WO2017163278A1

WO2017163278A1 - Product demand forecasting system, product demand forecasting method, and product demand forecasting program

Info

Publication number: WO2017163278A1
Application number: PCT/JP2016/001758
Authority: WO
Inventors: 圭介梅津
Original assignee: 日本電気株式会社
Priority date: 2016-03-25
Filing date: 2016-03-25
Publication date: 2017-09-28
Also published as: JP6451894B2; JPWO2017163278A1

Abstract

According to the present invention, a learner 81 learns a forecasting model on the basis of learning data including, for each of one or more products, an elapsed time since the sales launch of the product, words included in the name of the product, and a total quantity demanded for the product since the sales launch of the product. A forecaster 82 forecasts a quantity demanded for a target product, which is a product for which no learning data exists. Specifically, the forecaster 82 forecasts the quantity demanded for the target product for a target forecasting period, on the basis of the forecasting model and one or more words included in the name of the target product.

Description

Product demand forecasting system, product demand forecasting method, and product demand forecasting program

The present invention relates to a commodity demand prediction system, a commodity demand prediction method, and a commodity demand prediction program for predicting commodity demand.

A method of learning a prediction model based on past transaction results of products and performing future demand prediction based on the prediction model is widely known. For example, a predictive model is created based on learning data including data such as past sales performance, store opening hours, campaign information, weather information, etc., and the amount of product demand. A predicted value is obtained by substituting a value.

On the other hand, when there is no transaction record in the past like a new product, there is no learning data for the product, so a prediction model cannot be created by the method described above. In view of this, there has been proposed a method for performing demand prediction even when there is no information on the actual demand before release.

For example, Patent Document 1 describes a system for predicting demand for new products for which there is no past demand data. The system described in Patent Document 1 selects a product similar to a new product, calculates a base demand amount of the new product from the past demand amount of the similar product, and calculates a demand amount after the launch date of the new product. Ask.

Japanese Patent Laying-Open No. 2015-32034

However, in the system described in Patent Document 1, the determination as to whether a new product to be predicted and a product having past demand data are similar to each other depends on human subjectivity. Not self-evident. In other words, in the system described in Patent Document 1, an input of a product similar to a certain product is received from the user as a similar product, but the method of determining similarity is unknown. For this reason, there is a problem that the determination of whether or not they are similar depends on subjectivity such as past experience and intuition by skilled market staff.

Also, when there is no actual data of the same product, when learning a prediction model, it is conceivable to learn the prediction model by collecting historical data of past product groups similar to the new product. However, since it is not obvious which products are grouped as similar product groups, there is still a problem that the accuracy of the prediction model depends on the subjectivity of the experienced person.

Therefore, an object of the present invention is to provide a product demand prediction system, a product demand prediction method, and a product demand prediction program that can improve the demand prediction accuracy of a product for which learning data does not exist.

The product demand prediction system according to the present invention learns a prediction model based on learning data including an elapsed period from the start of sales of a product, a word included in the name of the product, and a demand quantity of the product after the start of sales. A learning device and a predictor for predicting the demand quantity of the target product, which is a product for which no learning data exists, based on the prediction model and the words included in the target product, It is characterized by forecasting demand quantity.

Another product demand prediction system according to the present invention is a product demand prediction system that predicts a demand quantity of a target product that is a product for which learning data including a past demand quantity does not exist. What is a target product sold in the past? Included in the forecast model that was learned based on the learning data including the elapsed period from the start of sales of different products, the words included in the name of the product, and the demand quantity of the product after the start of sales, and the target product And a predictor for predicting the demand quantity of the target product in the prediction target period based on the word.

The product demand prediction method according to the present invention learns a prediction model based on learning data including an elapsed period from the start of sales of a product, a word included in the name of the product, and a demand quantity of the product after the sales start. The demand quantity of the target product in the prediction target period is predicted based on the prediction model and words included in the target product that is a product for which no learning data exists.

