JP2019185377A - Information processor and information processing program - Google Patents

Abstract

To more simply estimate delivery.SOLUTION: A second provider device 20 includes an actual data acquisition part 212, a learning part 213, and an estimation part 214. The actual data acquisition part 212 acquires actual data which associates the quantity and the kind of an article with the quantity and the kind of a packaging material used for actually packaging an article corresponding to the quantity and the kind of the article. The learning part 213 performs machine learning on the basis of the actual data, and thereby constructs a learning model for estimating the packaging material. The estimation part 214 estimates the quantity and the kind of the packaging material used for actually packaging the article corresponding to the quantity and the kind of the article to be estimated on the basis of the learning model constructed by the learning part 213.SELECTED DRAWING: Figure 1

Description

  The present invention relates to an information processing apparatus and an information processing program.

  In recent years, there has been a demand for more efficient delivery of articles. For this purpose, for example, a technique is used in which a delivery is estimated using a computer, and an article is delivered based on the estimation result.

An example of a technique for performing such delivery estimation is disclosed in Patent Document 1. In the technique disclosed in Patent Document 1, a master including data such as the volume (length × width × height) of each article to be delivered is prepared in advance. When the input of the number of articles to be delivered is accepted, the volume of all the articles to be delivered is estimated by referring to this master.
By using this estimated volume, it is possible to further allocate vehicles for delivery, calculate delivery charges, and the like.

JP 2017-010340 A

  In order to perform estimation appropriately using a general technique such as the technique disclosed in Patent Document 1 described above, it is necessary to prepare a master including data such as the volume of each article to be delivered. is there.

  However, for example, a business operator who only undertakes delivery does not have data such as the volume of goods, and it is difficult to prepare such a master in advance. Even if you own data such as the volume of an article, in an industry where the life cycle of the product is short, it is necessary to update the master by measuring the volume of the product whenever a new product is sold. This was complicated and complicated.

  The present invention has been made in view of such a situation, and an object thereof is to provide an information processing apparatus and an information processing program for more simply performing estimation related to delivery.

In order to achieve the above object, an information processing apparatus of one embodiment of the present invention provides:
A record data acquisition means for acquiring record data associated with the quantity and type of the article and the quantity and type of the packing material used to actually pack the article corresponding to the quantity and type of the article;
A learning means for constructing a learning model for performing estimation related to the packing material by performing machine learning based on the result data;
Based on the learning model constructed by the learning means, estimating means for estimating the quantity and type of packing material used for actually packing the article corresponding to the quantity and type of the article to be estimated;
Is provided.

  According to the present invention, it is possible to estimate delivery more easily.

It is a block diagram showing an example of the whole composition of the information processing system concerning one embodiment of the present invention. It is a block diagram which shows an example of a structure of the 1st provider apparatus which concerns on one Embodiment of this invention. It is a block diagram which shows an example of a structure of the 2nd provider apparatus which concerns on one Embodiment of this invention. It is a block diagram which shows an example of a structure of the 3rd provider apparatus which concerns on one Embodiment of this invention. It is a 1st schematic diagram which shows the concept of learning and estimation in one Embodiment of this invention. It is a 2nd schematic diagram which shows the concept of learning and estimation in one Embodiment of this invention. It is a 3rd schematic diagram which shows the concept of learning and estimation in one Embodiment of this invention. It is a flowchart explaining the flow of the 1st learning algorithm in the learning of the 2nd provider apparatus which concerns on one Embodiment of this invention. It is a schematic diagram which shows an example of the decision tree produced | generated by the 2nd learning algorithm in the learning of the 2nd provider apparatus which concerns on one Embodiment of this invention. It is a sequence diagram which shows the flow at the time of the learning process which the information processing system which concerns on one Embodiment of this invention performs. It is a flowchart explaining the flow of the estimation process which the 2nd provider apparatus which concerns on one Embodiment of this invention performs. It is a schematic diagram which shows the concept of the fare calculation in the modification of one Embodiment of this invention. It is a 1st schematic diagram which shows the concept of learning and estimation in the modification of one Embodiment of this invention. It is a 2nd schematic diagram which shows the concept of learning and estimation in the modification of one Embodiment of this invention.

Hereinafter, an exemplary embodiment of the present invention will be described with reference to the accompanying drawings.
[System configuration]
FIG. 1 is a block diagram showing the overall configuration of the information processing system S according to the present embodiment. As illustrated in FIG. 1, the information processing system S includes a first operator device 10, a second operator device 20, a third operator device 30, and a network 40.

  The first operator device 10, the second operator device 20, and the third operator device 30 are connected to each other via a network 40 so that they can communicate with each other. Communication between the devices via the network 40 may be performed according to an arbitrary communication method, and the communication method is not particularly limited. The network 40 is realized by, for example, the Internet, a LAN (Local Area Network), a mobile phone network, or a combination of these. Moreover, the 1st provider apparatus 10, the 2nd provider apparatus 20, and the 3rd provider apparatus 30 are implement | achieved by the electronic devices which have information processing functions, such as a personal computer and a server apparatus.

  The first business operator device 10, the second business operator device 20, and the third business operator device 30 are devices used by different business operators. That is, the first business operator device 10 is a device used by the first business operator, the second business operator device 20 is a device used by the second business operator, and the third business operator device 30 is the third business device. It is a device used by business operators.

