JP2018032297A - Sales prediction apparatus, program and control method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a sales prediction apparatus, a program and a control method, capable of precisely predict sales of a merchandise even if the sales of the merchandise abruptly change.SOLUTION: A server device 1 calculates, in reprint determination processing, a prediction value of a cumulative sales quantity in prescribed days Δn from the present and a prediction range in view of an error of the prediction value, according to a transition of a past cumulative sales quantity. The server device 1 compares the calculated prediction value and prediction range of the cumulative sales quantity and a value obtained by multiplying a current cumulative purchase amount by a factor α to determine a recommendation degree of reprint. Here, if a sudden change of sales occurs, the server device 1 detects a point of time when the change occurs as a change point, and calculates a prediction value of a cumulative sales quantity according to the cumulative sales quantity having the change point as a starting point.SELECTED DRAWING: Figure 7

Description

  The present invention relates to a technique for predicting sales of goods.

  2. Description of the Related Art Conventionally, a technique for determining whether or not a duplicate of a book is necessary is known. For example, in Patent Document 1, based on the sales of books to the present time, the number of predicted sales for a predetermined period sold during a predetermined period from the present and the estimated number of final sales for which the target book is finally sold are calculated. There is disclosed a technique for outputting the difference between the number of stocks and the predicted number of books, whichever is smaller, as the recommended number of extra books to be printed. Patent Document 2 discloses an NM prediction method for predicting the cumulative number of books sold up to the Mth day by multiplying the cumulative number of books sold up to the Nth day after release by an NM coefficient.

JP 2008-282269 A JP 2007-058848 A

  Product sales may change drastically due to word-of-mouth or sales promotion effects via the Internet, etc.In such cases, the forecast of sales may be out of sales results, such as loss of sales opportunities for products, occurrence of excess inventory, etc. Cause. Therefore, the present invention provides a sales forecasting device, a program, and a control method capable of accurately forecasting the sales of a product even when the sales of the product rapidly change due to word-of-mouth or sales promotion effects, etc. The main issue is to do.

  In one aspect of the present invention, the sales prediction device includes a cumulative sales number calculation unit that calculates a first predicted value of the cumulative number of sales of the product, based on the sales of the product up to the present time, and a record of the cumulative number of sales. A change point detecting means for detecting a change point at a time when the tendency of the sales changes based on a comparison between the value and the first predicted value obtained by predicting the actual value of the cumulative number of sales, When the change point detection unit detects the change point, the cumulative sales number calculation unit calculates a second predicted value of the total sales number of the product based on the sales from the change point to the present.

  The sales forecasting device includes cumulative sales number calculation means and change point detection means. The cumulative sales number calculation means calculates a first predicted value of the cumulative sales number of the product based on the sales of the product up to now. The change point detection means detects a change point at which the sales trend has changed based on a comparison between the actual value of the cumulative number of sales and the first predicted value that predicted the actual value of the cumulative number of sales. Here, when the change point detection unit detects the change point, the cumulative sales number calculation unit calculates a second predicted value of the total sales number based on the sales from the change point to the present. By doing this, the sales forecasting device can accurately calculate the predicted value of the cumulative number of sales that matches the trend change even if the sales trend of the product changes due to word-of-mouth or sales promotion effects. Can do.

  In one aspect of the sales prediction apparatus, the change point detection means has an error between the actual value of the cumulative number of sales and the first predicted value that predicts the actual value of the cumulative number of sales is a predetermined value or more. When the error occurrence date continues for a predetermined number of days, the first day of the error occurrence date that continues for the predetermined number of days is detected as a change point. In this way, the sales forecasting device detects a change point only when the error occurrence date continues for a predetermined number of days, and misunderstands that the sales trend has changed when a temporary change in sales occurs. Setting the change point unnecessarily can be preferably suppressed.

  In another aspect of the sales forecasting device, the change point detecting means is configured to determine the cumulative number of sales based on the actual value of the cumulative number of sales and the first predicted value that predicts the actual value of the cumulative number of sales. It is determined whether the date corresponding to the actual value is the error occurrence date, information related to the determination is stored in the storage unit, and the current actual value of the cumulative number of sales and the current actual number of the cumulative number of sales are stored. When it is determined that the current date is the error occurrence date based on the first predicted value for which the value is predicted, the storage unit is referred to and the error occurrence date is continuous for the predetermined number of days. To determine. According to this aspect, the sales prediction device can detect the date as the changing point retroactively and accurately calculate the predicted value of the cumulative number of sales after the sales trend suddenly changes.

  In another aspect of the sales forecasting device, the product is a book, and the sales forecasting device is based on the cumulative total number of purchases to the book and the second predicted value. A duplication recommendation level determination unit that determines a recommendation level of a duplication version of a book, and an output unit that outputs information on the recommendation level are further provided. Here, “recommended degree of overprinting” indicates the degree to which overprinting is recommended, and is divided into at least two stages. The output means outputs information on the recommended degree of recommendation of the overprint. Here, “output recommendation level information” means a mode in which recommendation level information is displayed on a connected display or the like, and display information or the like necessary for displaying recommendation level information is displayed on another device. The mode which transmits to is also included. According to this aspect, even if the sales trend suddenly changes, the sales prediction apparatus appropriately determines the recommended degree of duplication of the book, and provides the user with a guideline for determining whether the duplication is necessary or not. Can be presented.

  In the aspect of the sales prediction apparatus, the duplication recommendation degree determination means includes a prediction range including a number obtained by multiplying the cumulative number of purchases by a variable coefficient, and the second prediction value or the second prediction value. By comparing at least one of them, the recommendation level of the duplicate of the book is determined. In this way, the sales prediction device compares the number obtained by multiplying the cumulative number of purchases by a variable coefficient with at least one of the second predicted value or the prediction range of the cumulative sales number, and determines the recommended degree of book duplication. decide. In this case, the sales prediction device can accurately determine the recommended degree of duplication of the book by adjusting a variable coefficient that is multiplied by the total number of purchases.

  In another aspect of the present invention, a program causes a computer to function as any one of the sales prediction apparatuses described above. By installing this program in a computer and causing it to function, the sales prediction apparatus according to the present invention can be configured.