Another product demand prediction method according to the present invention is a product demand prediction method for predicting a demand quantity of a target product that is a product for which learning data including a past demand quantity does not exist. What is a target product sold in the past? Included in the forecast model that was learned based on the learning data including the elapsed period from the start of sales of different products, the words included in the name of the product, and the demand quantity of the product after the start of sales, and the target product The demand quantity of the target product in the prediction target period is predicted on the basis of the word.

The product demand prediction program according to the present invention is based on learning data including an elapsed period from the start of sales of a product, a word included in the name of the product, and a demand quantity of the product after the start of sales. And a prediction process that predicts the demand quantity of the target product, which is a product for which no learning data exists, in the prediction process, based on the prediction model and the words included in the target product, The demand quantity of the target product in is predicted.

Another product demand prediction program according to the present invention is a product demand prediction program applied to a computer that predicts a demand quantity of a target product that is a product for which learning data including a past demand quantity does not exist. Forecasts learned based on learning data that includes the elapsed time from the start of sales of a product that is different from the target product sold on, the words included in the name of the product, and the demand quantity of that product since the start of sales Based on the model and the words included in the target product, a prediction process for predicting the demand quantity of the target product in the prediction target period is executed.

According to the present invention, it is possible to improve the demand prediction accuracy of a product for which no learning data exists.

It is a block diagram which shows the structural example of one Embodiment of the goods demand prediction system by this invention. It is explanatory drawing which shows the example of learning data. It is explanatory drawing which shows the other example of learning data. It is a flowchart which shows the operation example of 1st Embodiment. It is explanatory drawing which shows the example of a prediction model. It is explanatory drawing which shows the other example of learning data. It is a block diagram which shows the outline | summary of the goods demand prediction system by this invention. It is a block diagram which shows the other outline | summary of the goods demand prediction system by this invention.

For example, since a new product has no sales record in the past, a prediction model cannot be created from the sales result of the product. Accordingly, the inventor has focused on the word included in the name of the new product, not the past sales performance of the new product itself, and has come up with the idea of using the past sales performance of the product including the word. Specifically, in the present invention, a word included in the name of a product (more specifically, whether or not each word is included in each product) is an explanatory variable, and the demand quantity of the product (for example, the number of transactions) , Sales, orders, etc.). Hereinafter, embodiments of the present invention will be described with reference to the drawings.

Embodiment 1. FIG.
FIG. 1 is a block diagram showing a configuration example of a first embodiment of a commodity demand prediction system according to the present invention. The commodity demand prediction system 100 of this embodiment includes a storage unit 10, a learning device 20, a predictor 30, and an output unit 40.

The storage unit 10 stores learning data used by the learning device 20 described later to create a prediction model. The storage unit 10 is realized by, for example, a magnetic disk device. The learning device 20 and the storage unit 10 described later may be connected via a wired or wireless LAN (Local Area Network), or may be connected via the Internet.

The prediction model is information representing the correlation between explanatory variables and objective variables. A prediction model is a component for predicting the result of a prediction object, for example by calculating the target variable based on an explanatory variable. The prediction model may be described as “model”, “learning model”, “estimation model”, “prediction formula”, “estimation formula”, or the like.

The storage unit 10 stores learning data including an elapsed period from the start of sales of a product, a word included in the name of the product, and a demand quantity of the product after the sales start. For example, when the demand quantity is managed on a daily basis, the storage unit 10 includes the number of days elapsed from the sale start date of the product, the word included in the product name, and the demand quantity of the product after the sale start date. Store learning data. Hereinafter, the unit of the aggregation period of the demand quantity included in the learning data may be referred to as a unit period. For example, when there is daily learning data, the unit period is one day.

As a specific example of this embodiment, it is assumed that a certain factory predicts how much a new product should be manufactured as a demand quantity. If the quantity to be manufactured can be predicted, it is also possible to predict the raw materials necessary for manufacturing a new product at the factory. As the learning data, for example, sales data (for example, POS (Point データ of sale) data) of a new product acquired in the past at a store is used. For example, when “Bento” is a new product, it is preferable to use the sales data of the lunch box among the past sales data as the learning data. In addition, data for a certain period (for example, 1st to 7th days) from the release date is used to predict the sales trend of new products.