In the present embodiment, as an example for explanation, it is assumed that the third business operator is a business operator that operates a store that sells products (for example, a convenience store). In addition, it is assumed that the first business operator is a business operator such as a manufacturer that ships a product (for example, food) in response to an order from the third business operator. Further, it is assumed that the second business operator is a business operator who delivers the product shipped by the first business operator to a store operated by the third business operator in response to an order from the third business operator.
In this case, as shown in FIG. 1, an article that is a product moves in the order of the first business operator, the second business operator, and the third business operator.

In the information processing system S having such a configuration, learning processing and estimation processing are performed.
Specifically, in the learning process, the second company device 20 uses the quantity and type of the article and the quantity and type of the packing material used to actually pack the article corresponding to the quantity and type of the article. Acquires performance data associated with. Further, in the learning process, the second operator device 20 constructs a learning model for performing estimation regarding the packing material by performing machine learning based on the acquired result data.
Further, in the estimation process, the second business entity device 20 uses the constructed learning model, and the quantity of the packing material used for actually packing the article corresponding to the quantity and the type of the article to be estimated, and Estimate the type.

  By referring to the estimation result, the second business operator can predict how many products are packed in which packing material the product corresponding to the order data. That is, the second business operator can predict how another business operator (first business operator) will pack based on the ordering data.

Therefore, the second business operator can perform efficient delivery by preliminarily assigning vehicles necessary for delivery, determining delivery routes, and the like based on the prediction result. Further, in this case, the second business operator does not need to provide a master including data such as the volume of each item to be delivered or a master related to packing materials for packing the item.
That is, according to the information processing system S, it is possible to make an estimation regarding delivery more easily.

In the drawing, each device is illustrated one by one, but this is merely an example, and the information processing system S may include any number of these devices. Further, in response to this, there may be a plurality of businesses that use each device.
Next, the configuration of each of these devices will be described in detail.

[Configuration of first operator device]
Next, the structure of the 1st provider apparatus 10 is demonstrated with reference to the block diagram of FIG. As shown in FIG. 2, the first operator apparatus 10 includes a CPU (Central Processing Unit) 11, a ROM (Read Only Memory) 12, a RAM (Random Access Memory) 13, a communication unit 14, and a storage unit 15. And an input unit 16 and a display unit 17. These units are bus-connected by signal lines and transmit / receive signals to / from each other.

The CPU 11 executes various processes according to a program recorded in the ROM 12 or a program loaded from the storage unit 15 to the RAM 13.
The RAM 13 appropriately stores data necessary for the CPU 11 to execute various processes.

The communication unit 14 performs communication control for the CPU 11 to communicate with other devices included in the information processing system S.
The storage unit 15 includes a semiconductor memory such as a DRAM (Dynamic Random Access Memory) and stores various data.

The input unit 16 includes various buttons and a touch panel, or an external input device such as a mouse and a keyboard, and inputs various information according to a user's instruction operation.
The display unit 17 is composed of a liquid crystal display or the like, and displays an image corresponding to the image data output from the CPU 11.

When the first business operator apparatus 10 operates, as shown in FIG. 2, an inventory data management unit 111, an order data reception unit 112, and a shipping instruction unit 113 function in the CPU 11.
An inventory data storage unit 151 is set in one area of the storage unit 15.

  The inventory data storage unit 151 stores inventory data of articles manufactured by the first company (hereinafter referred to as “products” as appropriate). For example, data such as the number of products in stock, the number of products to be manufactured, the date and time of manufacture of products, and the expiration date (for example, expiration date) of products are stored as inventory data.

  The inventory data management unit 111 includes a function of managing inventory data stored in the inventory data storage unit 151. The inventory data management unit 111 is, for example, the operation of the first operator accepted by the input unit 16, the data received via the communication unit 14 from another device (not shown) used by the first operator, etc. Based on the above, management is performed by updating the inventory data stored in the inventory data storage unit 151 to the latest contents.

  The order data receiving unit 112 includes a function of receiving order data from the third business operator device 30 via the communication unit 14. The order data includes, for example, data such as the type of product delivered to each store operated by the third business operator, the number of products, and the delivery date and time. The order data receiving unit 112 also includes a function of outputting the received order data to the shipping instruction unit 113.

  The shipping instruction unit 113 includes a function of outputting shipping instruction information based on the latest inventory data stored in the inventory data storage unit 151 and the order data input from the order data receiving unit 112. The shipping instruction information includes, for example, information on which product in which warehouse is shipped to which location by what date and what time. Further, the shipping instruction unit 113 may also output information such as, for example, the shortage of stock for the product included in the ordering data. The output by the shipping instruction unit 113 is realized by, for example, displaying on the display unit 17 or printing on a paper medium from a printing apparatus (not shown).

  The first business operator performs shipment according to the order data by referring to the shipping instruction. In the present embodiment, as an example for explanation, it is assumed that the first business operator prepares a product corresponding to the ordering data from stock and delivers the prepared product to the delivery site of the second business operator. However, even if the first company prepares a product corresponding to the order data from the inventory and the second company goes to receive the product up to the warehouse of the first company, etc., this embodiment is used. It is possible to apply.

[Configuration of second operator device]
Next, the structure of the 2nd provider apparatus 20 is demonstrated with reference to the block diagram of FIG. As shown in FIG. 3, the second operator device 20 includes a CPU 21, a ROM 22, a RAM 23, a communication unit 24, a storage unit 25, an input unit 26, and a display unit 27. These units are bus-connected by signal lines and transmit / receive signals to / from each other.