  In still another aspect of the present invention, a sales method is a control method executed by a sales prediction apparatus, and calculates a first predicted value of a cumulative number of sales of a book based on sales up to the present time of a specified book. Based on a comparison between the number calculation step, the actual value of the cumulative number of sales, and the first predicted value that predicted the actual value of the cumulative number of sales, a change point that is the time when the trend of the sales has changed is detected And the accumulated sales number calculating step, when the changing point detecting step detects the changing point, the accumulated sales number of the product based on the sales from the changing point to the present. The second predicted value is calculated. By executing this control method, the sales forecasting device accurately calculates the predicted value of the cumulative number of sales based on the changed trend even if the sales trend of the product changes due to word-of-mouth or sales promotion effects. can do.

  The sales forecasting apparatus according to the present invention can accurately calculate the predicted value of the cumulative number of sales in accordance with the change in the trend even when the sales trend of the product changes due to word-of-mouth or sales promotion effects.

1 shows a configuration of a duplication plate determination system according to an embodiment. The block diagram of a server apparatus is shown. A block diagram of the publisher terminal is shown. (A) The transition of the actual value and the estimated value of the cumulative number of sales when a sudden change in sales occurs. (B) It is the graph which showed the actual value of the cumulative number of sales from the elapsed days N1 day regarded as a change point, and the predicted value of the cumulative number of sales calculated based on the transition of the cumulative number of sales after the change point. (C) It is a graph at the time of reflecting the graph of said (B) on the predicted value of the cumulative number of sales. It is a graph which shows transition of typical sales copies every day from the release date of a book. It is an example of the flowchart which shows the process sequence which displays a sales information screen. It is an example of the flowchart of a change point detection process. The example of a display of the sales information screen 15 days after a sale date is shown. The example of a display of the sales information screen 90 days after a sale date is shown.

  Hereinafter, preferred embodiments for carrying out the present invention will be described with reference to the drawings.

[Configuration of duplicate judgment system]
FIG. 1 shows a configuration of a duplicate plate determination system 100 according to the present embodiment. The overprint determination system 100 is a system for determining the recommended degree of overprint (that is, reprinting) by predicting the sales of books issued by each publisher. The overprint determination system 100 includes a server device 1 and a POS installed in a distributor or bookstore. It has a terminal 3 and a publisher terminal 4 installed at each publisher. The server device 1, the POS terminal 3, and the publisher terminal 4 communicate with each other via the network 2.

  The server device 1 receives the sales information “S1” indicating the number of sales for each book stored in the POS terminal 3 from each POS terminal 3, and recommends duplicate copies of each book based on the received sales information S1. Determine the degree. When the server apparatus 1 receives a request signal “S2” designating an arbitrary book from the publisher terminal 4, the server device 1 also displays a screen (“sales information screen”) showing the sales performance and sales forecast of the designated book. Display information “S3” for displaying “)” is transmitted to the publisher terminal 4. The server device 1 is an example of a “sales prediction device” in the present invention.

  The POS terminal 3 is installed at each agency or bookstore, and manages the number of purchases, sales, returns, and inventory in each book. Here, “Number of purchases” refers to the number of copies purchased at the bookstore, “Number of sales” refers to the number of copies of the book sold, and “Number of returns” is returned from the bookstore to the agency or publisher. The “city stock” indicates the number of books in the bookstore. Then, the POS terminal 3 transmits information such as the number of purchases for each book, the number of sales, the number of returned goods, and the number of stocks in the market to the server device 1 as sales information S1.

  The publisher terminal 4 is a terminal used by, for example, a person in charge of a publisher who makes a judgment on a duplicate of a book, and transmits a request signal S2 designating an arbitrary book to the server device 1. The publisher terminal 4 receives the display information S3 from the server device 1 as a response signal to the request signal S2, and displays a sales information screen. Further, the publisher terminal 4 transmits information such as the number of stocks held in the warehouse by the publisher (also referred to as “publisher stock number”) to the server device 1.

[Configuration of server device]
FIG. 2 is a block diagram of the server device 1. The server device 1 includes a storage unit 13, a communication unit 14 that performs data communication, and a control unit 15. Each of these elements is connected to each other via a bus line 10.

  The storage unit 13 is configured by a memory such as a hard disk or a flash memory. The storage unit 13 stores a program executed by the control unit 15, and stores information necessary for the control unit 15 to generate the display information S3. For example, the storage unit 13 stores a book information DB 131 in which information such as a book ID, a title, a genre, a price, and a release date is registered for each book. In addition, the storage unit 13 notifies the publisher terminal 4 of the total number of purchases, the total number of sales, the total number of returned goods, and the number of in-stock items calculated by the control unit 15 based on the sales information S1 received from the POS terminal 3. The sales information DB 132 is stored in which the number of publisher publishers and the like are stored in association with each book ID.

  The control unit 15 includes a CPU (Central Processing Unit), a ROM (Read Only Memory), a RAM (Random Access Memory), and the like (not shown), and performs various controls on each component in the server device 1.

  For example, the control unit 15 purchases each book based on the number of purchases, the number of sales, the number of returned goods, and the number of in-stock items included in the sales information S1 received by the communication unit 14 from each POS terminal 3. Calculate or recognize the cumulative number, the cumulative number of sales, the cumulative number of returned goods, and the number of stocks in the market, and update the sales information DB 132. Similarly, the control unit 15 updates the sales information DB 132 when receiving information such as the number of publishers in stock, the number of first edition copies, and the number of duplicate copies from the publisher terminal 4. In addition, when the communication unit 14 receives the request signal S2, the control unit 15 determines the recommendation level of the duplicate of the book (also referred to as “designated book”) specified by the request signal S2 (“overprint determination”). Also referred to as “processing”. The control unit 15 functions as “cumulative sales number calculating means”, “change point detecting means”, “overprint recommendation degree determining means”, “output means”, and a program executed by the computer in the present invention.

[Publisher terminal configuration]
FIG. 3 is a block diagram of the publisher terminal 4. The publisher terminal 4 includes a display unit 41 such as a display, an input unit 42 such as a mouse, a keyboard, and a touch panel, a storage unit 43, a communication unit 44 that performs data communication, and a control unit 45. Each of these elements is connected to each other via a bus line 40.

  The storage unit 43 is configured by a memory such as a hard disk or a flash memory. The storage unit 43 stores a program executed by the control unit 45.