In addition, the demand means not only the demand of each store in a certain factory but also the demand of customers viewed from the store. For example, if the number of daily lunch boxes sold at a certain store is 10, from the store's viewpoint, the demand quantity of the lunch boxes for that day will be 10. In addition, if the number of lunches per day that are wholesaled to 100 stores in a certain area is 10 each, the demand quantity of the lunches for that day is 1000 from the viewpoint of the factory.

In this embodiment, since a word included in the name of a product is used as an explanatory variable, the storage unit 10 stores whether or not a word used as an explanatory variable is included in each product. In this embodiment, the memory | storage part 10 shall not memorize | store the learning data of the goods (henceforth target goods) made into the object of prediction. Examples of target products include new products and products that have not been handled so far among existing products.

FIG. 2 is an explanatory diagram illustrating an example of learning data stored in the storage unit 10. FIG. 2 exemplifies learning data including the number of days elapsed from the sales start date of a product sold in the past, the words included in the name of the product, and the demand quantity of the product on each elapsed date after the sales start date. ing. The transaction actual number (demand quantity) illustrated in FIG. 2 is, for example, a total value of the sales quantity and the order quantity of each store.

In the example shown in FIG. 2, variable 1 represents the number of days elapsed from the sales start date, and variables 2 to 5 indicate whether or not a predetermined word is included in the product name (if included, 1 When not included, 0) is represented. In the example shown in FIG. 2, the variable 2 is “whether or not the product name includes“ within the curtain ”(1 if included, 0 if not included, the same applies hereinafter)” and the variable 3 is “Japanese-style“ "Whether or not" "is included," variable 4 represents "whether or not" product name includes "season" "", and variable 5 represents "whether or not" product name includes "ginger" ". Further, the actual number of transactions represents the demand quantity.

For example, the first row of the table illustrated in FIG. 2 shows the demand for the product “Luxury Japanese Style Lunch Box” identified by the product ID = P101 on the first day from the sales start date at the store identified by the store ID = S101. It shows that the quantity was 1180 pieces. Further, since the word “gorgeous”, “Japanese style”, “curtain inside”, and “bento” are included in the “gorgeous Japanese style curtain lunch box”, 1 is set to the corresponding variable 2 and variable 3.

In addition, the memory | storage part 10 may memorize | store the word itself contained in the name of goods. Moreover, the memory | storage part 10 memorize | stores only the name of goods, and a word may be extracted from a goods name using methods, such as a morphological analysis, on the side to utilize. However, in order to speed up the learning process described later, the storage unit 10 preferably stores whether or not a word corresponding to the explanatory variable is included.

In the example shown in FIG. 2, the learning data includes the number of days elapsed from the sales start date of the product, the word included in the name of the product, and the demand quantity of the product on each elapsed date after the sales start date. However, the learning data may include other variables. Other variables include information indicating the characteristics of each elapsed date, information indicating the characteristics of each product, and the like. As information indicating the characteristics of the elapsed date, for example, the ratio of holidays to the number of elapsed days from the sales start date can be considered. As information indicating the characteristics of a product, for example, information indicating the amount of nutrients contained in the product can be considered.

FIG. 3 is an explanatory diagram showing another example of learning data. In FIG. 3, information indicating the number of days elapsed from the sales start date of a product sold in the past, the word included in the name of the product, the demand quantity of the product on each elapsed date after the sales start date, and the characteristics of each elapsed date And learning data including information indicating the characteristics of the product. Specifically, in the example shown in FIG. 3, variables 6 to 8 are added to the example shown in FIG.

Variable 6 represents, as information indicating the characteristics of each elapsed day, whether or not the day is a holiday (1 if it is a holiday, 0 if it is not a holiday). As information indicating the characteristics of each product, variable 7 represents the price of the product, and variable 8 represents the calorie of the product.

The information indicating the characteristics of each elapsed date and the information indicating the characteristics of each product may not necessarily be included in the learning data. However, it is more preferable to create a prediction model in consideration of these variables because the prediction accuracy becomes higher.

Also, the information indicating the characteristics of each elapsed date is not limited to information indicating whether the elapsed date is a holiday. Information indicating the characteristics of each elapsed day may include, for example, information such as the day of the week, whether it is a consecutive holiday, or the week of the month. Such information is also used as explanatory variables of the prediction model.