  Here, the function as hardware of each part is equivalent to the function as hardware of each part of the same name, which is different from only the reference numeral, provided in the first business operator apparatus 10 described above. Therefore, the overlapping description is omitted.

When the second operator device 20 operates, as shown in FIG. 3, in the CPU 21, the order data reception unit 211, the result data acquisition unit 212, the learning unit 213, and the estimation unit 214 function.
In addition, a result data storage unit 251 and a learning model storage unit 252 are set in one area of the storage unit 25.

  In the performance data storage unit 251, information related to packing performed by the first company for delivering the product is stored as performance data. For example, information associating order data corresponding to one delivery with the quantity and type of packing material used for packing the product corresponding to the order data is stored as performance data.

  The learning model storage unit 252 stores a learning model constructed by machine learning performed by a learning unit 213 described later. Details of the machine learning performed by the learning unit 213 and the details of the learning model will be described later in the column “Learning and Estimation”.

  The order data receiving unit 211 includes a function of receiving order data from the third business operator device 30 via the communication unit 24. The order data receiving unit 211 outputs the received order data to the result data acquiring unit 212.

  The performance data acquisition unit 212 includes a function of acquiring performance data. In the present embodiment, when the second business operator receives the product corresponding to the ordering data, the second business operator measures the quantity and type of the packing material actually used for packing, and measures it by operating the input unit 26. The result is input to the record data acquisition unit 212. The performance data acquisition unit 212 is based on the operation of the second operator accepted by the input unit 26, the data received via the communication unit 24 from another device (not shown) used by the second operator, and the like. Thus, the quantity and type of the packing material used for packing the product corresponding to the order data are acquired.

  Then, the order data receiving unit 211 includes the order data input from the order data receiving unit 211 and the quantity and type of packing materials used for packing the product corresponding to the order data acquired as described above. Acquire results data by linking. Further, the order data receiving unit 211 stores the acquired result data in the result data storage unit 251.

  The learning unit 213 includes a function of constructing a learning model for performing estimation related to the packing material by performing machine learning based on the result data stored in the result data storage unit 251.

Based on the learning model constructed by the learning unit 213, the estimation unit 214 is a function for estimating the quantity and type of packaging materials used to actually pack the product corresponding to the quantity and type of the product that is the estimation target. including. In addition, the estimation unit 214 includes a function of outputting an estimation result. The output by the estimation unit 214 is realized by, for example, displaying on the display unit 27, printing on a paper medium from a printing apparatus (not shown), and the like.
Details of the machine learning performed by the learning unit 213, the details of the learning model, and the details of the estimation performed by the estimation unit 214 will be described later in the section “Learning and Estimation”.

[Configuration of Third Operator Device]
Next, the structure of the 3rd provider apparatus 30 is demonstrated with reference to the block diagram of FIG. As shown in FIG. 4, the third operator device 30 includes a CPU 31, a ROM 32, a RAM 33, a communication unit 34, a storage unit 35, an input unit 36, and a display unit 37. These units are bus-connected by signal lines and transmit / receive signals to / from each other.

  Here, the function of each part as hardware is equivalent to the function as hardware of each part of the same name, which is different from only the reference numeral, provided in the first provider apparatus 10 and the second provider apparatus 20 described above. Therefore, the overlapping description is omitted.

When the third business operator device 30 operates, as shown in FIG. 4, the order data generation unit 311 and the order data transmission unit 312 function in the CPU 11.
An order data storage unit 351 is set in one area of the storage unit 35.

  The order data storage unit 351 stores order data. For example, data such as the type of product delivered to each store operated by the third business operator, the number of products, and the delivery date are stored as ordering data.

  The order data generation unit 311 includes a function of generating order data stored in the order data storage unit 351. The order data generation unit 311 is, for example, the operation of the third business operator accepted by the input unit 36, the data received via the communication unit 34 from another device (not shown) used by the third business operator, etc. Order data is generated based on the above. Then, the order data generation unit 311 stores the generated order data in the order data storage unit 351.

  The order data transmission unit 312 transmits the order data stored in the order data storage unit 351 to the first business entity device 10 and the second business entity device 20 corresponding to the order data via the communication unit 34. including. The order data transmitted by the order data transmitting unit 312 is received by the order data receiving unit 112 and the order data receiving unit 211 described above.

[Learning and estimation]
Next, details of machine learning performed by the learning unit 213, details of the learning model, and details of estimation performed by the estimation unit 214 will be described with reference to FIGS.
5 to 7 are schematic diagrams showing the concepts of machine learning and estimation. The learning unit 213 constructs and updates a learning model by performing machine learning using the result data stored in the result data storage unit 251 as teacher data.

Specifically, as shown in FIG. 5, the learning unit 213 receives the order data D11 included in the record data, and the shipment data D12 included in the record data as an output. The relationship with the data D12 is learned. Then, the learning unit 213 uses the order data D21 included in the unknown data (data whose output has not yet been determined) as an input, and the relationship between the learned order data D11 and the shipment data D12 is the shipping data D22. Estimate based on Then, the estimation unit 214 outputs the estimated shipment data D22 as an estimation result.
The details of the machine learning performed by the learning unit 213, the details of the learning model, and the details of the estimation performed by the estimation unit 214 have been described above.