  The control unit 45 includes a CPU, a ROM, a RAM, and the like (not shown), and performs various controls on each component in the publisher terminal 4. The control unit 45 performs a predetermined process by executing a program stored in the storage unit 43. For example, the control unit 45 receives an input for designating a designated book to be displayed on the sales information screen through the input unit 42, and sends a request signal S2 including the ID of the book designated by the input through the communication unit 44 to the server device 1. Send to. And the control part 45 displays the sales information screen based on display information S3 on the display part 41, when display information S3 is received from the server apparatus 1 as a response of request signal S2. In addition, the control unit 45 transmits information such as the number of publishers in stock, the number of first edition copies, and the number of duplicate copies to the server device 1 through the communication unit 44.

[Overlapping judgment processing]
Next, the duplicate determination process executed by the server device 1 will be described. Schematically, in the duplication determination process, the server apparatus 1 predicts based on the transition of the past cumulative number of sales, taking into account the predicted value of the cumulative number of sales after a predetermined number of days “Δn” from the present and the error of the predicted value. Calculate the range. Then, the server device 1 compares the calculated estimated value and estimated range of the cumulative sales number with a value obtained by multiplying the current cumulative total number of purchases by a predetermined coefficient “α”, thereby determining the recommended degree of duplication. decide. At this time, when the sales trend changes due to a sales promotion effect, word-of-mouth, or the like, the server device 1 detects the time when the change occurs (also referred to as “change point”), and after the change point. Calculate the predicted value of cumulative sales based on the cumulative sales. As a result, the server apparatus 1 accurately predicts the cumulative number of sales even when the sales trend changes suddenly, and presents the recommended degree of duplication with high reliability to the user, thereby suitably reducing sales opportunity loss and excess inventory. Let

(1) Method for Determining Elapsed Days Δn Here, first, a method for determining the elapsed days Δn will be described. The elapsed days Δn is set to, for example, a number of days that is equal to or longer than a period (that is, a lead time) from when an instruction for reprinting the target book is issued until it is actually reprinted and arranged in the bookstore. The server device 1 may receive the user input of the elapsed days Δn on the sales information screen and receive the input information from the publisher terminal 4 to determine the elapsed days Δn.

(2) Prediction of Cumulative Sales before Detection of Change Point Next, a method for calculating the predicted value of the cumulative sales at the time after the lapse of the number of elapsed days Δn from now (also referred to as “duplicate judgment reference date”) will be described. Here, calculation of the predicted value of the cumulative number of sales when the change point is not detected will be described, and calculation of the predicted value of the cumulative number of sales when the change point is detected will be described later. Note that the predicted value of the cumulative number of sales when the change point is not detected is an example of the “first predicted value” in the present invention, and the predicted value of the cumulative number of sales when the change point is detected is It is an example of the “second predicted value” in the invention.

  Before detecting the change point, the server device 1 calculates the predicted value of the cumulative number of sales on the duplication determination reference date based on the transition of the cumulative number of sales from the release date of the book stored in the sales information DB 132 to the present. calculate. For example, the server device 1 sets the number of days elapsed from the book release date to a variable “x” and the total number of sales to a variable “y”, and totals the sales totaled for each date from the book release date to the present. Using the number as a sample, an approximate expression representing the transition of the cumulative number of sales is calculated based on regression analysis, the least square method, or the like. Then, the server device 1 calculates a predicted value of the cumulative number of sales on the duplicate determination reference date based on the calculated approximate expression. In addition, when calculating the approximate expression, the server device 1 considers only the change in the cumulative number of sales within the predetermined number of days from the present, instead of considering all the changes in the cumulative number of sales from the book release date to the present. An approximate expression may be calculated.

  In another example, the server device 1 uses the sales data of other books of the same genre as the target book based on the NM prediction method described in Patent Document 2 and the like, so that the sales on the duplication determination reference date Calculate the cumulative number. In this case, the server device 1 uses the variable “x” as the cumulative number of sales from the release date to an arbitrary time “N” day, and the variable “y” as the cumulative sales from the release date to an arbitrary time “N + Δn”. And an approximate expression by the least square method of variables x and y for each genre of books released in the past is calculated. And the server apparatus 1 calculates the value of the variable y obtained by substituting the present cumulative sales number for x using the said approximate expression as a predicted value of the cumulative sales number on the duplication determination reference date.

(3) When the trend of sales suddenly changes due to the detection of the change point, the predicted sales promotion effect based on the change point, the word of mouth, etc., the above-described method for calculating the predicted value of the cumulative number of sales increases the prediction error. FIG. 4A shows the transition of the actual value and the predicted value of the cumulative number of sales when a sudden change in sales occurs. In FIG. 4A, the solid line graph shows the transition of the actual value of the cumulative number of sales, and the broken line graph shows the transition of the predicted value of the cumulative number of sales without considering the change point.

  In the example of FIG. 4A, the rate of increase in the cumulative number of sales has risen sharply due to sales promotion effects, word-of-mouth, etc., on the “N1” day since the release date. On the other hand, the calculation of the forecast value of the cumulative number of sales without considering the change point takes into account the actual value of the cumulative number of sales before the N1 day since the release date. The actual value of the cumulative number of sales has a large fluctuation with respect to the calculated predicted value of the cumulative cumulative number of sales, and the prediction accuracy of the cumulative number of sales is reduced.

  In consideration of the above, the server device 1 detects the point of time when the rate of increase in the cumulative number of sales is rapidly increasing (that is, the number of days N1 since the release date) as a change point, and sales after the detected change point. Calculate the predicted value of the cumulative number of sales based on the transition of the cumulative number.

  Here, a method for detecting the change point will be specifically described. First, the server device 1 stores the calculated predicted value in the storage unit 13 every time it calculates the predicted value of the cumulative number of sales after Δn days from the present in the state before the change point is detected. Thereafter, when Δn day has elapsed from the calculation of the predicted value and the cumulative sales number is predicted, the server apparatus 1 predicts the current cumulative sales number predicted “Δn days before the date (“ past The predicted cumulative sales number Sp ”is also compared with the current cumulative sales number (also referred to as“ current cumulative number Sc ”) recorded in the sales information DB 132. The server device 1 then refers to the date when the current cumulative sales number Sc exceeds the past predicted cumulative sales number Sp by a predetermined threshold (also referred to as “threshold value Sth”) or more as a change point candidate (also referred to as “change point candidate”). ) And stored in the storage unit 13. The threshold value Sth is set based on an experiment or the like as an error value that can be assumed when there is no change in the sales trend, for example.