Moreover, in order to classify each product, the learning data may include information indicating the classification of the product. For example, when the product is food, the learning data may include information indicating whether “meat” or “fish” is included, or information indicating “rice” or “bread”. For example, since the product “full-boiled ginger-grilled lunch” is “boiled rice” including “meat”, the learning data may include variables indicating “meat” and “rice boiled rice”. Good. Further, the learning data may include information indicating the classification hierarchically. The learning data may include hierarchical classification information such as “meat> pork”.

The learning device 20 creates a prediction model based on the learning data described above. Specifically, the learning device 20 creates one prediction model with the demand quantity as an objective variable and variables (information) included in the learning data as explanatory variables. The method for creating the prediction model is arbitrary. The learning device 20 may create a prediction model using a generally known method. Since a method for creating a prediction model is widely known, detailed description thereof is omitted.

The predictor 30 predicts the demand quantity of the target product (that is, the product for which the learning data described above does not exist). Specifically, the predictor 30 predicts the demand quantity of the target product in the prediction target period based on the prediction model created by the learning device 20 and the words included in the target product.

As the demand quantity of the target product in the forecast target period, for example, the total demand quantity from the start of sales until the predetermined period elapses (from the first day of sales to the Mth day) or the day of the predetermined period after the start of sales ( The demand quantity on the (Nth day), the total demand quantity within a certain period (from the Nth day to the Mth day) after a predetermined period from the start of sales, and the like are conceivable.

Hereinafter, the prediction method when the explanatory variables illustrated in FIG. 3 are used will be described with reference to specific examples. Here, it is assumed that the demand quantity from the start of sales of the newly released product “seasonal vegetable sandwich ginger flavor” is predicted.

The prediction model using the variable of the learning data illustrated in FIG. 3 is expressed by, for example, Expression 1 illustrated below. Here, f is an arbitrary function representing a prediction formula.
Demand quantity D = f (variable 1, variable 2,..., Variable 7, variable 8) (Equation 1)

Hereinafter, the case where the demand quantity on the second day from the start of sales is predicted will be described. Here, it is assumed that the predicted date is a holiday, the price of the newly released product is 300 yen, and the calorie is 1000 kcal. In order to predict the number of transactions on the second day from the start of the new release, the variable 1 = 2. In addition, since the newly released product “seasonal vegetable sandwich ginger flavor” includes “season”, “vegetable”, “sand”, “ginger” and “flavor”, variable 2 = 0, variable 3 = 0, variable 4 = 1, variable 5 = 1. Note that the particle "no" may be included in the word candidates. Since the predicted date is a holiday, the variable 6 = 1. Moreover, since the price is 300 yen and the calorie is 1000 kcal, variable 7 = 300 and variable 8 = 1000. The predictor 30 predicts the demand quantity D on the second day by substituting these variables into Equation 1 shown above.

The same applies when calculating the total demand quantity within a certain period after a predetermined period has elapsed since the start of sales. For example, when calculating the total demand quantity from the second day to the fifth day of the sales start, the predictor 30 performs the third day, the fourth day, and the fifth day similarly to the case of forecasting the demand quantity on the second day. The daily demand quantity may be calculated, and the demand quantity for each day may be added to calculate the total demand quantity. The same applies when calculating the total demand quantity from the start of sales until a predetermined period elapses.

The output unit 40 outputs the prediction result by the predictor 30. The output unit 40 is realized by a display device, for example.

The learning device 20 and the predictor 30 are realized by a CPU of a computer that operates according to a program (commodity demand prediction program). For example, the program may be stored in the storage unit 10, and the CPU may read the program and operate as the learning device 20 and the predictor 30 according to the program. Further, the function of the product demand prediction system may be provided in the SaaS (Software as a Service) format.

Further, each of the learning device 20 and the predictor 30 may be realized by dedicated hardware. Moreover, a part or all of each component of each device may be realized by a general-purpose or dedicated circuit (circuitry), a processor, or a combination thereof. These may be configured by a single chip or may be configured by a plurality of chips connected via a bus. Part or all of each component of each device may be realized by a combination of the above-described circuit and the like and a program.