  Next, examples of specific data structures of the ordering data D11 and the shipping data D12 will be described with reference to FIG. In addition, with reference to FIG. 7, examples of specific data structures of the order data D21 and the shipping data D22 will be described. The specific examples illustrated in these drawings are merely examples for description, and the data structure of each data is not limited to the specific examples illustrated in these drawings.

  As shown in FIG. 6, the order data D <b> 11 that serves as an input for learning includes each item ordered at each day (here, March 1 to March 5 as an example). Each quantity and type is included. For example, in the case of the record of the product “Yakisoba Bread”, each of the product group 1 “cooking bread” that is a group to which the product belongs, and the product group 2 “vegetables” that is a group in which the product group 1 is further conceptualized, Includes daily order quantity. The order data D11 also includes the quantity of each product group ordered at each day at a certain store. For example, in the case of the record of the product group “cooking bread”, the total number of orders placed each day for each product belonging to this product group is included. For example, in the example of FIG. 6, the total number of orders for each day of the products “yakisoba bread”, “croquette bread”, “cutlet sandwich”, and “pirosiki bread” belonging to the product group “cooking bread” is included. .

  However, this is only an example, and information different from the information described above may be used as the product group. For example, a larger number of product groups may be set. Further, the number of product group hierarchies may be further increased, and product group 3 “food”, which is a group in which product group 2 is further conceptualized, may be added. In addition, the product groups may not have a hierarchical relationship and may have no hierarchical relationship. For example, the product group 1 is information indicating the product type such as “cooking bread”, the product group 2 is “product for men / women”, and the product group 1 such as “price sale product / special sale product”. It is good also as information which does not have a hierarchy. That is, the product group may be generated based on arbitrary property information indicating the product attribute.

  Further, as shown in FIG. 6, the shipping data D12 that is an output in learning is used by the first operator for actually packing each product ordered in each day at a certain store. Includes the quantity and type of packaging material that was present. For example, “shipping” and “crate” exist in the shipping data D12 as the types of packing materials. In addition, the shipment data D12 includes the quantity of “watch weight” and “crate” used for the actual packaging by the first operator at each day in a certain store. In addition, a magnitude | size and a shape may be contained as a kind of packing material. For example, even if it is a watch weight, it may be handled as different types of packing materials with a watch weight having a large size and a watch weight having a smaller size.

  The learning unit 213 learns the relationship between the ordering data D11 and the shipping data D12 on each day by machine learning. The relationship learning may be performed on the relationship between the quantity and type of each product included in the ordering data D11 and the quantity and type of packing materials included in the shipping data D12. It may be performed on the relationship between the quantity and type of each product group included and the quantity and type of packaging material included in the shipping data D12.

On the other hand, as shown in FIG. 7, the ordering data D21 is data including the quantity and type of products on the date to be estimated (in the figure, “March 6” is assumed as an example). For example, the order data D21 may be input by the second operator via the input unit 26, or the new order data received by the order data receiving unit 211 may be used as the order data D21.
Further, the order data D22 is data output as an estimation result by the estimation unit 214, and is a packing material used for actually packing a product corresponding to the quantity and type of the product targeted for estimation by the shipping data D21. This data includes the quantity and type of

  In this way, the estimation unit 214 estimates the shipping data D22 when the ordering data D21 is input based on the relationship between the ordering data D11 and the shipping data D12 learned by the learning unit 213.

In FIG. 6, the order data D11 ordered on a plurality of days (in this case, March 1 to March 5 as an example) at a certain store, or the certain store The case where learning is performed using the shipping data D12 shipped on each of a plurality of days (here, as an example, March 1 to March 5) has been described as an example. FIG. 7 illustrates an example in which the shipping data D22 is estimated using the order data D21 ordered on a certain day (here, March 6 as an example) at this certain store. did.
However, the present invention is not limited to this. For example, learning may be performed by using order data ordered at a plurality of stores on a certain day or shipment data shipped to a plurality of stores on this certain day. Good. In addition, the estimation may be performed using order data ordered on a certain day at a certain store other than the plurality of stores. Furthermore, for example, learning may be performed using order data ordered at each of a plurality of stores on each of a plurality of days, and shipment data shipped to each of a plurality of stores on each of the plurality of days. . These points will be described later as <Fourth Modification>.

  Next, as an example of a method for the learning unit 213 to perform learning, two algorithms, a first learning algorithm and a second learning algorithm, will be described below.

[First learning algorithm]
The first learning algorithm is a method for estimating the size of each product based on the result data. A flow when the learning unit 213 performs learning based on the first learning algorithm will be described with reference to the flowchart of FIG.

  In step S101, the learning unit 213 selects an index for evaluating the capacity of the packaging material. The index is selected from, for example, the volume of the container that is the packing material, the bottom area of the container, and the like. This selection may be performed based on a predetermined condition, or may be performed based on the operation of the second operator.

In step S102, the learning unit 213 obtains the average size of all the products by dividing the total total of the packing material capacity by the total amount of all the products based on the result data of the period for learning the results. The unit 213 assumes that all products have the same size, and sets the average size of all products obtained in step S102 as the size of each product.

In step S <b> 104, the learning unit 213 obtains the average number of orders received per day within the period for each product based on the result data of the period for learning the results.
In step S <b> 105, the learning unit 213 calculates, based on the result data of the period for which the result is learned, the quantity ordered for that day for each currently set size for each day within the period. Multiply and estimate the number of packaging materials required for the current setting.

