmit it to a terminal used by the user. This facilitates the operation of inventory replenishment of products that can be actually operated.

[0034] For the method of outputting product inventory prediction data, for example, a spreadsheet application or a BI tool may be used, or a new user interface system may be used. Also, by displaying the product inventory prediction data in an interval from the defined 5% point to the 95% point, etc., it becomes easier for the user to grasp the trend of the product inventory.

[0035] Here, with reference to FIG. 10, an output screen of product inventory prediction data in which the product inventory prediction data is displayed in the interval from the 5% point to the 95% point will be described. In the graph displayed on the output screen shown in FIG. 10, the inventory quantity y of the product up to today is represented by a solid line. After today, the transition of the 5% point of the product inventory quantity y is represented by graph g1, the transition of the 50% point of the product inventory quantity y is represented by graph g2, and the transition of the 95% point of the product inventory quantity y is represented by graph g3. Thus, the user can visually grasp the trend of the product inventory. For example, by performing inventory replenishment on day d0, which is obtained by subtracting the 95% point value of the supply LT (the longest value of the supply LT in the interval from the 5% point to the 95% point) from day d1 (the day when the inventory quantity y of graph g1 becomes 0) when the inventory quantity y of the product may become 0 at the shortest within the interval from the 5% point to the 95% point, an operation of inventory replenishment that is operationally feasible and in line with the actual situation can be realized.

[0036] The product inventory calculation device 3 may be configured to output an alert regarding the inventory level of the product based on the product inventory quantity prediction data. The functional configuration of the product inventory calculation device 3 in this case will be described with reference to FIG. 11. In the example of FIG. 11, in addition to a demand prediction data acquisition unit 31, a product supply LT prediction data acquisition unit 32, a product inventory actual result data storage unit 33, an instruction remaining data storage unit 34, a product inventory quantity prediction data generation unit 35, and a product inventory quantity prediction data output unit 36, the product inventory calculation device 3 includes a product inventory alert condition storage unit 37 that stores a product inventory alert condition, which is a condition for outputting an alert regarding the ratio of supplying a product to a predetermined demand with respect to the probability distribution of the daily product inventory quantity indicated by the product inventory quantity prediction data, and a product inventory alert unit 38 that outputs an alert when the product inventory alert condition is satisfied. The method of outputting the alert may be, for example, screen display or transmission to the terminal used by the user. The product inventory alert condition is an example of an inventory alert condition. The product inventory alert unit 38 is an example of an inventory alert unit.

[0037] For example, the product inventory alert condition stored in the product inventory alert condition storage unit 37 is that the product service rate, which is the ratio of supplying a product to the demand, becomes 90% or less. The product inventory alert unit 38 calculates the product service rate based on the demand prediction data and the product inventory quantity prediction data, and determines whether the product service rate becomes 90% or less. When the product inventory alert unit 38 determines that the product service rate becomes 90% or less, it outputs an alert. When the product inventory alert unit 38 outputs an alert, the user can know the possibility that the product service rate becomes 90% or less for the production of the product, and can take countermeasures, for example, by adjusting the production plan.

[0038] Next, the functional configuration of the component inventory calculation device 4 will be described with reference to FIG. 12. The component inventory calculation device 4 includes a product inventory quantity prediction data acquisition unit 41 that acquires product inventory quantity prediction data from the product inventory calculation device 3, a component supply LT prediction data acquisition unit 42 that acquires component supply LT prediction data from the supply LT prediction device 2, a component configuration master storage unit 43 that stores a component configuration master indicating the component configuration of a product, a component inventory performance data storage unit 44 that stores component inventory performance data indicating the inventory performance of components, a scheduled receipt data storage unit 45 that stores scheduled receipt data indicating the quantity of components for which an order has been confirmed but not recorded in the component inventory performance, and a component inventory quantity prediction data generation unit 46 that calculates the probability distribution of the component inventory quantity based on the product inventory quantity prediction data, the component inventory performance data, the scheduled receipt data, the component supply LT prediction data, and the component configuration master, and generates component inventory quantity prediction data including the probability distribution of the component inventory quantity. The component inventory quantity prediction data is an example of inventory quantity prediction data. The scheduled receipt data is an example of inventory replenishment schedule data. The component inventory performance data is an example of inventory performance data. The component inventory quantity prediction data generation unit 46 is an example of an inventory quantity prediction data generation unit.

[0039] The component inventory quantity prediction data generation unit 46 executes MCMC sampling on the probability distribution indicated by the product inventory quantity prediction data and the probability distribution indicated by the component supply LT prediction data, and estimates the probability distribution of the daily component inventory quantity based on the component configuration master, the component inventory performance data, and the scheduled receipt data. An increase in the product inventory quantity means the production of the product, that is, the demand for components. The component inventory quantity prediction data generation unit 46 generates component inventory quantity prediction data including the probability distribution of the daily component inventory quantity. For the probability distribution of the daily component inventory quantity, an interval such as from the 5% point to the 95% point can be defined.

[0040] The component inventory quantity prediction data output unit 47 outputs component inventory quantity prediction data including the probability distribution of the daily component inventory quantity. The component inventory quantity prediction data output unit 47 generates, for example, browsing-use component inventory quantity prediction data from the component inventory quantity prediction data, and displays it on the screen or transmits it to the terminal used by the user. Thereby, the operation of replenishing the inventory of components that can be actually operated becomes easy.