In addition, when some or all of the components of each device are realized by a plurality of information processing devices and circuits, the plurality of information processing devices and circuits may be centrally arranged or distributedly arranged. May be. For example, the information processing apparatus, the circuit, and the like may be realized as a form in which each is connected via a communication network, such as a client and server system and a cloud computing system.

Next, the operation of the product demand prediction system of this embodiment will be described. FIG. 4 is a flowchart illustrating an operation example of the commodity demand prediction system 100 according to the first embodiment.

The learning device 20 learns the prediction model based on the learning data (step S11). The predictor 30 predicts the demand quantity of the target product in the prediction target period based on the prediction model and the words included in the target product (step S12).

As described above, in the present embodiment, the learning device 20 is based on learning data including an elapsed period from the start of sales of a product, a word included in the name of the product, and a demand quantity of the product after the start of sales. Learn the prediction model. And the predictor 30 predicts the demand quantity of the goods (namely, object goods) with which learning data does not exist. Specifically, the predictor 30 predicts the demand quantity of the target product in the prediction target period based on the prediction model and the words included in the target product. With such a configuration, it is possible to improve the demand prediction accuracy of a product for which no learning data exists.

That is, in the product demand prediction system of the present embodiment, a prediction model is created based on objective information called product names. For this reason, prediction that does not depend on the subjectivity becomes possible, and the demand prediction accuracy can be improved. Moreover, the prediction which paid its attention to the name of goods like this embodiment corresponds also with the viewpoint of judging the goods visually by the consumer who saw the new goods.

Also, for example, from the viewpoint of a factory that manufactures new products, the amount of raw materials required before the release of new products can be ascertained in advance, so it is possible to suppress the risk of excessive or insufficient raw materials. In addition, from the viewpoint of a store that sells new products, it is possible to suppress risks such as stocking of new products and loss of opportunities.

Next, a modification of the first embodiment will be described. In 1st Embodiment, it illustrated about the case where a prediction model is represented by one prediction formula which is illustrated by Formula 1. On the other hand, a prediction model is not limited to the aspect represented by one prediction formula. The learning device 20 may create a prediction model in which a prediction expression is determined according to the value of a variable that identifies the target product. And the predictor 30 may specify a prediction formula according to the value of the variable which specifies object goods from the created prediction model, and may predict the demand quantity of object goods using the specified prediction formula.

FIG. 5 is an explanatory diagram illustrating an example of a prediction model in which a prediction formula is determined according to a value of a variable that identifies a target product. FIG. 5 illustrates a prediction model in which a selected prediction formula is represented by a tree structure. In the example illustrated in FIG. 5, the prediction formula 1 is used when the price of the target product is less than 300 yen, and the prediction formula 2 is used when the price of the target product is 300 yen or more.

Embodiment 2. FIG.
Next, a second embodiment of this embodiment will be described. The configuration of the commodity demand prediction system of this embodiment is the same as the configuration of the first embodiment. In the first embodiment, it is assumed that a certain factory predicts how much a new product should be manufactured as a demand quantity. This embodiment demonstrates the case where the demand quantity which considered the attribute of the store is estimated. The store attributes include, for example, information for identifying individual stores (for example, information on store types and chain stores), and regions to which the stores belong (for example, Kanto district, Kansai district, Tokyo, etc.) It is.

FIG. 6 is an explanatory diagram showing still another example of learning data stored in the storage unit 10. In addition to the variables 1 to 8 illustrated in FIG. 3, the learning data illustrated in FIG. 6 includes a variable 9 indicating a sales area as a store attribute. For example, in the Kanto region, the variable 9 = 1, and in the Kansai region, the variable 9 = 2.

The learning device 20 creates one prediction model including the demand quantity as an objective variable and a variable representing the store attribute as an explanatory variable. That is, in the present embodiment, the learning device 20 does not create a prediction model for each store or region, but creates a prediction model that includes the attributes of these stores as explanatory variables.