  In step S <b> 106, the learning unit 213 selects a day on which the difference between the number of packing materials estimated in step S <b> 105 and the number of actual packing materials is the largest, based on the result data of the period for learning the results. Here, whether or not the divergence is the largest can be determined based on, for example, the divergence of the difference between the estimated number of packaging materials and the actual number of packaging materials or the divergence of the ratio.

  In step S107, the learning unit 213 determines that the number of actual orders received is the most dissimilar from the average number of orders obtained in step S104 on the day selected in step S106 based on the actual data of the period in which the actual results are learned. Ask for a product to do.

  In step S108, the learning unit 213 increases or decreases the setting size of the product selected in step S107 so that the estimated number of packing materials approaches the number of actual packing materials. In this case, you may make it provide a restriction | limiting in the range or increase / decrease rate of increase / decrease.

In step S <b> 109, the learning unit 213 determines whether or not the estimated number of packing materials matches the actual number of packing materials. If they do not match, it is determined No in step S109, and the process returns to step S107. Then, the process is resumed. In addition, since the increase / decrease is performed in step S108, in step S107 performed again, a product having the next largest difference from the product selected in the previous step S107 is selected.
On the other hand, if they match, it is determined Yes in step S109, and the process proceeds to step S110. Note that, even if they do not match, if the processing from step S107 to step S109 is repeated N times (N is an arbitrary natural number), it may be determined as Yes in step S109.

  In step S110, the learning unit 213 determines whether or not the termination condition of the first learning algorithm is satisfied. Here, the end condition can be arbitrarily determined. For example, there is no day when the estimated number of packing materials and the number of actual packing materials deviate, or the processing up to step S109 is repeated a predetermined number of times. This can be an end condition. If the end condition is satisfied, it is determined Yes in step S110, and the first learning algorithm ends. On the other hand, if the end condition is not satisfied, No is determined in step S110, and the process returns to step S106. Then, the process is resumed. In addition, since the increase / decrease is performed in step S108, in step S106 performed again, the day with the largest divergence after the day selected in the previous step S106 is selected.

  According to the first learning algorithm described above, an estimated value of the size of each product can be output, so that a human can check and correct it as necessary. That is, since the relationship between the actual data and the estimated value is a black box, there is no problem that it is difficult to understand how the computer has determined the estimated value.

[Second learning algorithm]
The second learning algorithm uses each day's ordering data (for example, the quantity of each product or the quantity of each product group) as input, and records as a combination of records that output the actual number of packaging materials corresponding to it as deep training data. A general supervised (or semi-supervised) machine learning algorithm such as learning is used for learning.

Although any machine learning algorithm can be used, in general, data in the world of distribution and logistics is not universal because of frequent product renovation and price / fluctuation fluctuations. Methods such as learning may not be appropriate.
Instead, in general, the relationship between input and output is relatively simple for data in the world of distribution and logistics. For example, it is easy to see that there is a relationship that, in general, the quantity of large-sized products increases, so the amount of cargo increases.

Considering such characteristics of teacher data, the ability to perform complex discrimination is inferior, but it is considered to perform learning using a decision tree such as a random forest that can obtain good learning results from a small number of teacher data. It is done.
As an example, a case where the learning unit 213 performs learning using a simple decision tree as the second learning algorithm will be described. In this example, when order data is input, a decision tree for determining whether or not one or more of a certain type of packing material is required is created.

  (1) In this case, the learning unit 213 divides the performance data into two groups, that is, when one or more target packing materials are required and in other cases.

  (2) Next, the learning unit 213 obtains an average value for each index on the input side for each of the two groups divided in (1) above, and the value is the most distant between the two groups. Find indicators. For example, in the order data D11 and the shipment data D12 described above with reference to FIG. 6, when it is determined whether or not there is one or more crate, a group of actual data with zero crate and one crate. In the above groups, the most disparate index is the number of “dairy products”, and the average number of “dairy products” in each group is 0 and 35.6.

  (3) The index found in (2) above is used as the branch index at this level, and the intermediate value between the two groups is used as the branch value. For example, in the case of the order data D11 and the shipment data D12 described above with reference to FIG. 6, the actual data is divided into two depending on whether the number of dairy products is 17.8 or more. However, this is only an example, and the branch index may be obtained by another method.

  (4) Each group divided in the above (3) confirms whether the target packing material is properly separated into only 0 data and 1 or more data. If it is not properly separated, the process returns to (1) again for the group and is divided in the same procedure.

  FIG. 9 shows a conceptual image of a completed decision tree that is completed by performing the above-described processing by the second learning algorithm. By using the decision tree thus completed, it is possible to determine the number of packing materials used for packing. FIG. 9 shows a conceptual image and is not actually created according to the above procedure for the ordering data D11 and the shipping data D12 described above with reference to FIG.

  Similarly, a decision tree that determines whether or not there are 1, 2, 3,..., M (where M is an arbitrary natural number) or more for each of other types of crate and number. Create This makes it possible to determine the number used for packing for each of a plurality of types of packing materials.

  According to the second learning algorithm, it is possible to perform estimation even when the container type and number cannot be obtained by the total size of the products to be simply loaded due to the problem of the cushioning material and the combination. That is, estimation can be performed even when the total amount of packages after packing is not determined simply by the sum of the sizes of the items in the order data.