[0041] As a method for outputting the component inventory quantity prediction data, for example, a spreadsheet application or a BI tool may be used, or a new user interface system may be used. Also, by displaying the component inventory quantity prediction data in an interval from the defined 5% point to the 95% point, etc., it becomes easier for the user to grasp the transition of the component inventory quantity.

[0042] The component inventory calculation device 4 may be configured to output an alert regarding the component inventory quantity based on the component inventory quantity prediction data. The functional configuration of the component inventory calculation device 4 in this case will be described with reference to FIG. 13. In the example of FIG. 13, in addition to the product inventory quantity prediction data acquisition unit 41, the component supply LT prediction data acquisition unit 42, the component configuration master storage unit 43, the component inventory actual result data storage unit 44, the incoming scheduled data storage unit 45, the component inventory quantity prediction data generation unit 46, and the component inventory quantity prediction data output unit 47, the component inventory calculation device 4 stores a component inventory alert condition storage unit 48 that stores a component inventory alert condition, which is a condition for outputting an alert regarding the ratio of supplying components for the production of a predetermined product with respect to the probability distribution of the daily component inventory quantity indicated by the component inventory quantity prediction data, and a component inventory alert unit 49 that outputs an alert when the component inventory alert condition is satisfied. The method of outputting the alert may be, for example, displaying it on the screen or transmitting it to the terminal used by the user. The component inventory alert condition is an example of an inventory alert condition. The component inventory alert unit 49 is an example of an inventory alert unit.

[0043] For example, the component inventory alert condition stored in the component inventory alert condition storage unit 48 is set such that the component service rate, which is the ratio of supplying components for product production, becomes 90% or less. The component inventory alert unit 49 calculates the component service rate based on the product inventory quantity prediction data and the component inventory quantity prediction data, and determines whether the component service rate becomes 90% or less. When the component inventory alert unit 49 determines that the component service rate becomes 90% or less, it outputs an alert. When the component inventory alert unit 49 outputs an alert, the user can know the possibility that the component service rate becomes 90% or less, and can take countermeasures, for example, by adjusting the order point of the components.

[0044] Subsequently, the flow of each process executed in the inventory prediction system 100 will be described with reference to FIGS. 14 to 17. The demand prediction process shown in FIG. 14 starts, for example, when an instruction to calculate the inventory quantity is input to the inventory prediction system 100. When an instruction to calculate the inventory quantity is input to the inventory prediction system 100, the prior distribution generation unit 14 of the demand prediction device 1 uses the shipment plan data for each unit of each hierarchy of the product stored in the shipment plan data storage unit 11, the error data stored in the error data storage unit 12, and the statistical model stored in the statistical model storage unit 13 to generate a prior distribution for each unit of each hierarchy of the product (step S11). The prior distribution generation unit 14 sends prior distribution data indicating the generated prior distribution for each unit of each hierarchy of the product to the demand prediction data generation unit 18.

[0045] When the demand prediction data generation unit 18 receives the prior distribution data for each unit of each hierarchy of the product from the prior distribution generation unit 14, it calculates the likelihood based on the demand actual result data for each unit of each hierarchy of the product up to the previous day and the demand model (step S12). The demand prediction data generation unit 18 multiplies the likelihood by the prior distribution indicated by the prior distribution data, executes MCMC sampling, and calculates the demand prediction probability distribution, which is the posterior distribution for each unit of each hierarchy of the product (step S13). The demand prediction data generation unit 18 generates demand prediction data including the demand prediction probability distribution of the product at the minimum unit (step S14).

[0046] In the example of FIG. 3, the demand prediction data includes the demand prediction probability distributions of models M1 to M10. Model M1 includes products with model names T1 to T10, and model M10 includes products with model names T91 to T100. As shown in FIG. 3, assuming that T is the model name, M is the model name, and S[T,t] is the number of units sold by model name on the t-th day, S[T,t] follows a Poisson distribution with mean μ[T]. μ[T] follows a normal distribution with mean ν[M] and variance σ[M]2. ν[M] follows a normal distribution with mean ξ and variance δ2.

[0047] Returning to FIG. 14, the demand prediction data output unit 19 outputs the demand prediction data generated by the demand prediction data generation unit 18 to the product inventory calculation device 3 (step S15) and ends the process.

[0048] The supply LT prediction process shown in FIG. 15 starts, for example, when an instruction to calculate the inventory level is input to the inventory prediction system 100. When an instruction to calculate the inventory level is input to the inventory prediction system 100, the supply LT prediction data generation unit 23 of the supply LT prediction device 2 calculates the probability distribution of the supply LT of the minimum unit of the product based on the actual value of the supply LT of the minimum unit of the product included in the product supply performance data stored in the product supply performance data storage unit 21 (step S21), and generates product supply LT prediction data including the probability distribution of the supply LT of the minimum unit of the product (step S22). The supply LT prediction data output unit 24 outputs the product supply LT prediction data including the probability distribution of the supply LT of the minimum unit of the product to the product inventory calculation device 3 (step S23). The supply LT prediction data generation unit 23 calculates the probability distribution of the supply LT of the component based on the actual value of the supply LT of the component included in the component supply performance data (step S24), and generates component supply LT prediction data including the probability distribution of the supply LT of the component (step S25). The supply LT prediction data output unit 24 outputs the component supply LT prediction data including the probability distribution of the supply LT of the component to the component inventory calculation device 4 (step S26) and ends the process.