For example, in areas where there are few new store openings or stores, it is difficult to collect a large amount of learning data for new products. However, in this embodiment, since the learning device 20 can also use learning data of other regions that exist in large quantities by creating one prediction model, demand prediction can be performed even for regions with a small amount of learning data. Accuracy can be improved.

Note that the method of prediction by the predictor 30 is the same as in the first embodiment. For example, a prediction model using the learning data variables illustrated in FIG. 6 is represented by Expression 2 illustrated below. Here, g is an arbitrary function representing a prediction formula.
Demand quantity D = g (variable 1, variable 2,..., Variable 7, variable 8, variable 9) (Equation 2)

Hereinafter, the case where the demand quantity on the second day from the start of sales of a certain product in the Kansai area is predicted. As in the first embodiment, the values of variables 1 to 8 are set based on the elapsed period from the start of sale, the presence / absence of words, price, and calories. Furthermore, in order to predict the demand quantity in the Kansai area, the variable 9 = 2. The predictor 30 predicts the demand quantity D on the second day by substituting these variables into Equation 2 shown above.

As described above, in the present embodiment, the learning device 20 creates one prediction model including the demand quantity as the objective variable and the variable representing the store attribute as the explanatory variable. Therefore, in addition to the effects of the first embodiment, it is possible to perform demand prediction for each region. Further, by making one prediction model to be created, it is possible to improve the accuracy of demand prediction even in an area where learning data is small.

Next, the outline of the present invention will be described. FIG. 7 is a block diagram showing an outline of a commodity demand prediction system according to the present invention. The product demand prediction system 80 (for example, the product demand prediction system 100) according to the present invention includes an elapsed period from the start of sales of the product (for example, the number of days elapsed from the sales start date), a word included in the name of the product, and sales. A learning device 81 (for example, learning device 20) that learns a prediction model based on learning data including the demand amount of the product after the start, and a demand quantity (for example, ordering) of a target product that is a product for which no learning data exists A predictor 82 (for example, the predictor 30).

The predictor 82 predicts the demand quantity of the target product in the prediction target period based on the prediction model and the words included in the target product.

With such a configuration, it is possible to improve the demand prediction accuracy of products for which no learning data exists.

Also, the learning device 81 may create one prediction model that uses the demand quantity as an objective variable and includes a variable that represents a store attribute as an explanatory variable. According to such a configuration, it is possible to perform demand prediction from the viewpoint of the store. Further, by making one prediction model to be created, it is possible to improve the accuracy of demand prediction even in an area where learning data is small.

In addition, the learning device 81 obtains information indicating characteristics when a predetermined period has elapsed from the start of sales of the product (for example, characteristics of an elapsed date from the sales start date, whether it is a holiday, etc.), and characteristics of the product. The prediction model may be learned based on learning data including at least one piece of information (for example, price). According to such a configuration, it is possible to make a prediction in consideration of the sales date and the characteristics of the product.

Specifically, the predictor 82 determines the total demand quantity from the start of sales until the predetermined period elapses, the demand quantity for the unit period after the elapse of the predetermined period from the start of sales, or after the elapse of the predetermined period from the start of sales. The total demand quantity within a certain period of time may be predicted. Further, as the explanatory variable, for example, information including a word position in the product name may be used as the explanatory variable. For example, information such as whether a certain word is included at the head of the product name or whether a certain word is included at the end of the product name may be used as the explanatory variable.

Further, the learning device 81 may create a prediction model in which a prediction formula is determined according to the value of a variable that identifies the target product. And the predictor 82 may specify a prediction formula according to the value of the variable which specifies object goods from the created prediction model, and may predict the demand quantity of object goods using the specified prediction formula.

Specifically, a product for which no learning data exists may be a newly released product.

FIG. 8 is a block diagram showing another outline of the commodity demand prediction system according to the present invention. The product demand prediction system illustrated in FIG. 8 is a product demand prediction system 90 (for example, a product demand prediction system 100) that predicts the demand quantity of a target product that is a product for which learning data including past demand quantities does not exist. , Learned based on learning data including the elapsed time from the start of sales of a product different from the target product sold in the past, the word included in the name of the product, and the demand quantity of the product after the start of sales A predictor 91 (for example, predictor 30) that predicts the demand quantity of the target product in the prediction target period based on the prediction model and the words included in the target product.