  Then, the server device 1 similarly compares the current cumulative sales number Sc with the past predicted cumulative sales number Sp after the date regarded as the change point candidate, and continuously displays the current sales for a predetermined number of days or more continuously from the stored change point candidates. When the cumulative number Sc exceeds the past predicted cumulative sales number Sp by a threshold Sth or more, the change point candidate is regarded as a change point. The above-mentioned predetermined number of days is a threshold value that is set to prevent erroneous detection of a point of time when the sales temporarily fluctuate even though there is no change in the sales trend, and is based on, for example, experiments It is set in advance to an appropriate value. A specific processing procedure of the change point detection method will be described later with reference to the flowchart of FIG. The date when the current cumulative sales number Sc exceeds the past predicted cumulative sales number Sp by a threshold Sth or more is an example of the “error occurrence date” in the present invention.

  Next, a method for calculating the predicted value of the cumulative number of sales when a change point is detected will be described. When the server device 1 detects a change point, the server device 1 predicts the number of sales to be added to the cumulative number of sales at the change point from the actual sales value after the detected change point, thereby predicting the cumulative number of sales. Calculate the value.

FIG. 4B is a graph showing the actual value of the cumulative number of sales from the number of elapsed days N1 regarded as the change point and the predicted value of the cumulative number of sales calculated based on the transition of the cumulative number of sales after the change point. It is. In the example of FIG. 4B, the accumulated sales number offset value based on the actual value “S _N1 ” of the accumulated sales number on the elapsed _N1 day, that is, the accumulated sales starting from the change point. Numbers are shown. As shown in FIG. 4 (B), when detecting a change point, the server device 1 applies regression analysis, least squares, or the like using the actual value of the cumulative number of sales from the elapsed day N1 as a sample. Thus, a predicted value of an increase (offset value) from the actual value S _{N1 of the} cumulative number of sales at the change point is calculated. As described above, when the change point is detected, the server device 1 predicts the change in the cumulative number of sales based on the cumulative number of sales starting from the change point.

FIG. 4C is a graph when the cumulative number of sales calculated from the change point is reflected in the predicted value of the cumulative number of sales from the release date. In FIG. 4C, the predicted value (see the broken line) of the cumulative number of sales after the elapsed day N1 that is the changing point is the offset value of the predicted value of the cumulative number of sales shown in the broken line graph of FIG. the actual value S _N1 sales cumulative number of the change point is set to a value obtained by adding. As a result, in the example of FIG. 4C, the prediction accuracy of the predicted value of the cumulative number of sales after the change point is improved as compared with the example of FIG.

  As described above, in this embodiment, the server device 1 detects a change point, and sets the change point as a starting point when calculating the predicted value of the cumulative number of sales, thereby causing a sudden change in sales at the change point. The predicted value of the corresponding cumulative number of sales can be calculated accurately.

(4) Setting Forecast Range Next, a method for setting the forecast range for the cumulative number of sales will be described. Hereinafter, the upper limit value of the prediction range is also referred to as “prediction upper limit value”, and the lower limit value of the prediction range is also referred to as “prediction lower limit value”.

  The server device 1 sets the prediction range to a range in which the expected error range of the prediction value is set up and down with respect to the calculated prediction value. For example, the server device 1 calculates a mean square error (that is, standard deviation) between the approximate expression used to calculate the predicted value and the sample value used to calculate the approximate expression, and the larger the standard deviation is, the greater the standard deviation is. , Set the prediction range to be wide. In another example, the server device 1 sets the prediction upper limit value and the prediction lower limit value so that the difference between the prediction value and each of the prediction upper limit value and the prediction lower limit value becomes a predetermined ratio (for example, 20%) of the prediction value. The prediction range is determined by setting.

(5) the method for determining the recommended degrees will now be described a method of determining the recommendation degree. The server device 1 compares the predicted value and the predicted upper limit value of the cumulative number of sales on the duplication determination reference date with the number obtained by multiplying the current total number of purchases by a variable coefficient α (0 <α ≦ 1). This determines the degree of recommendation. A method for setting the coefficient α will be described later.

  In this embodiment, as an example, the server device 1 represents the above-mentioned recommendation level in three stages. On the sales information screen, the highest recommendation level is “◎”, the next highest recommendation level is “◯”, and the lowest recommendation level. The degree of recommendation (that is, the degree of recommendation when duplication is not recommended) is represented by “−”. Then, when the predicted value of the cumulative number of sales on the duplication determination reference date is larger than the cumulative number of purchases multiplied by the coefficient α, the server device 1 does not have to have an appropriate amount in the market before the duplication determination reference date. It is determined that it is necessary to increase the cumulative number of purchases, and the recommendation level of the duplicate is set to “◎”, which is the highest. Further, the server apparatus 1 recommends duplication when the predicted value is equal to or less than the cumulative number of purchases multiplied by the coefficient α and the predicted upper limit value is larger than the cumulative number of purchases multiplied by the coefficient α. Set the degree to “O”. In other words, in this case, the server device 1 determines that it is necessary to increase the cumulative number of purchases in order to maintain an appropriate market inventory rate when the sales increase more than predicted, and the recommended degree of duplication is determined. Set to “O”. On the other hand, when the predicted upper limit value is equal to or less than the cumulative number of purchases multiplied by the coefficient α, the server device 1 determines that an appropriate commercial stock quantity is secured at the time of the duplication determination reference date without performing duplication. , The recommendation level of the overprint is set to “−” (that is, no overprint is required).

(6) How to set coefficients alpha Next, the procedure for setting the coefficient alpha. The coefficient α is a variable value that changes according to a change in the current date. Hereinafter, an example of setting the coefficient α will be described.

  For example, the server device 1 changes the coefficient α in accordance with the number of days that have elapsed from the release date to the present. Specifically, the server device 1 increases the coefficient α as the elapsed days described above are larger. Thereby, the server apparatus 1 increases the value of the cumulative number of purchases multiplied by the coefficient α as the number of days elapsed from the release date increases, and makes it difficult to calculate a high recommendation level. For example, the server device 1 sets the coefficient α to a different value (for example, 0.5 before the lapse and 0. Set to 8). The server device 1 may store a map of the number of days elapsed from the release date and the set value of the coefficient α, and set the coefficient α with reference to the map.

  In another example, the server device 1 determines the coefficient α based on the total number of returned goods instead of or in addition to the number of days elapsed from the release date to the present. For example, the server device 1 stores a map or the like of the total number of returned goods and the setting information of the coefficient α, and sets the coefficient α by referring to the map or the like. At this time, the server device 1 increases the coefficient α as the cumulative number of returned goods is larger.