[0049] Note that the supply LT prediction data output unit 24 may be configured to generate, for example, product supply LT prediction data for viewing from the product supply LT prediction data, generate component supply LT prediction data for viewing from the component supply LT prediction data, display it on the screen, or transmit it to the terminal used by the user. In this case, in step S23, the supply LT prediction data output unit 24 performs MCMC sampling on the probability distribution indicated by the product supply LT prediction data, generates the product supply LT prediction data for viewing, and outputs it. In step S26, the supply LT prediction data output unit 24 performs MCMC sampling on the probability distribution indicated by the component supply LT prediction data, generates the component supply LT prediction data for viewing, and outputs it.

[0050] The product inventory calculation process shown in FIG. 16 starts, for example, when the product inventory calculation device 3 receives demand prediction data from the demand prediction device 1 and product supply LT prediction data from the supply LT prediction device 2. When receiving the demand prediction data and the product supply LT prediction data, the product inventory quantity prediction data generation unit 35 of the product inventory calculation device 3 performs MCMC sampling on the demand prediction probability distribution indicated by the demand prediction data and the probability distribution indicated by the product supply LT prediction data, and estimates the probability distribution of the inventory quantity of the product in the minimum unit by day based on the product inventory actual data and the instruction remaining data (step S31). The product inventory quantity prediction data generation unit 35 generates product inventory quantity prediction data including the probability distribution of the inventory quantity of the product in the minimum unit by day (step S32).

[0051] The product inventory quantity prediction data output unit 36 outputs the product inventory quantity prediction data including the probability distribution of the inventory quantity of the product in the minimum unit by day and the demand prediction data to the component inventory calculation device 4 (step S33), and ends the process.

[0052] Note that in step S33, the product inventory quantity prediction data output unit 36 may, for example, generate product inventory quantity prediction data for viewing from the product inventory quantity prediction data, display it on the screen, or transmit it to the terminal used by the user.

[0053] The component inventory calculation process shown in FIG. 17 starts, for example, when the component inventory calculation device 4 receives product inventory quantity prediction data from the product inventory calculation device 3 and component supply LT prediction data from the supply LT prediction device 2. When receiving the product inventory quantity prediction data and the component supply LT prediction data, the component inventory quantity prediction data generation unit 46 of the component inventory calculation device 4 performs MCMC sampling on the probability distribution indicated by the product inventory quantity prediction data and the probability distribution indicated by the component supply LT prediction data, and estimates the probability distribution of the daily inventory quantity of components based on the component configuration master, component inventory performance data, and incoming schedule data (step S41). The component inventory quantity prediction data generation unit 46 generates component inventory quantity prediction data including the probability distribution of the daily inventory quantity of components (step S42).

[0054] The component inventory quantity prediction data output unit 47 outputs the component inventory quantity prediction data including the probability distribution of the daily inventory quantity of components (step S43) and ends the process.

[0055] Note that in step S43, the component inventory quantity prediction data output unit 47 may, for example, generate browsing-use component inventory quantity prediction data from the component inventory quantity prediction data, display it on the screen, or transmit it to the terminal used by the user.

[0056] According to the inventory prediction system 100 according to the embodiment, since the hierarchical Bayesian method is used as the method for predicting the demand of products, even if the data serving as the samples of the lower hierarchy is insufficient, the demand can be predicted. Moreover, since the supply LT is handled as a probability distribution based on the actual results, it is possible to predict the inventory quantities of products and components closer to the actual situation, and it is also operable in actual operation even when the data serving as the samples is insufficient or when the market situation changes rapidly, and it is possible to realize an inventory replenishment operation that suits the actual situation.

[0057] In Bayesian statistics, the parameter is assumed to vary, and the statistic is estimated with a probability distribution. In Bayesian statistics, the statistic is output as a probability distribution (posterior distribution), and the posterior distribution can be updated based on newly acquired data. Compared with conventional statistics, Bayesian statistics has the advantage that it can stochastically estimate the statistic even when there is insufficient performance data. Due to these advantages, it can be applied to business fields with rapid changes. Also, when there are instabilities in the data such as shortages, missing values, and outliers, the prediction result with a wide tail of the probability distribution is output, so that the user can visually grasp the instability of the inventory quantity prediction.

[0058] In the above embodiment, the inventory quantity was calculated on a daily basis, but the calculation unit of the inventory quantity is not limited to this, and units such as half-day, two days, one week, and one month may also be used.

[0059] In the above embodiment, it was described that products are produced and parts are ordered, but this is not limiting, and parts may be produced.

[0060] In the above embodiment, the inventory prediction system 100 calculates the inventory quantities of products and parts, but this is not limiting, and it may be configured to calculate only the inventory quantity of products. In this case, the inventory prediction system 100 may not include the component inventory calculation device 4. Also, the object for which the inventory quantity is calculated may be an article for which inventory prediction is performed, as long as it has a hierarchy for classifying the articles.