Even with such a configuration, it is possible to improve the demand prediction accuracy of products for which no learning data exists.

DESCRIPTION OF SYMBOLS 10 Memory | storage part 20 Learning device 30 Predictor 40 Output part 100 Commodity demand prediction system

Claims

A learning device for learning a prediction model based on learning data including an elapsed period from the start of sales of a product, a word included in the name of the product, and a demand quantity of the product after the sales start,
A predictor for predicting the demand quantity of the target product, which is a product for which the learning data does not exist,
The predictor predicts a demand quantity of the target product in a prediction target period based on the prediction model and words included in the target product.
The commodity demand prediction system according to claim 1, wherein the learning device creates one prediction model including the demand quantity as an objective variable and a variable representing a store attribute as an explanatory variable.
The learning device learns a prediction model based on learning data including at least one of information indicating a characteristic when a predetermined period has elapsed from the start of sales of the product and information indicating a characteristic of the product. The commodity demand prediction system according to claim 2.
The predictor is the total demand quantity from the start of sales until the elapse of a predetermined period, the demand quantity for the unit period after the elapse of the predetermined period from the start of sales, or the total quantity within a certain period after elapse of the predetermined period from the start of sales The commodity demand prediction system according to any one of claims 1 to 3, wherein a demand quantity is predicted.
The learner creates a prediction model in which the prediction formula is determined according to the value of the variable that identifies the target product,
The predictor specifies a prediction formula according to a value of a variable that specifies the target product from the generated prediction model, and predicts a demand quantity of the target product using the specified prediction formula. The commodity demand prediction system according to any one of the above.
The product demand prediction system according to any one of claims 1 to 5, wherein the product for which learning data does not exist is a newly released product.
A product demand prediction system that predicts the demand quantity of a target product that is a product for which learning data including past demand quantities does not exist,
Learning based on learning data including an elapsed period from the start of sales of a product different from the target product sold in the past, a word included in the name of the product, and a demand quantity of the product after the sales start A product demand prediction system comprising: a predictor that predicts a demand quantity of the target product in a prediction target period based on the prediction model that has been made and a word included in the target product.
Learning a prediction model based on learning data including an elapsed period from the start of sales of the product, a word included in the name of the product, and a demand quantity of the product after the start of sales,
Based on the prediction model and a word included in a target product that is a product for which the learning data does not exist, a demand quantity of the target product in a prediction target period is predicted.
The commodity demand forecasting method according to claim 8, wherein a demand model is used as an objective variable, and one forecasting model including a variable representing a store attribute as an explanatory variable is created.
A product demand prediction method for predicting a demand quantity of a target product, which is a product for which learning data including a past demand quantity does not exist,
Learning based on learning data including an elapsed period from the start of sales of a product different from the target product sold in the past, a word included in the name of the product, and a demand quantity of the product after the sales start A demand forecast method for a product, characterized in that a demand quantity of the target product in a forecast target period is predicted based on the predicted model and a word included in the target product.
On the computer,
A learning process for learning a prediction model based on learning data including an elapsed period from the start of sales of a product, a word included in the name of the product, and a demand quantity of the product after the sales start, and
A prediction process for predicting a demand quantity of a target product that is a product for which the learning data does not exist;
A product demand prediction program for predicting a demand quantity of the target product in a prediction target period based on the prediction model and words included in the target product in the prediction process.
On the computer,
The merchandise demand forecasting program according to claim 11, wherein a learning model is used to create one forecasting model including demand quantity as an objective variable and a variable representing a store attribute as an explanatory variable.
A product demand prediction program applied to a computer that predicts the demand quantity of a target product that is a product for which learning data including past demand quantities does not exist,
In the computer,
Learning based on learning data including an elapsed period from the start of sales of a product different from the target product sold in the past, a word included in the name of the product, and a demand quantity of the product after the sales start The product demand prediction program for performing the prediction process which estimates the demand quantity of the said target goods in a prediction target period based on the made prediction model and the word contained in the said target goods.