JP2003162619A - Sales prediction apparatus and method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a sales prediction apparatus capable of making appropriate sales prediction while avoiding complexity of processing without reference to existence of sales achievements. <P>SOLUTION: A host computer 6 receives incoming data and outgoing data from a computer 2 in a distribution center performing intensively entering/ dispatching and storing. Store register devices 4 are provided to each store in whole country and sends sales data to the host computer 6. The host computer 6 places an order to a computer in a manufacturer 3, receives above described data to summarize at a predetermined time every day and communicates to a database server 9 to update a database 8. A sale prediction computer 10 is connected to the database server 9 through a LAN and makes the sales prediction based on the data stored into the database 8. <P>COPYRIGHT: (C)2003,JPO

Description

Detailed Description of the Invention

[0001]

TECHNICAL FIELD OF THE INVENTION The present invention relates to a system for predicting sales of goods and services.

[0002]

2. Description of the Related Art In order to sell an efficient product without excess or deficiency of the product, sales forecast is performed to determine the order quantity.

For example, Japanese Patent Laid-Open No. 5-120314 discloses
The following sales forecast system is disclosed. In this system, based on the past sales of the target product,
Predict future sales. Sales of the product for the last 3 months, sales for the last 3 months of the whole country, January (3
The accuracy is increased by predicting the sales in consideration of all sales for (months).

However, this system is premised on that the same product was sold in the previous year, and there was a problem that it was not possible to make a prediction for a product that has no sales record. For this reason, there is a problem that it cannot be applied to cases where there are large fluctuations in the products handled or industries where new products are actively introduced. In addition, since the past sales record of each product is referenced for each product for prediction, when the number of products handled is enormous, the process becomes complicated and difficult to realize. There was also.

Japanese Unexamined Patent Publication No. 8-278959 discloses a system for forecasting sales of products that have no sales record. In this system, products are classified in detail by attributes such as color, pattern, size, grade, and the sales record is recorded. For products that have no sales record, products with the most similar attributes are selected from already recorded products, and sales are predicted based on the sales record of the products. In addition, the sales forecast is made in time series, and when there is a difference between the forecast and the actual result, the forecast is corrected.

[0006] According to this system, it is possible to predict sales for products that have no sales record. However, in order to improve the accuracy of prediction, it is necessary to record detailed attributes of the product, and as a result, there arises a problem that a huge number of attributes must be recorded and accumulated.

The present invention solves the above problems and provides a sales forecasting apparatus capable of making an appropriate sales forecasting regardless of the sales record and avoiding complicated processing. The purpose is to do.

[0008]

[Means for Solving the Problems] (1) (2) (3) The sales forecasting device of the present invention acquires the sales record of each transaction object, and based on the sales record, determines the sales time of each transaction object. As at least one indicator, each transaction object is grouped, based on the sales time of the transaction object that makes the prediction, the transaction object that makes the prediction determines which group the transaction object makes, and the transaction object that makes the prediction, It is characterized in that the sales forecast of the transaction target is calculated based on the sales record of the transaction target group at the past response time which is similar to the time of the forecast.

(4) (5) (6) The sales forecasting device of the present invention is a record in which the sales results of each group are recorded for the transaction objects grouped with the sales time of each transaction object as at least one index. It is configured to be accessible to the department, and based on the sales time of the trading object to make a prediction, it is determined which group the trading object to make the prediction belongs to, and the prediction is made to the trading object to make the prediction. It is characterized in that the sales record of the transaction target group at a past corresponding period having similarity with the period is acquired from the recording unit, and the sales forecast of the transaction target is calculated based on the sales record.

That is, each transaction object is grouped according to its sales time, it is judged to which group the transaction object for which prediction is made belongs, and the sales forecast is made based on the sales record of the group. Since the trading objects are grouped according to the sales period, the classification of the trading objects is clear, and highly accurate forecasting can be performed.
Further, even if the number of items to be handled increases, there is no fear that the classification will increase extremely and the processing will become complicated.

(7) The sales forecasting apparatus of the present invention is characterized in that forecasting is performed for each predetermined unit period, and sales forecasts are calculated for a plurality of unit periods for each predetermined unit period. Therefore, time-series sales forecast can be performed.

(8) The sales forecasting apparatus of the present invention is characterized in that the sales forecasting calculates at least a relative sales ratio in each unit period. Therefore, at least the trend of sales can be predicted in time series.

(9) The sales forecasting apparatus according to the present invention is configured such that a relative sales ratio and a past sales result for an elapsed unit time for which a sales result has already been obtained among a plurality of unit times, and a non-elapsed unit for making a sales forecast. It is characterized by making sales forecasts for non-elapsed unit periods based on the relative sales ratio for the periods.

Therefore, based on the relative sales ratio calculated in time series, it is possible to predict future sales in consideration of the actual results in the elapsed unit time.