[0061] The hardware configurations of the demand prediction device 1, the supply LT prediction device 2, the product inventory calculation device 3, and the component inventory calculation device 4 will be described with reference to FIG. 18. As shown in FIG. 18, the demand prediction device 1, the supply LT prediction device 2, the product inventory calculation device 3, and the component inventory calculation device 4 include a temporary storage unit 101, a storage unit 102, a calculation unit 103, an input unit 104, a transmission / reception unit 105, and a display unit 106. The temporary storage unit 101, the storage unit 102, the input unit 104, the transmission / reception unit 105, and the display unit 106 are all connected to the calculation unit 103 via a BUS.

[0062] The calculation unit 103 is, for example, a CPU (Central Processing Unit). The calculation unit 103 executes the processes of the prior distribution generation unit 14 and the demand prediction data generation unit 18 of the demand prediction device 1, the supply LT prediction data generation unit 23 of the supply LT prediction device 2, the product inventory quantity prediction data generation unit 35 and the product inventory alert unit 38 of the product inventory calculation device 3, and the component inventory quantity prediction data generation unit 46 and the component inventory alert unit 49 of the component inventory calculation device 4 according to the control program stored in the storage unit 102. Also, in a configuration where the demand prediction data output unit 19 of the demand prediction device 1, the supply LT prediction data output unit 24 of the supply LT prediction device 2, the product inventory quantity prediction data output unit 36 of the product inventory calculation device 3, and the component inventory quantity prediction data output unit 47 of the component inventory calculation device 4 generate browsing data, the calculation unit 103 executes the processes of the demand prediction data output unit 19 of the demand prediction device 1, the supply LT prediction data output unit 24 of the supply LT prediction device 2, the product inventory quantity prediction data output unit 36 of the product inventory calculation device 3, and the component inventory quantity prediction data output unit 47 of the component inventory calculation device 4 according to the control program stored in the storage unit 102.

[0063] The temporary storage unit 101 is, for example, a RAM (Random - Access Memory). The temporary storage unit 101 loads the control program stored in the storage unit 102 and is used as the working area of the calculation unit 103.

[0064] The storage unit 102 is a non-volatile memory such as a flash memory, a hard disk, a DVD-RAM (Digital Versatile Disc - Random Access Memory), a DVD-RW (Digital Versatile Disc - ReWritable), etc. The storage unit 102 prestores a program for making the calculation unit 103 perform the processes of the demand forecasting device 1, the supply LT forecasting device 2, the product inventory calculation device 3, and the part inventory calculation device 4, and also supplies data stored by this program to the calculation unit 103 according to an instruction from the calculation unit 103, and stores the data supplied from the calculation unit 103. The shipment plan data memory unit 11, error data memory unit 12, statistical model memory unit 13, demand model memory unit 15, demand actual data memory unit 16, and product group master memory unit 17 of the demand forecasting device 1, the product supply actual data memory unit 21 and the parts supply actual data memory unit 22 of the supply LT prediction device 2, the product inventory actual data memory unit 33, remaining order data memory unit 34, and product inventory alert condition memory unit 37 of the product inventory calculation device 3, and the parts configuration master memory unit 43, parts inventory actual data memory unit 44, inventory schedule data memory unit 45, and parts inventory alert condition memory unit 48 of the parts inventory calculation device 4 are configured in the memory unit 102.

[0065] In addition, the shipment plan data memory unit 11, error data memory unit 12, statistical model memory unit 13, demand model memory unit 15, demand actual data memory unit 16, and product group master memory unit 17 of the demand forecasting device 1, the product supply actual data memory unit 21, parts supply actual data memory unit 22, and supply LT alert condition memory unit 25 of the supply LT prediction device 2, the product inventory actual data memory unit 33, remaining order data memory unit 34, and product inventory alert condition memory unit 37 of the product inventory calculation device 3, and the parts configuration master memory unit 43, parts inventory actual data memory unit 44, inventory schedule data memory unit 45, and parts inventory alert condition memory unit 48 of the parts inventory calculation device 4 may be provided by an external device or system.

[0066] The input unit 104 is an input device such as a keyboard, a pointing device, or a voice input device, and an interface device that connects the input device to the BUS. Information input by the user is supplied to the calculation unit 103 via the input unit 104.

[0067] The transmission / reception unit 105 is a network termination device or a wireless communication device that connects to the network, and a serial interface or a LAN (Local Area Network) interface that connects to them. The transmission / reception unit 105 functions as the demand prediction data output unit 19 of the demand prediction device 1, the supply LT prediction data output unit 24 of the supply LT prediction device 2, the demand prediction data acquisition unit 31 of the product inventory calculation device 3, the product supply LT prediction data acquisition unit 32, and the product inventory quantity prediction data output unit 36, as well as the product inventory quantity prediction data acquisition unit 41, the component supply LT prediction data acquisition unit 42, and the component inventory quantity prediction data output unit 47 of the component inventory calculation device 4. Further, in a configuration where the supply LT prediction device 2, the product inventory calculation device 3, and the component inventory calculation device 4 output alerts to the terminal used by the user, the transmission / reception unit 105 functions as the supply LT alert unit 26 of the supply LT prediction device 2, the product inventory alert unit 38 of the product inventory calculation device 3, and the component inventory alert unit 49 of the component inventory calculation device 4.