[Learning process]
Next, the flow of learning processing executed by the information processing system S will be described with reference to the sequence diagram of FIG. Here, the learning process is a series of processes for realizing learning by the second business operator apparatus 20 through cooperation of the apparatuses included in the information processing system S.

In step S31, the order data generation unit 311 of the third business entity apparatus 30 generates order data.
In step S <b> 32, the order data transmission unit 312 of the third business entity device 30 transmits the result data generated by the order data generation unit 311 in step S <b> 31 to the first business operator device 10 and the second business operator device 20. To do.

In step S21, the order data receiving unit 211 of the second business operator apparatus 20 receives the order data transmitted in step S32.
In step S11, the order data reception unit 112 of the first business operator apparatus 10 receives the order data transmitted in step S32.

  In step S12, the shipping instruction unit 113 of the first business entity apparatus 10 generates shipping instruction information based on the order data received in step S11. Then, the shipping instruction unit 113 instructs the first company to ship by outputting the generated shipping instruction information. In response to this shipping instruction, the first operator performs shipping. Thereby, as shown by the broken line arrow in the figure, the product is moved to the delivery center of the second business operator. When the second business operator receives the product corresponding to the ordering data, the second business operator measures the quantity and type of the packing material actually used for packing, and the measurement result is obtained by operating the input unit 26. Input to the data acquisition unit 212.

  In step S22, the result data acquisition unit 212 of the second operator device 20 acquires the result data. In addition, the record data acquisition unit 212 stores the acquired record data in the record data storage unit 251 of the second operator device 20.

In step S23, the learning unit 213 of the second operator device 20 executes machine learning. For example, the learning unit 213 performs machine learning based on the first algorithm described above with reference to FIG. 8 or the second algorithm described above with reference to FIG. In addition, the learning unit 213 stores the learning model constructed (or updated) by machine learning in the learning model storage unit 252 of the second business operator apparatus 20. Thereby, the learning process ends.
Through the learning process described above, it is possible to create a learning model that is constructed based on performance data and updated as appropriate.

  In the learning process shown in the sequence diagram of FIG. 8, online learning is performed in which machine learning is immediately performed every time performance data that is teacher data is created. However, the present invention is not limited to this. For example, machine learning may be performed by batch learning. Batch learning is a learning method in which a plurality of performance data corresponding to repetition is collected while performance data is repeatedly created, and machine learning is performed using all the collected performance data.

[Estimation processing]
Next, the flow of estimation processing executed by the estimation unit 214 of the second operator apparatus 20 will be described with reference to the flowchart of FIG. Here, the estimation process is based on the learning model constructed (or updated) in the learning process, and the quantity of the packing material used to actually pack the article corresponding to the quantity and type of the article to be estimated, and It is a series of processes for estimating the type.

In step S41, the estimation unit 214 receives estimation target data.
In step S42, the estimation unit 214 executes estimation. Specifically, the estimation unit 214 uses the estimation target data received in step S41 as an input of the learning model stored in the learning model storage unit 252. And the estimation part 214 acquires the output of the learning model according to this input as an estimation result.

In step S43, the estimation unit 214 outputs the estimation result acquired in step S42.
By referring to the estimation result, the second business operator can predict how many products are packed in which packing material the product corresponding to the order data. That is, the second business operator can predict how another business operator (first business operator) will pack based on the ordering data.

  Such an estimation process can be performed when the second operator apparatus 20 acquires estimation target data (for example, current order data). For example, in the example of FIG. 10, estimation processing can be performed when order data is received in step S <b> 21, and can be performed before actually receiving shipped goods.

Therefore, the second business operator can perform efficient delivery by preliminarily assigning vehicles necessary for delivery, determining delivery routes, and the like based on the prediction result. Further, in this case, the second business operator does not need to provide a master including data such as the volume of each item to be delivered or a master related to packing materials for packing the item.
That is, according to the information processing system S, it is possible to make an estimation regarding delivery more easily.

  The present invention is not limited to the above-described embodiment, and modifications, improvements, and the like within the scope that can achieve the object of the present invention are included in the present invention.

For example, the series of processes described above can be executed by hardware or can be executed by software. In addition, each of the functional blocks described above may be configured by hardware alone, may be configured by software alone, or may be configured by a combination thereof.
In other words, the functional configuration illustrated in FIGS. 2, 3, and 4 is merely an example, and is not particularly limited. That is, it is sufficient if the information processing system S has a function capable of executing the above-described series of processes as a whole, and what functional block is used to realize this function is particularly shown in FIGS. It is not limited to the example illustrated in FIG.

  For example, the functional configuration included in the present embodiment can be realized by a processor that executes arithmetic processing, and various processors such as a single processor, a multiprocessor, and a multicore processor can be used in the present embodiment. In addition to what is constituted by a single processing apparatus, a combination of these various processing apparatuses and a processing circuit such as an ASIC (Application Specific Integrated Circuit) or an FPGA (Field-Programmable Gate Array) is included.

  For example, since the second provider device 20 has a large amount of computation associated with machine learning, for example, a technology called GPU (Graphics Processing Units) and GPGPU (General-Purpose computing on Graphics Processing Units) is installed in a personal computer. Therefore, if the GPU is used for arithmetic processing accompanying machine learning, high-speed processing can be performed. Furthermore, in order to perform higher-speed processing, a computer cluster is constructed by using a plurality of computers equipped with such GPUs, and parallel processing is performed by a plurality of computers included in the computer cluster. May be.