(10) The sales forecasting apparatus of the present invention is characterized in that the immediately preceding elapsed unit period is used as the elapsed unit period. Therefore, it is possible to make a sales forecast that accurately reflects the trend of sales results.

(11) The sales forecasting apparatus of the present invention is characterized by using a plurality of unit periods as the elapsed unit period. Therefore, even if the sales result fluctuates due to a sudden special factor, it is possible to make a stable sales forecast that is not affected by the fluctuation.

By using a plurality of unit periods immediately before, it is possible to make a stable sales forecast excluding special factors while reflecting the trend of sales results as much as possible.

(12) The sales forecasting apparatus of the present invention estimates the presence / absence of inventory shortage based on the arrival and sales of the sales results of the elapsed unit time, and when it judges that there is inventory shortage, The feature is that the sales forecast is made by using the corrected sales results obtained by correcting the sales results.

Therefore, it is possible to make an accurate sales forecast in consideration of fluctuations in sales results due to the occurrence of inventory shortage.

(16) The sales forecasting apparatus of the present invention is characterized in that sales forecasting is performed by using sales records at a plurality of corresponding periods. Therefore, it is possible to make a prediction based on a more accurate sales record.

(17) The sales forecasting apparatus of the present invention is characterized in that sales forecasting is performed in consideration of sales results before and / or after the corresponding period. Therefore, the sales forecast can be performed in consideration of the sales results of other groups whose sales period is before or after the sales period of the group to which the target of the transaction for which the prediction is made belongs. That is, it is possible to make a stable sales forecast that is not affected by a time-series trend shift.

(18) The sales forecasting apparatus according to the present invention is based on the overall sales schedule of a transaction target for a plurality of periods and the transition of the sales performance of the transaction target group to which the transaction target belongs at the corresponding time. It is characterized by calculating the sales forecast of the transaction target in each unit period.

Therefore, it is possible to make a time-series sales forecast in each unit period based on the overall sales schedule.

(19) The sales forecasting apparatus of the present invention is characterized in that it calculates the required delivery amount of the transaction subject based on the calculated sales forecasting and outputs delivery order data.

Therefore, the required delivery amount can be calculated based on the predicted sales.

(20) The sales forecasting apparatus of the present invention is characterized in that the predetermined unit period is set shorter than the lag period required from the ordering of the transaction target to the delivery.
Therefore, it is possible to determine the required delivery amount while considering the arrival schedule and sales record of the ordered transaction target.

(21) In the sales forecasting apparatus of the present invention, the required delivery amount is calculated by calculating the delivery amount required in the future unit period in consideration of the above-mentioned shift period, and it is necessary in the future unit period. The delivery amount is characterized by being based on the sales forecast.

Therefore, it is possible to accurately determine the required delivery amount while considering the shift period.

(22) In the sales forecasting apparatus of the present invention, the required delivery amount in the future unit period is equal to or has a predetermined relationship with the forecasted total sales for a predetermined number of unit periods prior to the unit period. Further, it is characterized in that the forecasted inventory amount in the unit period is determined, and the delivery amount is determined based on the forecasted inventory amount.

Therefore, it is possible to determine the appropriate inventory quantity based on the sales forecast and calculate the required delivery quantity. Also, depending on the forecast frequency and forecast accuracy, the appropriate delivery amount is determined by adjusting how many unit periods ahead of the relevant unit period the inventory amount equal to (or in a predetermined relationship with) the forecast sales amount. can do.

(23) The sales forecasting apparatus according to the present invention is characterized in that the sales target fluctuates in sales due to seasonal factors. Therefore, it is possible to perform accurate sales forecasting by grouping by sales time.

(24) The sales forecasting apparatus of the present invention is characterized in that the transaction objects are classified according to their attributes, and are grouped according to the combination of the attributes and the sales period.

Therefore, more appropriate grouping can be performed, and highly accurate sales forecasting can be performed.

(25) The sales forecasting apparatus of the present invention is characterized in that the sales time used for grouping is the sales start time. By grouping according to the sales period, it becomes possible to forecast sales from the beginning of sales.

(26) According to the sales forecasting method of the present invention, the sales target of each transaction object is acquired, and based on the sales time of the transaction object for which the prediction is made, the transaction object for which the prediction is made is any existing transaction object. It is characterized by calculating the sales forecast of the transaction target based on the sales record of the transaction target in the past period that has similarity to the time when the prediction is performed I am trying.

That is, each transaction object is classified according to its sales time, it is determined which existing transaction object the transaction object to be predicted corresponds to, and the sales forecast is made based on the sales record of the transaction object. I am trying. Since the transaction targets are classified according to the sales period, the classification of the transaction targets is clear, and highly accurate prediction can be performed. Further, even if the number of items to be handled increases, there is no fear that the classification will increase extremely and the processing will become complicated.

"Transaction target" means a product or service for which sales are predicted in the present invention. In the embodiment, this is an apparel product.