[0068] The display unit 106 is a display device such as an LCD (Liquid Crystal Display) or an organic EL (electroluminescence) display. In a configuration where the demand prediction device 1, the supply LT prediction device 2, the product inventory calculation device 3, and the component inventory calculation device 4 each display browsing data on the screen, the display unit 106 functions as the demand prediction data output unit 19 of the demand prediction device 1, the supply LT prediction data output unit 24 of the supply LT prediction device 2, the product inventory quantity prediction data output unit 36 of the product inventory calculation device 3, and the component inventory quantity prediction data output unit 47 of the component inventory calculation device 4. Further, in a configuration where the supply LT prediction device 2, the product inventory calculation device 3, and the component inventory calculation device 4 output alerts on the screen display, the display unit 106 functions as the supply LT alert unit 26 of the supply LT prediction device 2, the product inventory alert unit 38 of the product inventory calculation device 3, and the component inventory alert unit 49 of the component inventory calculation device 4.

[0069] The processes of the shipment plan data storage unit 11, error data storage unit 12, statistical model storage unit 13, prior distribution generation unit 14, demand model storage unit 15, demand actual result data storage unit 16, product group master storage unit 17, demand prediction data generation unit 18, and demand prediction data output unit 19 of the demand prediction device 1 shown in FIG. 2 are executed by the control program using resources such as the temporary storage unit 101, calculation unit 103, storage unit 102, input unit 104, transmission / reception unit 105, and display unit 106 for processing.

[0070] The processes of the product supply actual result data storage unit 21, component supply actual result data storage unit 22, supply LT prediction data generation unit 23, supply LT prediction data output unit 24, supply LT alert condition storage unit 25, and supply LT alert unit 26 of the supply LT prediction device 2 shown in FIGS. 7 and 8 are executed by the control program using resources such as the temporary storage unit 101, calculation unit 103, storage unit 102, input unit 104, transmission / reception unit 105, and display unit 106 for processing.

[0071] The processes of the demand prediction data acquisition unit 31, product supply LT prediction data acquisition unit 32, product inventory actual result data storage unit 33, instruction remaining data storage unit 34, product inventory quantity prediction data generation unit 35, product inventory quantity prediction data output unit 36, product inventory alert condition storage unit 37, and product inventory alert unit 38 of the product inventory calculation device 3 shown in FIGS. 9 and 11 are executed by the control program using resources such as the temporary storage unit 101, calculation unit 103, storage unit 102, input unit 104, transmission / reception unit 105, and display unit 106 for processing.

[0072] The processing of the product inventory quantity prediction data acquisition unit 41, the component supply LT prediction data acquisition unit 42, the component composition master storage unit 43, the component inventory actual result data storage unit 44, the incoming scheduled data storage unit 45, the component inventory quantity prediction data generation unit 46, the component inventory quantity prediction data output unit 47, the product inventory alert condition storage unit 48, and the product inventory alert unit 49 of the component inventory calculation device 4 shown in FIGS. 12 and 13 is executed by the control program using resources such as the temporary storage unit 101, the calculation unit 103, the storage unit 102, the input unit 104, the transmission / reception unit 105, and the display unit 106.

[0073] In addition, the above hardware configuration and flowchart are examples, and arbitrary changes and modifications are possible.

[0074] The central part for performing the processing of the demand prediction device 1, the supply LT prediction device 2, the product inventory calculation device 3, and the component inventory calculation device 4, such as the calculation unit 103, the temporary storage unit 101, the storage unit 102, the input unit 104, the transmission / reception unit 105, and the display unit 106, can be realized using a normal computer system without relying on a dedicated system. For example, a computer program for executing the above operations is stored and distributed on a computer-readable recording medium such as a flexible disk, a CD-ROM (Compact Disc - Read Only Memory), or a DVD-ROM (Digital Versatile Disc - Read Only Memory), and the demand prediction device 1, the supply LT prediction device 2, the product inventory calculation device 3, and the component inventory calculation device 4 for executing the above processing may be configured by installing the computer program on a computer. Also, the computer program may be stored in a storage device of a server device on a communication network represented by the Internet, and the demand prediction device 1, the supply LT prediction device 2, the product inventory calculation device 3, and the component inventory calculation device 4 may be configured by downloading the computer program by a normal computer system.

[0075] In addition, when the functions of the demand prediction device 1, the supply LT prediction device 2, the product inventory calculation device 3, and the component inventory calculation device 4 are realized by sharing between the OS (Operating System) and the application program, or by cooperation between the OS and the application program, etc., only the application program part may be stored in a recording medium or a storage device.

[0076] Also, it is possible to superimpose a computer program on a carrier wave and provide it via a communication network. For example, the computer program may be posted on a bulletin board (BBS, Bulletin Board System) on the communication network, and the computer program may be provided via the communication network. Then, by starting this computer program and executing it in the same manner as other application programs under the control of the OS, the above-described processing may be executed.