When a series of processing is executed by software, a program constituting the software is installed on a computer or the like from a network or a recording medium.
The computer may be a computer incorporated in dedicated hardware. The computer may be a computer capable of executing various functions by installing various programs, for example, a general-purpose personal computer.

  A recording medium including such a program may be provided to the user by being distributed separately from the apparatus main body in order to provide the program to the user, or provided to the user in a state of being pre-installed in the apparatus main body. Also good. A storage medium distributed separately from the apparatus main body is constituted by, for example, a magnetic disk (including a floppy disk), an optical disk, a magneto-optical disk, or the like. The optical disc is composed of, for example, a CD-ROM (Compact Disk-Read Only Memory), a DVD (Digital Versatile Disc), a Blu-ray (registered trademark) Disc (Blu-ray Disc), and the like. The magneto-optical disk is configured by an MD (Mini-Disk) or the like. The recording medium provided to the user in a state of being incorporated in the apparatus main body in advance is, for example, the ROM 12 in FIG. 2, the ROM 22 in FIG. 3, the ROM 32 in FIG. 4, or the storage unit in FIG. 15, a hard disk included in the storage unit 25 of FIG. 3 and the storage unit 35 of FIG.

In the present specification, the step of describing the program recorded on the recording medium is not limited to the processing performed in chronological order according to the order, but is not necessarily performed in chronological order, either in parallel or individually. The process to be executed is also included.
Further, in this specification, the term “system” means an overall apparatus including a plurality of apparatuses and a plurality of means.

As mentioned above, although several embodiment of this invention was described, these embodiment is only an illustration and does not limit the technical scope of this invention. The present invention can take other various embodiments, and various modifications such as omission and replacement can be made without departing from the gist of the present invention. These embodiments and modifications thereof are included in the scope and gist of the invention described in this specification and the like, and are included in the invention described in the claims and the equivalents thereof.
For example, the embodiment of the present invention may be modified as in the following modified example.

<First Modification>
In the above-described embodiment, the quantity and type of the packing material used for actually packing the article corresponding to the quantity and type of the article to be estimated are estimated. Not limited to this, other estimations may be performed. For example, estimation may be performed for a fare for delivering an article corresponding to the quantity and type of the article to be estimated.

  For this purpose, a combination of characteristics in delivery (for example, origin, landing, type of luggage, and vehicle type necessary for delivery, etc.) is handled as “product” in the above-described embodiment, and transportation means and freight are “packing materials”. And handle. In this way, it is possible to estimate the type and number of vehicles and the fare for a transportation request for a certain luggage set.

The fare estimation in this modification will be described in more detail with reference to FIG. The specific example illustrated in FIG. 12 is merely an example for description, and the data structure of each data is not limited to the specific example illustrated in FIG.
In this modification, as input of teacher data, instead of the order data D11 described above, delivery data D31 in which characteristics in delivery and a load amount are associated is used. A combination of characteristics in the delivery is considered as one “product”. Each characteristic is considered as one “product group”. Further, the quantity of cargo is considered as “product quantity”.
In addition, as output of the teacher data, fare data D32 indicating a fare that is a cost generated in the delivery is used instead of the shipment data D12. And this fare is considered “packaging material”.

  By thinking in this way, the learning unit 213 receives the relationship between the delivery data D31 and the fare data D32 when the delivery data D31 is input and the fare data D32 is output, as in the above-described embodiment. Can learn.

  Note that the delivery data D31 and the fare data D32 are actual data in actual delivery, respectively, and the actual data acquisition unit 212 is based on user input and the like, like the ordering data D11 and the shipping data D12. To be acquired.

Further, the estimation unit 214 has the relationship between the learned delivery data D31 and the fare data D32 as the fare data D42 when the delivery data D41 included in unknown data (data whose output has not yet been determined) is input. Estimate based on Then, the estimation unit 214 outputs the estimated fare data D42 as an estimation result.
According to this modification described above, it is possible to estimate a fare for delivering an article corresponding to the quantity and type of an article that is an estimation target.

  Further, the present modification may be further modified to estimate other information other than the fare. For example, the type and number of vehicles required for delivery may be estimated by considering one vehicle as one large packing material (that is, one large container).

<Second Modification>
When an unknown product (for example, a new product) is included in the ordering data D21 to be estimated, estimation may be performed using a product group. In this case, for example, the learning unit 213 learns the relationship between the shipment data D12 and the product group included in the order data D11, not the products included in the order data D11.

When the order data D21 to be estimated includes an unknown product, the estimation unit 214 performs estimation based on the product group of the unknown product.
In this way, even when an unknown product is included in the ordering data D21 to be estimated in the ordering data D21, estimation can be performed with higher accuracy.

<Third Modification>
In the above-described embodiment, in the learning process, the second business operator apparatus 20 constructs a learning model for performing estimation related to the packing material by performing machine learning based on the acquired result data.
Further, in the estimation process, the second company device 20 uses the quantity of the packing material used for actually packing the article corresponding to the quantity and the type of the article to be estimated based on the constructed learning model, and The type was estimated. The second operator used this estimation result.

Not only this but the function which performs a learning process and an estimation process may be implement | achieved by the 1st provider apparatus 10 (or the 3rd provider apparatus 30). And a 1st provider (or 3rd provider) may estimate with these apparatuses, and you may make it notify an estimation result with respect to a 2nd provider.
Even in such a configuration, the second operator can know how the other operator (first operator) packs the product corresponding to the ordering data. it can. That is, also by this modification, the same effect as the above-mentioned embodiment can be produced.