  In general, the greater the total number of returns, the lower the need for duplicate printing. Therefore, when the total number of returned products is large, the server device 1 increases the value of the total number of purchases multiplied by the coefficient α, makes it difficult to calculate a high recommendation degree, and suppresses duplication. Thereby, the server apparatus 1 can set appropriately the recommendation degree of a duplicate and can suppress suitably that excess inventory generate | occur | produces.

  In yet another example, the server device 1 sets the coefficient α based on the number of publishers in stock in place of or in addition to the above-described elapsed days and the total number of returns. Specifically, the server device 1 increases the coefficient α as the number of publishers in stock increases. In general, when there is a large amount of publisher inventory, it is possible to increase the number of purchases without using the publisher inventory to perform duplication. Therefore, in this example, when the number of publishers in stock is large, the server device 1 increases the cumulative number of purchases multiplied by the coefficient α, and makes it difficult to calculate a high recommendation level. Thereby, the server apparatus 1 can suppress suitably promoting unnecessary duplication.

  The server device 1 may determine the coefficient α by receiving user input of the coefficient α on the sales information screen and receiving the input information from the publisher terminal 4.

  Here, a supplementary explanation will be given of the effect of making the coefficient α variable. FIG. 5 is a graph showing a typical change in the number of sales for each day from the book release date. As shown in FIG. 5, generally, the sales period of books is, in order from the period close to the release date, an initial speed period in which sales accelerate immediately after the release date, a duplication period that is a period in which duplication is first determined, Increased sales period during which sales increase, falling period during which sales begin to decrease, write-down period during which the price of books is lowered to promote reduction in inventory, cutting, etc. It is divided into a cutting period, which is a period for eliminating inventory.

  In the present embodiment, the server device 1 sets the coefficient α to a relatively low value during a period in which sales are expected to increase immediately after release (for example, an initial speed period and a duplication period). Thereby, the server apparatus 1 makes it easy to set the recommendation level of the duplicate to be relatively high, and actively recommends the duplicate. Thereby, the server apparatus 1 can suppress suitably the sales opportunity loss by a stockout.

  On the other hand, the server device 1 increases the coefficient α according to the number of days elapsed from the release date. As a result, when a considerable number of days have passed since the release date, the server device 1 increases the coefficient α to make it difficult to set the recommendation level of the duplicate to be relatively high, and carefully recommend the duplicate. Thereby, the server apparatus 1 can suppress suitably the excess inventory by the misunderstanding of a demand.

[Processing flow]
FIG. 6 is an example of a flowchart showing a processing procedure for displaying the sales information screen.

  First, the server apparatus 1 receives from the publisher terminal 4 a request signal S2 that designates a book for which a sales information screen is to be displayed (step S101). In this case, the server device 1 extracts the bibliographic items and the like of the designated book designated by the request signal S2 from the book information DB 131, and extracts the cumulative sales number, the cumulative purchase number, etc. of the designated book from the sales information DB 132 ( Step S102).

  Next, the server apparatus 1 performs a change point detection process (step S103). The change point detection process will be specifically described with reference to the flowchart of FIG. And the server apparatus 1 determines whether the change point was detected as a result of a change point detection process (step S104). If the server device 1 determines that the change point has been detected by the change point detection process (step S104; Yes), the calculation start point of the predicted value of the cumulative sales number is shifted to the change point (step S105). Based on the transition of the cumulative number of sales starting from (see FIG. 4B), the predicted value and the prediction range of the cumulative number of sales are calculated (step S106). By doing in this way, the server apparatus 1 can calculate the predicted value of the cumulative number of sales corresponding to the sudden change in the sales tendency due to the sales promotion effect, word of mouth, or the like.

  On the other hand, if the server device 1 does not detect the change point by the change point detection process (step S104; No), based on the transition of the cumulative number of sales recorded in the sales information DB 132, the predicted value of the cumulative sales number and the prediction The range is calculated (step S106). In this case, for example, the server device 1 uses the cumulative sales number counted from each date from the date of publication of the book to the present date as a sample, and approximates the transition of the cumulative sales number based on regression analysis, least squares, etc. By calculating the formula, the predicted value of the cumulative number of sales on the duplicate determination reference date is calculated. In step S106, the server device 1 stores the calculated predicted cumulative sales number in the storage unit 13 together with the prediction target date as the past predicted cumulative sales number Sp used in the change point detection process in step S103. Let me.

  Next, the server device 1 determines the recommendation degree using the coefficient α (step S107). Specifically, the server apparatus 1 recommends by comparing the predicted value and the predicted upper limit value of the cumulative number of sales on the duplication determination reference date calculated in step S106 with the cumulative number of purchases multiplied by the coefficient α. Determine the degree. In this case, it is preferable that the server apparatus 1 promotes unnecessary duplication by setting the coefficient α based on at least one of the elapsed days from the release date to the present, the cumulative number of returned goods, the number of publishers in stock, and the like. Can be suppressed.

  Next, the server device 1 generates display information S3 on the sales information screen reflecting the recommendation degree determined in step S107, and transmits the display information S3 to the publisher terminal 4 that is the transmission source of the request signal S2 ( Step S108).

  FIG. 7 is an example of a flowchart of the change point detection process executed in step S103 of FIG.

  First, the server device 1 calculates the past cumulative sales number calculated in the past and stored in the storage unit 13 as the current sales cumulative number Sc extracted from the sales information DB 132 and the predicted value of the cumulative sales number corresponding to the current date. Sp is compared (step S201). Then, the server device 1 determines whether or not the current cumulative sales number Sc is higher than the past predicted cumulative sales number Sp by a threshold Sth or more (step S202). Then, when the server device 1 determines that the current cumulative sales number Sc is higher than the past predicted cumulative sales number Sp by a threshold Sth or more (step S202; Yes), whether or not the change point candidate is already stored in the storage unit 13. Determination is made (step S203). Here, the change point candidate is stored in the storage unit 13 in association with the count number “Ct” obtained by counting the number of consecutive days from the date change point candidate for which the current cumulative sales number Sc is higher than the past predicted cumulative sales number Sp by a threshold Sth or more. It is remembered. Then, when the server device 1 determines that the change point candidate is already stored in the storage unit 13 (step S203; Yes), the count day Ct stored in association with the change point candidate is increased by “1”. (Step S204).