[0077] Although the preferred embodiments have been described in detail above, the present invention is not limited to the above-described embodiments, and various modifications and substitutions can be made to the above-described embodiments without departing from the scope described in the claims.

[0078] Hereinafter, aspects of the present disclosure will be summarized as appendices.

[0079] (Appendix 1) An inventory prediction system that performs demand prediction using the hierarchical Bayesian method and calculates the inventory quantity of an article, a demand prediction device that generates a prior distribution for each unit of each layer obtained by classifying the article, calculates a demand prediction probability distribution that is a posterior distribution for each unit of each layer of the article based on the prior distribution, and generates demand prediction data including the demand prediction probability distribution of the article at the minimum unit; a supply LT prediction device that calculates a probability distribution of the supply lead time of the article at the minimum unit and generates supply LT prediction data including the probability distribution of the supply lead time of the article at the minimum unit; Based on the demand prediction data and the supply LT prediction data, an inventory calculation device estimates the probability distribution of the inventory level of the item at the minimum unit, and generates inventory prediction data including the probability distribution of the inventory level of the item at the minimum unit. An inventory prediction system comprising the same. (Appendix 2) The demand prediction device A prior distribution generation unit that generates a prior distribution for each unit of each hierarchy of the item using demand plan data indicating the demand plan for each unit of each hierarchy of the item, error data indicating the error between the demand plan and the demand actual performance for each unit of each hierarchy of the item, and a statistical model representing the relationship between the demand plan data and the error data; Based on the prior distribution for each unit of each hierarchy of the item, a demand model representing the demand trend for each unit of each hierarchy of the item, demand actual performance data indicating the demand actual performance for each unit of each hierarchy of the item, and an item group master indicating the hierarchy for classifying the item, a demand prediction probability distribution for each unit of each hierarchy of the item is calculated using the hierarchical Bayes method, and a demand prediction data generation unit that generates the demand prediction data. Comprising The inventory prediction system according to Appendix 1. (Appendix 3) The supply LT prediction device Based on supply actual performance data indicating the supply actual performance including the supply lead time of the item, a supply LT prediction data generation unit that calculates the probability distribution of the supply lead time of the item at the minimum unit and generates the supply LT prediction data. The inventory prediction system according to Appendix 1 or 2. (Appendix 4) The supply LT prediction device Further comprising a supply LT alert unit that outputs an alert when the probability distribution of the supply lead time of the item at the minimum unit satisfies a supply LT alert condition regarding the delivery date of the item, which is predetermined. The inventory prediction system according to Appendix 3. (Appendix 5) The inventory calculation device Based on the demand prediction data, the supply LT prediction data, the inventory performance data indicating the inventory performance of the item at the minimum unit, and the inventory replenishment schedule data indicating the quantity of the item at the minimum unit for which replenishment has been determined but not yet recorded in the inventory performance, a probability distribution of the inventory level of the item at the minimum unit is estimated, and an inventory level prediction data generation unit that generates the inventory level prediction data is provided. The inventory prediction system according to any one of Appendices 1 to 4. (Appendix 6) The inventory calculation device The inventory calculation device further includes an inventory alert unit that outputs an alert when the probability distribution of the inventory level of the item at the minimum unit satisfies the inventory alert condition regarding the ratio of supplying the item to the demand, which is predetermined. The inventory prediction system according to Appendix 5. (Appendix 7) The item is a product and a component, The inventory calculation device includes a product inventory calculation device and a component inventory calculation device, The product inventory calculation device Based on the demand prediction data including the demand prediction probability distribution of the product at the minimum unit, the supply LT prediction data including the probability distribution of the supply lead time of the product at the minimum unit, the inventory performance data indicating the inventory performance of the product at the minimum unit, and the inventory replenishment schedule data indicating the quantity of the product at the minimum unit for which production has been determined but not yet recorded in the inventory performance, a probability distribution of the inventory level of the product at the minimum unit is estimated, and the inventory level prediction data including the probability distribution of the inventory level of the product at the minimum unit is generated. The component inventory calculation device Based on the inventory level prediction data including the probability distribution of the inventory level of the product at the minimum unit, the inventory performance data indicating the inventory performance of the component, the inventory replenishment schedule data indicating the quantity of the component for which an order has been determined but not yet recorded in the inventory performance, the supply LT prediction data including the probability distribution of the supply lead time of the component, and the component configuration master indicating the component configuration of the product, a probability distribution of the inventory level of the component is estimated, and the inventory level prediction data including the probability distribution of the inventory level of the component is generated. The inventory prediction system according to Appendix 5 or 6. (Appendix 8) Demand prediction data including the demand prediction probability distribution of the item at the minimum unit, which is the posterior distribution calculated using the hierarchical Bayesian method based on the prior distribution for each unit of each hierarchy in which the item is classified, supply LT prediction data including the probability distribution of the supply lead time of the item at the minimum unit, inventory performance data indicating the inventory performance of the item at the minimum unit, and inventory replenishment schedule data indicating the quantity of the item at the minimum unit for which replenishment has been determined but not recorded in the inventory performance, based on these, estimate the probability distribution of the inventory level of the item, and include an inventory level prediction data generation unit that generates inventory level prediction data including the probability distribution of the inventory level of the item. Inventory calculation device. (Appendix 9) An inventory prediction method for performing demand prediction using the hierarchical Bayesian method and estimating the inventory level of an item, Performed by a demand prediction device, Generate a prior distribution for each unit of each hierarchy in which the item is classified, calculate a demand prediction probability distribution, which is the posterior distribution for each unit of each hierarchy of the item based on the prior distribution, and generate demand prediction data including the demand prediction probability distribution of the item at the minimum unit; Performed by a supply LT prediction device, Calculate the probability distribution of the supply lead time of the item at the minimum unit, and generate supply LT prediction data including the probability distribution of the supply lead time of the item at the minimum unit; Performed by an inventory calculation device, Based on the demand prediction data and the supply LT prediction data, estimate the probability distribution of the inventory level of the item at the minimum unit, and generate inventory level prediction data including the probability distribution of the inventory level of the item at the minimum unit; An inventory prediction method comprising the above steps. (Appendix 10) A computer, Demand prediction data including the probability distribution of demand prediction for the item at the minimum unit, which is the posterior distribution calculated using the hierarchical Bayesian method based on the prior distribution for each unit of each hierarchy for classifying the item, supply LT prediction data including the probability distribution of the supply lead time of the item at the minimum unit, inventory performance data indicating the inventory performance of the item at the minimum unit, and inventory replenishment schedule data indicating the quantity of the item at the minimum unit for which replenishment has been determined but not recorded in the inventory performance, estimate the probability distribution of the inventory level of the item, and generate inventory level prediction data including the probability distribution of the inventory level of the item. A program that functions as