<Fourth Modification>
In the above-described embodiment, learning using the order data D11 and shipping data D12 shown in FIG. 6 and estimation using the order data D21 and shipping data D22 shown in FIG. 7 are taken as specific examples. I explained.
However, the present invention is not limited to this. For example, learning may be performed by using order data ordered at a plurality of stores on a certain day or shipment data shipped to a plurality of stores on this certain day. Good. In addition, the estimation may be performed using order data ordered on a certain day at a certain store other than the plurality of stores. In this case, an example of data used for learning is shown as ordering data D51 and shipping data D52 in FIG. In this case, an example of data used for estimation is shown as ordering data D61 and shipping data D62 in FIG.

As shown in FIG. 13, the order data D51 is the order data ordered at a plurality of stores (here, store A to store E as an example) on a certain day (here, March 5 as an example). It is. Further, as shown in FIG. 13, the shipment data D52 includes shipments shipped to a plurality of stores (here, store A to store E as an example) on a certain day (here, March 5 as an example). There is data.
In this modification, the learning unit 213 learns the relationship between the ordering data D51 and the shipping data D52 in each store by machine learning.

On the other hand, as shown in FIG. 14, the ordering data D61 includes the store to be estimated (store X as an example in the drawing) on the date to be estimated (in the figure, “March 6” as an example). ) Data including product quantity and type.
Further, the order data D62 is data output as an estimation result by the estimation unit 214, and is a packing material used for actually packing a product corresponding to the quantity and type of the product targeted for estimation by the shipment data D61. This data includes the quantity and type of

In this way, the estimation unit 214 estimates the shipment data D62 when the order data D61 is input based on the relationship between the order data D51 and the shipment data D52 learned by the learning unit 213.
As described above, as in this modification, the order data ordered at each of a plurality of stores on a certain day and the shipping data shipped to each of a plurality of stores on this certain day are used. Even if learning is performed, the same effects as those of the above-described embodiment can be obtained.

  In addition, this modification is further modified, for example, using order data ordered at each of a plurality of stores on each of a plurality of days, and shipment data shipped to each of a plurality of stores on each of the plurality of days. Even if learning is performed, the same effects as those of the above-described embodiment can be obtained.

10 1st company apparatus 20 2nd company apparatus 30 3rd company apparatus 11, 21, 31 CPU
12, 22, 32 ROM
13, 23, 33 RAM
14, 24, 34 Communication unit 15, 25, 35 Storage unit 16, 26, 36 Input unit 17, 27, 37 Display unit 40 Network 111 Inventory data management unit 112 Order data reception unit 113 Shipping instruction unit 151 Inventory data storage unit 211 Order data reception unit 212 Result data acquisition unit 213 Learning unit 214 Estimation unit 251 Result data storage unit 252 Learning model storage unit 311 Order data generation unit 312 Order data transmission unit 351 Order data storage unit S Information processing system

Claims (7)

A record data acquisition means for acquiring record data associated with the quantity and type of the article and the quantity and type of the packing material used to actually pack the article corresponding to the quantity and type of the article;
A learning means for constructing a learning model for performing estimation related to the packing material by performing machine learning based on the result data;
Based on the learning model constructed by the learning means, estimating means for estimating the quantity and type of packing material used for actually packing the article corresponding to the quantity and type of the article to be estimated;
An information processing apparatus comprising: The achievement data acquisition means acquires the achievement data when the first main body packs an article,
The learning means constructs the learning model by performing machine learning based on the performance data when the first main body packs an article,
The estimation means outputs a result of the estimation to a second subject;
The information processing apparatus according to claim 1. The learning means constructs a learning model for estimating the size of each type of article by performing machine learning based on the result data.
The information processing apparatus according to claim 1 or 2. The learning means performs machine learning based on the result data to determine a quantity of packing material necessary for actually packing an article corresponding to the quantity and type of the article to be estimated. Build a learning model for each type of packaging material,
The information processing apparatus according to claim 1 or 2. The learning means classifies each type of article based on a predetermined category, and builds a learning model for performing estimation on the packing material by performing machine learning based on the actual data after the classification,
The estimation means performs the estimation based on the classification corresponding to the article when the article to be estimated does not exist in the actual data.
The information processing apparatus according to any one of claims 1 to 4. The actual data acquisition means uses actual data associating actual data for fare estimation with information indicating characteristics in delivery of goods and fare generated for actually performing delivery corresponding to the information indicating the characteristics. Acquired,
The learning means constructs a learning model for performing estimation on a fare generated by delivery by performing machine learning based on the actual data for the fare estimation,
The estimation means estimates a fare generated for actually delivering an article to be estimated based on a learning model for making an estimate related to the fare constructed by the learning means.
The information processing apparatus according to any one of claims 1 to 5. An information processing program for causing a computer to function as an information processing apparatus,
A record data acquisition means for acquiring record data associated with the quantity and type of the article and the quantity and type of the packing material used to actually pack the article corresponding to the quantity and type of the article;
A learning means for constructing a learning model for performing estimation related to the packing material by performing machine learning based on the result data;
Based on the learning model constructed by the learning means, estimating means for estimating the quantity and type of packing material used for actually packing the article corresponding to the quantity and type of the article to be estimated;
An information processing program for causing the computer to function as an information processing apparatus comprising:

Images