  Next, the server device 1 determines whether or not the count number of days Ct has reached a predetermined number of days (step S205). When the server device 1 determines that the count number of days Ct has reached the predetermined number of days (step S205; Yes), the server device 1 detects a change point candidate as a change point (step S206). That is, in this case, the server device 1 determines that the sales have continued to be higher than expected from the change point candidates, and that the sales trend has changed due to sales promotion effects, word of mouth, etc. A change point candidate that is a date on which the cumulative sales number Sc is higher than the past predicted cumulative sales number Sp by a threshold Sth or more is regarded as a change point. On the other hand, when the server device 1 determines that the count number of days Ct has not reached the predetermined number of days (step S205; No), the server device 1 determines that it is too early to consider the change point candidate as the change point, and performs the processing of the flowchart. finish. By doing in this way, the server apparatus 1 can suppress suitably detecting erroneously the time point when sales fluctuated temporarily although there is no change in the sales trend.

  On the other hand, in step S202, if the current cumulative sales number Sc is not higher than the past predicted cumulative sales number Sp by the threshold value Sth or more (step S202; No), the server device 1 deletes the record if there is a change point candidate record. (Step S207). That is, in this case, the server device 1 regards the sudden change in sales at the change point candidate stored in the storage unit 13 as temporary, and regards the change point candidate as not a change point, and records it in the storage unit 13. Delete the record of the change point candidate.

  In step S203, when the server apparatus 1 determines that the change point candidate is not stored in the storage unit 13 (step S203; No), the current date is regarded as the change point candidate and is set to the initial value “1”. It is stored in the storage unit 13 together with the set count number.

[Display example of sales information screen]
Next, a display example of the sales information screen will be described with reference to FIGS. In the following, the elapsed days Δn is set to “30 days”, and the coefficient α is set to “0.5” when it is less than 90 days from release, and is set to “0.8” after 90 days after release. Will be described.

  FIG. 8 is a display example of a sales information screen displayed by the publisher terminal 4 after 15 days from the release date, which is less than 90 days from the release date. Here, the publisher terminal 4 transmits to the server apparatus 1 a request signal S2 designating the book “Secret of XX” whose release date is “February 24” based on the user input, and receives it as a response. The sales information screen of FIG. 8 is displayed based on the display information S3. The sales information screen of FIG. 8 includes a graph display field 20, a bibliographic information display field 21, an issue information display field 22, a POS information display field 23, and a duplication determination display field 24.

  First, the display in the graph display column 20 will be described.

  The server device 1 displays on the graph display field 20 two-dimensional coordinates with the number of copies related to sales of the designated book as the vertical axis and the date from the release date as the horizontal axis, and the graphs A1 to A2 on the two-dimensional coordinates. A5, B2 to B4 are displayed. Here, the graph A1 is a graph showing the transition of the cumulative number of purchases from the release date (February 24) to the present (March 9) (that is, the cumulative number of products sent to the bookstore), and the graph A2 is , Is a graph showing the change in the number of stocks in the market from the release date to the present, Graph A3 is a graph showing the change in the cumulative number of purchases multiplied by the coefficient α, Graph A4 is the graph from the release date to the present It is a graph which shows the transition of the total number of sales, and graph A5 is a graph which shows the transition of the total number of returned goods from a sale date to the present. The graph B2 is a graph showing the prediction of the change in the number of stocks in the market, the graph B3 is a graph in which the cumulative number of purchases multiplied by the coefficient α is fixed to the current value, and the graph B4 is the cumulative sales. It is a graph which shows prediction of transition of a number.

  The server device 1 calculates an approximate line indicating the transition of the cumulative number of sales based on the transition of the cumulative number of sales, etc., as described in the section “Overlapping determination process” described above, and displays it as a graph B4. In this example, the server device 1 detects March 3 when the cumulative number of sales has changed significantly as a change point, and changes the cumulative number of sales from March 3 to the present (March 9). Based on this, an approximate line of the graph B4 is calculated. As a result, an approximate line indicating the transition of the cumulative number of sales that accurately reflects the sales trend changed from March 3 can be calculated.

  Furthermore, the server apparatus 1 calculates the predicted value and the predicted range of the cumulative number of sales on the duplication determination reference date (April 8) based on the approximate line shown in the graph B4, and the plot 23 indicating the predicted value A range mark 33 indicating the prediction range is displayed. Further, the server device 1 predicts the cumulative number of sales on the duplicate determination reference date (March 29, April 18) when the elapsed days Δn are “20 days” and “40 days”, respectively. The prediction ranges are calculated, and the plots 22 and 24 indicating the prediction values and the range marks 32 and 34 indicating the prediction ranges are respectively displayed.

  Further, the server device 1 predicts a change in the number of in-city stocks indicated by the graph B2 by assuming that the cumulative number of sales changes based on the graph B4 and that the cumulative number of purchases does not change. That is, the server device 1 calculates the number of in-city stocks indicated by the graph B2 as being decreased by the increase in the cumulative number of sales indicated by the graph B4. As a result, in the example of FIG. 8, the commercial stock quantity indicated by the graph B <b> 2 is 0 after March 27. That is, when the cumulative number of sales changes based on the graph B4, the graph B2 indicates that the commercial stock will disappear if the cumulative number of purchases is not increased.

  Next, the bibliographic information display column 21, the issue information display column 22, the POS information display column 23, and the overprint determination display column 24 will be described.

  In the bibliographic information display column 21, the server device 1 extracts the price and release date of the designated book from the book information DB 131 and displays them in the bibliographic information display column 21. In the issuance information display column 22, the server apparatus 1 refers to the sales information DB 132 to recognize the first edition copy number, the duplicate copy number, and the issuance copy number of the designated book, and displays them in the issue information display column 22. . In the example of FIG. 8, the first copy number is 8000 copies, and the overprint copy number is 0. Further, in the POS information display field 23, the server device 1 displays the current cumulative number of products sent to the bookstore (ie, the cumulative total number of purchases), the current cumulative total number of sales, the current total number of returned products, The number of stocks in the market, the current number of publishers in stock, and the number of days elapsed from the release date to the present are displayed. The server device 1 acquires these pieces of information by referring to the sales information DB 132.