[0080] Note that the present disclosure can be implemented in various embodiments and variations without departing from the broad spirit and scope of the present disclosure. Also, the above-described embodiments are for explaining this disclosure and do not limit the scope of the present disclosure. That is, the scope of the present disclosure is indicated by the claims rather than the embodiments. And various modifications made within the scope of the claims and within the scope of the meaning equivalent to the present disclosure are considered to be within the scope of this disclosure.

[0081] This application is based on Japanese Patent Application No. 2022-87913, filed on May 30, 2022. The entire specification, claims, and drawings of Japanese Patent Application No. 2022-87913 are incorporated herein by reference.

Explanation of Reference Numerals

[0082] 1 Demand prediction device, 2 Supply LT prediction device, 3 Product inventory calculation device, 4 Component inventory calculation device, 11 Shipping plan data storage unit, 12 Error data storage unit, 13 Statistical model storage unit, 14 Prior distribution generation unit, 15 Demand model storage unit, 16 Demand actual result data storage unit, 17 Product group master storage unit, 18 Demand prediction data generation unit, 19 Demand prediction data output unit, 21 Product supply actual result data storage unit, 22 Component supply actual result data storage unit, 23 Supply LT prediction data generation unit, 24 Supply LT prediction data output unit, 25 Supply LT alert condition storage unit, 26 Supply LT alert unit, 31 Demand prediction data acquisition unit, 32 Product supply LT prediction data acquisition unit, 33 Product inventory actual result data storage unit, 34 Instruction remaining data storage unit, 35 Product inventory quantity prediction data generation unit, 36 Product inventory quantity prediction data output unit, 37 Product inventory alert condition storage unit, 38 Product inventory alert unit, 41 Product inventory quantity prediction data acquisition unit, 42 Component supply LT prediction data acquisition unit, 43 Component composition master storage unit, 44 Component inventory actual result data storage unit, 45 Scheduled incoming data storage unit, 46 Component inventory quantity prediction data generation unit, 47 Component inventory quantity prediction data output unit, 48 Component inventory alert condition storage unit, 49 Component inventory alert unit, 100 Inventory prediction system, 101 Temporary storage unit, 102 Storage unit, 103 Calculation unit, 104 Input unit, 105 Transmission / reception unit, 106 Display unit, g1~g3 Graphs.

Claims

1. An inventory prediction system that performs demand prediction using the hierarchical Bayesian method and calculates the inventory level of an item, comprising: A demand prediction device that generates a prior distribution for each unit of each hierarchy obtained by classifying the item, calculates a demand prediction probability distribution that is a posterior distribution for each unit of each hierarchy of the item based on the prior distribution, and generates demand prediction data including the demand prediction probability distribution for the item at the minimum unit; A supply LT prediction device that calculates a probability distribution of the supply lead time for the item at the minimum unit and generates supply LT prediction data including the probability distribution of the supply lead time for the item at the minimum unit; An inventory calculation device that estimates a probability distribution of the inventory level of the item at the minimum unit based on the demand prediction data and the supply LT prediction data, and generates inventory level prediction data including the probability distribution of the inventory level of the item at the minimum unit; An inventory prediction system comprising the above components.

2. The demand prediction device includes: A prior distribution generation unit that generates a prior distribution for each unit of each hierarchy of the item using demand plan data indicating the demand plan for each unit of each hierarchy of the item, error data indicating the error between the demand plan and the demand actual result for each unit of each hierarchy of the item, and a statistical model representing the relationship between the demand plan data and the error data; A demand prediction data generation unit that calculates the demand prediction probability distribution for each unit of each hierarchy of the item using the hierarchical Bayesian method based on the prior distribution for each unit of each hierarchy of the item, a demand model representing the demand trend for each unit of each hierarchy of the item, demand actual result data indicating the demand actual result for each unit of each hierarchy of the item, and an item group master indicating the hierarchy for classifying the item, and generates the demand prediction data; The inventory prediction system according to Claim 1, comprising the above components.