  In the duplication determination display column 24, the server apparatus 1 displays the recommendation degree of duplication, the coefficient α, the elapsed days Δn, the reference number of copies, the upper limit number of copies, and the lower limit number of copies. Here, the reference number of copies is the number of copies that serves as a guide for the number of overprints when overprinting. Further, the upper limit number of copies is a recommended number of duplicate copies when it is assumed that the cumulative number of sales reaches the predicted upper limit value at the time of the duplicate determination reference date. Similarly, the lower limit number of copies is the recommended number of duplicate copies when it is assumed that the cumulative number of sales reaches the predicted lower limit value at the time of the duplicate determination reference date.

  In the example of FIG. 8, the server apparatus 1 recommends the degree of recommendation because the predicted value of the cumulative number of sales indicated by the plot 23 is larger than the cumulative number of purchases multiplied by the coefficient α indicated by the graph B3 on the duplication determination reference date. Is set to “◎”.

  In addition, when the cumulative number of sales transitions according to the graph B4, the server device 1 indicates the number of commercial stocks that are deficient (minus) on the duplication determination reference date, that is, the number of additional copies required for the commercial stock numbers. Calculated as the number of copies. Further, when it is assumed that the cumulative sales number becomes the predicted upper limit value corresponding to the upper end position of the range mark 33 on the duplication determination reference date, the server apparatus 1 sets the necessary additional number of copies to the stock quantity in the market as the upper limit number of copies. Set. Furthermore, the server apparatus 1 sets the required additional number of copies to the stock quantity in the market as the lower limit number of copies when it is assumed that the cumulative number of sales will be the predicted lower limit corresponding to the lower end position of the range mark 33 on the duplicate determination reference date. To do.

  Then, the user of the publisher terminal 4 who has viewed the sales information screen of FIG. 8 can preferably grasp that the duplication is necessary because the recommendation level of the duplication determination display column 24 is “◎”. . Further, the user of the publisher terminal 4 who has browsed the sales information screen of FIG. 8 can preferably recognize the guide for the number of duplicate copies based on the guide number of copies in the duplicate determination display column 24. Then, the user of the publisher terminal 4 determines the necessity of the final duplication and the number of duplication copies in consideration of the number of publisher stocks and the like.

  FIG. 9 shows a display example of a sales information screen that is displayed when 90 days have passed since the release date. In the example of FIG. 9, the growth of the cumulative number of sales shown in the graph A3 has slowed, and the cumulative number of returned goods shown in the graph A5 starts to gradually increase.

  In the example of FIG. 9, the server device 1 displays the graphs A <b> 1 to A <b> 5 and B <b> 3 on the two-dimensional coordinates of the graph display column 20 with reference to the sales information DB 132 as in the example of FIG. 8. In addition, the server device 1 displays the graph B4 by calculating an approximate line indicating a predicted value of the cumulative number of sales based on the transition of the cumulative number of sales from the release date to the present. Further, the server device 1 recognizes the duplication determination reference date (June 23) based on the set elapsed days Δn (30 days) and calculates the predicted value and the prediction range of the cumulative number of sales on the duplication determination reference date. Then, a plot 23 and a range mark 33 representing these are displayed. Further, the server device 1 predicts a change in the number of in-stock items on the assumption that the cumulative number of sales changes and the total number of purchases does not change based on the graph B4, and displays the graph B2. Further, the server device 1 displays the bibliographic information display column 21, the issue information display column 22, and the POS information display column 23 with reference to the book information DB 131 and the sales information DB 132.

  Next, the overprint determination display field 24 of FIG. 9 will be described. In FIG. 9, since 90 days have passed since the release date, the server device 1 increases the coefficient α, which was “0.5” in the example of FIG. 8, to “0, 8”. As a result, the cumulative number of purchases (see graph B3) multiplied by the coefficient α functioning as a threshold value for determining the degree of recommendation of the overlapping plates increases as the coefficient α increases. Therefore, it is difficult to calculate a high recommendation degree as the coefficient α increases.

  In the example of FIG. 9, the server apparatus 1 has the predicted upper limit value of the cumulative number of sales corresponding to the upper end position of the range mark 33 lower than the total number of purchases multiplied by the coefficient α indicated by the graph B3. For this reason, the recommendation level in the duplication determination display column 24 is set to “−”. In addition, the server device 1 displays the shortage of the stock quantity in the market (see graph B2) on the duplication determination reference date when it is assumed that the cumulative number of sales shifts to the predicted value indicated by the plot 23 as an approximate number of copies. Yes. In this case, the reference number of copies is a negative value, indicating that the commercial stock remains without being purchased. Similarly, the server apparatus 1 displays the shortage of the stock quantity in the market on the duplicate determination reference date when it is assumed that the cumulative number of sales changes to the forecast upper limit value or the forecast lower limit value as the upper limit number or the lower limit number, respectively. ing.

  As described above, in the example of FIG. 9, the prediction of the cumulative number of sales is caused by increasing the coefficient α according to the passage from the release date and the decrease in the estimated increase in the cumulative number of sales. The upper limit value is lower than the cumulative number of purchases multiplied by the coefficient α. As a result, the server apparatus 1 does not recommend overprinting. Thereby, the server apparatus 1 can suppress suitably that excess inventory is caused by unnecessary duplication.

[Effects of overprint determination system]
According to the overprint determination system 100 of the embodiment, the server apparatus 1 compares the number obtained by multiplying the cumulative number of purchases by the coefficient α with the predicted value and the predicted upper limit value of the cumulative number of sales, thereby duplicating the specified book. The recommendation level is determined and displayed on the sales information screen. At this time, when a sudden change in sales occurs, the server device 1 detects the time when the change has occurred as a change point, and the cumulative number of sales is determined based on the change in the cumulative number of sales starting from the change point. Calculate the predicted value. As a result, the server device 1 accurately predicts the cumulative number of sales even when a sudden change in sales occurs due to sales promotion effects, word of mouth, etc., and presents the recommended degree of reliable overprint to the user. Thus, loss of sales opportunities and excess inventory can be suitably reduced.

[Modification]
Next, a modified example suitable for the above-described embodiment will be described. The following modifications may be applied in any combination to the above-described embodiment.

(Modification 1)
The configuration of the demand prediction system in FIG. 1 is an example, and the configuration to which the present invention is applicable is not limited to this.

  For example, instead of the configuration of FIG. 1, the overprint determination system 100 does not include the server device 1, the publisher terminal 4 includes the book information DB 131 and the sales information DB 132 instead of the server device 1, and the POS terminal 3. The sales information S1 may be received from and the sales information DB 132 may be updated. In this case, the publisher terminal 4 generates the display information of the sales information screen by executing the duplication determination process executed by the server device 1 based on the sales information DB 132 and the like, and displays the sales information screen on the display unit 41. Let In this case, the publisher terminal 4 functions as a “sales prediction device” in the present invention, and the control unit 45 of the publisher terminal 4 performs “cumulative sales number calculation means”, “change point detection means”, “ It functions as a “duplicate recommendation degree determining means”, “output means”, and a program executed by the computer.

  In another example, the server device 1 may be composed of a plurality of servers. For example, the server device 1 includes a server having a book information DB 131, a server for storing the sales information DB 132 and updating it based on the sales information S1, and a server for generating the display information S3 by executing duplication determination processing. Also good. In this case, the server that executes the duplicate version determination process functions as a “sales prediction device” in the present invention, and exchanges information necessary for the assigned process with another server via the network 2.

(Modification 2)
In the embodiment, the server device 1 determines the recommended degree of duplication by comparing the prediction value and the prediction upper limit value of the cumulative number of sales with the total number of purchases multiplied by the coefficient α. Instead of this, the server device 1 compares at least one of the predicted value of the cumulative sales number, the predicted upper limit value, and the predicted lower limit value with the cumulative number of purchases multiplied by the coefficient α. The degree of recommendation may be determined.

  For example, when only the predicted value of the cumulative number of sales is to be compared, the server device 1 sets the recommendation degree to “◎” when the predicted value is equal to or greater than the total number of purchases multiplied by the coefficient α, and the predicted value is the coefficient When less than the total number of purchases multiplied by α, the recommendation level is “−”. In this example, the server device 1 may not calculate the prediction upper limit value and the prediction lower limit value (that is, the prediction range). In another example, when the predicted upper limit value and the predicted lower limit value are to be compared, the server device 1 sets the recommendation degree to “◎” when the cumulative number of purchases multiplied by the coefficient α is lower than the predicted lower limit value. When the value is equal to or greater than the lower limit value and less than the predicted upper limit value, the recommendation level is set to “◯”, and when the value is equal to or higher than the predicted upper limit value, the recommendation level is set to “−”.

  As described above, the server device 1 suitably determines the recommendation level of the duplicated printing by using at least one of the predicted value of the cumulative number of sales, the predicted upper limit value, and the predicted lower limit value. be able to.

(Modification 3)
The server device 1 may detect, as a change point, a time point at which the cumulative sales number starts to decrease instead of or in addition to the time point at which the cumulative sales number starts to increase rapidly. Good.

  In this case, when the period in which the current total sales number Sc is lower than the past predicted total sales number Sp by a threshold Sth or more becomes a predetermined number of days or more, the server device 1 detects the beginning of the period as a change point. In this way, the server device 1 can accurately estimate the transition of the cumulative number of sales even if the sales have fallen sharply due to low evaluations such as word-of-mouth. Excess inventory due to errors can be suitably suppressed.

(Modification 4)
The duplicate version determination system 100 may predict the cumulative number of sales for a product other than a book in the same manner as in the embodiment, and determine whether or not the product needs to be additionally manufactured.

  Even in this case, the server device 1 detects the time when the change occurs when a sudden change in sales of the product due to the sales promotion effect or word-of-mouth occurs, and the sales after the change point are detected. Calculate the predicted value of the cumulative number of sales based on the transition of the cumulative number. Also according to this aspect, the server device 1 accurately predicts the cumulative number of sales, presents a highly reliable recommendation regarding the necessity of additional product manufacture, and suitably reduces the sales opportunity loss and excess inventory of the product. Can be made.

DESCRIPTION OF SYMBOLS 1 ... Server apparatus 2 ... Network 3 ... POS terminal 4 ... Publisher terminal

Claims (7)

A cumulative sales number calculating means for calculating a first predicted value of the cumulative number of sales of the product based on the sales of the product to date;
Based on a comparison between the actual value of the cumulative number of sales and the first predicted value that predicts the actual value of the cumulative number of sales, a change point detecting means for detecting a change point at which the tendency of the sales changes And comprising
When the change point detection unit detects the change point, the cumulative sales number calculation unit calculates a second predicted value of the total sales number of the product based on the sales from the change point to the present. A sales forecasting device.   In the case where the error occurrence date in which the error between the actual value of the cumulative number of sales and the first predicted value in which the actual value of the cumulative number of sales is predicted is equal to or greater than a predetermined value continues for a predetermined number of days. The sales prediction apparatus according to claim 1, wherein the first day of the error occurrence day that continues for the predetermined number of days is detected as a changing point. The change point detecting means includes
Based on the actual value of the cumulative number of sales and the first predicted value obtained by predicting the actual value of the cumulative number of sales, it is determined whether or not the date corresponding to the actual value of the cumulative number of sales is the error occurrence date To store information on the determination in a storage unit,
When it is determined that the current date is the error occurrence date based on the current actual value of the cumulative number of sales and the first predicted value obtained by predicting the current actual value of the cumulative number of sales, the storage The sales prediction device according to claim 2, wherein it is determined whether or not the error occurrence date continues for the predetermined number of days with reference to a section. The product is a book,
A duplicate recommendation degree determining means for determining a recommendation degree of duplicates of the book based on the cumulative number of purchases obtained by accumulating the number of purchases to the book and the second predicted value;
Output means for outputting information of the recommendation level;
The sales forecasting device according to any one of claims 1 to 3, further comprising:   The duplication recommendation degree determination means compares the number obtained by multiplying the cumulative number of purchases by a variable coefficient with at least one of the second predicted value or the predicted range including the second predicted value, The sales forecasting apparatus according to claim 4, wherein a recommendation level of the duplicate version of the book is determined.   A program that causes a computer to function as the sales prediction device according to any one of claims 1 to 5. A control method executed by the sales prediction device,
A cumulative sales number calculating step of calculating a first predicted value of the cumulative number of sales of the book based on the sales of the specified book to date;
A change point detection step of detecting a change point at which the sales trend changes based on a comparison between the actual value of the cumulative number of sales and the first predicted value for which the actual value of the cumulative number of sales has been predicted. And having
When the change point detection step detects the change point, the cumulative sales number calculation step calculates a second predicted value of the total sales number of the product based on sales from the change point to the present. Characteristic control method.

Images