3. The supply LT prediction device includes: A supply LT prediction data generation unit that calculates a probability distribution of the supply lead time for the item at the minimum unit based on supply actual result data indicating the supply actual result including the supply lead time of the item, and generates the supply LT prediction data. The inventory prediction system according to Claim 1 or 2, comprising the above components.

4. The supply LT prediction device further includes: A supply LT alert unit that outputs an alert when the probability distribution of the supply lead time for the item at the minimum unit satisfies a supply LT alert condition regarding the delivery date of the item, which is predetermined. The inventory prediction system according to Claim 3, comprising the above components.

5. The inventory calculation device includes: ​ Based on the demand prediction data, the supply LT prediction data, the inventory performance data indicating the inventory performance of the item at the minimum unit, and the inventory replenishment schedule data indicating the quantity of the item at the minimum unit for which replenishment has been determined but not recorded in the inventory performance, estimate the probability distribution of the inventory level of the item at the minimum unit, and include an inventory level prediction data generation unit that generates the inventory level prediction data. The inventory prediction system according to claim 1 or 2.

6. The inventory calculation device further includes an inventory alert unit that outputs an alert when the probability distribution of the inventory level of the item at the minimum unit satisfies a predetermined inventory alert condition regarding the ratio of supplying the item to the demand. The inventory prediction system according to claim 5.

7. The item is a product and a component, the inventory calculation device includes a product inventory calculation device and a component inventory calculation device, the product inventory calculation device Based on the demand prediction data including the probability distribution of the demand prediction of the product at the minimum unit, the supply LT prediction data including the probability distribution of the supply lead time of the product at the minimum unit, the inventory performance data indicating the inventory performance of the product at the minimum unit, and the inventory replenishment schedule data indicating the quantity of the product at the minimum unit for which production has been determined but not recorded in the inventory performance, estimate the probability distribution of the inventory level of the product at the minimum unit, and generate the inventory level prediction data including the probability distribution of the inventory level of the product at the minimum unit. The component inventory calculation device Based on the inventory level prediction data including the probability distribution of the inventory level of the product at the minimum unit, the inventory performance data indicating the inventory performance of the component, the inventory replenishment schedule data indicating the quantity of the component for which an order has been determined but not recorded in the inventory performance, the supply LT prediction data including the probability distribution of the supply lead time of the component, and the component configuration master indicating the component configuration of the product, estimate the probability distribution of the inventory level of the component, and generate the inventory level prediction data including the probability distribution of the inventory level of the component. The inventory prediction system according to claim 5.

8. Demand prediction data including the demand prediction probability distribution of the item at the minimum unit, which is the posterior distribution calculated using the hierarchical Bayesian method based on the prior distribution for each unit of each hierarchy into which the item is classified, supply LT prediction data including the probability distribution of the supply lead time of the item at the minimum unit, inventory performance data indicating the inventory performance of the item at the minimum unit, and inventory replenishment schedule data indicating the quantity of the item at the minimum unit for which replenishment has been determined but not recorded in the inventory performance, based on these, an inventory quantity prediction data generation unit that estimates the probability distribution of the inventory quantity of the item and generates inventory quantity prediction data including the probability distribution of the inventory quantity of the item. Inventory calculation device.

9. An inventory prediction method for performing demand prediction using the hierarchical Bayesian method and estimating the inventory quantity of an item, Performed by a demand prediction device, Generating a prior distribution for each unit of each hierarchy into which the item is classified, calculating a demand prediction probability distribution, which is the posterior distribution for each unit of each hierarchy of the item based on the prior distribution, and generating demand prediction data including the demand prediction probability distribution of the item at the minimum unit; Performed by a supply LT prediction device, Calculating the probability distribution of the supply lead time of the item at the minimum unit and generating supply LT prediction data including the probability distribution of the supply lead time of the item at the minimum unit; Performed by an inventory calculation device, Based on the demand prediction data and the supply LT prediction data, estimating the probability distribution of the inventory quantity of the item at the minimum unit and generating inventory quantity prediction data including the probability distribution of the inventory quantity of the item at the minimum unit; An inventory prediction method comprising the above steps.

10. A computer, Based on the prior distribution for each unit of each hierarchy into which the item is classified, demand prediction data including the demand prediction probability distribution of the item at the minimum unit, which is the posterior distribution calculated using the hierarchical Bayesian method, supply LT prediction data including the probability distribution of the supply lead time of the item at the minimum unit, inventory performance data indicating the inventory performance of the item at the minimum unit, and inventory replenishment schedule data indicating the quantity of the item at the minimum unit for which replenishment has been determined but not recorded in the inventory performance, an inventory quantity prediction data generation unit that estimates the probability distribution of the inventory quantity of the item and generates inventory quantity prediction data including the probability distribution of the inventory quantity of the item. A program for causing it to function as such.

Images