"Sales forecast" means to forecast unknown sales, regardless of whether it is the past or the future. It is a concept that includes not only product sales forecasts but also demand forecasts such as electricity usage. Further, not only when predicting the absolute amount / absolute number of sales, but like the index shown in the embodiment,
This is a concept that includes the case of predicting the tendency (relative proportion) of sales.

The "sales record" means the quantity actually sold.

The "grouping based on sales time" means to group the trading objects based on the sales time of the trading objects, and only the sales start date such as the sales start date, week, year, and the sales start day. Instead, it is a concept that includes grouping by the end of sales.

The "recording medium on which the program is recorded" means
A recording medium such as a ROM, a RAM, a flexible disk, a CD-ROM, a memory card, and a hard disk recording a program. The concept also includes communication media such as telephone lines and carrier paths. Not only a recording medium such as a hard disk that is connected to a CPU and directly executes a recorded program, but a CD-R that records a program to be executed after once installed in a hard disk or the like.
It is a concept including a recording medium such as OM.

The term "program" is a concept that includes not only a program directly executable by the CPU but also a source format program, a compressed program, an encrypted program, and the like.

[0043]

BEST MODE FOR CARRYING OUT THE INVENTION 1. Overall System Diagram FIG. 1 shows a system configuration when the sales forecasting system according to one embodiment of the present invention is applied to the management of apparel products. In this embodiment, the sales forecasting computer 10
Constitutes a sales forecasting device. The host computer 6 receives the receipt data and the shipment data from the computer 2 of the distribution center that centrally stores and stores and stores the products. Further, the store cash register device 4 is arranged in each store nationwide, and sends sales data to the host computer 6. The host computer 6 also places an order with the maker computer 3.

In this embodiment, the host computer 6
Although the data has been transferred to, the data may be sent via a recording medium such as a flexible disk. Alternatively, the data printed out on paper or the like may be input to the host computer 6. further,
Regarding the data output from the host computer 6,
A printout on a recording medium or paper may be used.

The host computer 6 receives these data, totalizes the data at a predetermined time every day, and communicates with the database server to update the database 8.
The sales forecasting computer 10 is connected to the database server 9 via a LAN and performs a sales forecasting process based on the data accumulated in the database server 8.

2. System Overview FIG. 2 schematically shows the flow of goods and the flow of information when the system of FIG. 1 is applied for sales forecast / order management. Headquarters 18 records sales results before the previous year. The head office 18 makes a sales forecast based on the sales schedule (or target) determined for each product and the past sales transition of the group to which the product belongs. Headquarters 18
Based on the calculated sales forecast, the manufacturer 19 is ordered. The maker 19 produces a product based on the order instruction and delivers the product to the distribution center 12.

Logistics center 12 that received the goods
Ships to the stores 14 and 16 in accordance with instructions from the headquarters 18. From the distribution center 12 to the headquarters 18,
Incoming and shipping data is sent. Each store 14,
At 16, the accepted product is sold. Each store 14,
Sales data is transmitted from 16 to the head office 18.

In the head office 18, each product is grouped based on the sales start time, and the sales transition data is recorded for each group. The head office 18 predicts future sales of products based on the sales data of each product (actual sales record) and the past sales trends of the group to which each product belongs.
Based on the calculated sales forecast, the headquarters 18
An additional order will be placed for item 9. After that, ~ is repeatedly executed.

As described above, highly accurate forecasting of sales and placing an order based on the forecasting of sales make it possible to construct a system with less excess inventory and less loss of sales opportunity.

3. Host Computer FIG. 3 shows the hardware configuration of the host computer 6. The host computer 6 includes a CPU 20, a hard disk 22, a display 24, a communication unit 26, a memory 28, a keyboard / mouse 30, a CD-ROM drive 32, a printer 34, and a flexible disk drive 36.

The communication section 26 is for communicating with the distribution center computer 2, the store cash register device 4, the maker computer 3, and the sales forecasting computer 10. The hard disk 22 stores an operating system and a program for data aggregation processing. Also,
The database 8 is built on the hard disk 22. In the database 8, aggregated sales / inventory data, product master data indicating product prices, and the like are recorded.

The program for data totalization processing is C
It is installed in the hard disk 22 from the CD-ROM 38 or the like via the D-ROM drive 32. FIG. 4 shows a flowchart of a program for the data totaling process.

In step S1, the CPU 20 sells data from the store cash register device 4 of each store (how many products have been sold) and warehousing data from the distribution center computer 2 (how many products have been received). Then, it receives how many pieces have been shipped) and order data (how many products have been ordered) from the sales forecasting computer 10, and stores them in the database 8 of the database server 9. This process is repeated until a predetermined time is reached (step S
2).

When the predetermined time comes, the received data is totaled (step S3). Usually, this aggregation process is often performed at night when sales data and warehousing data are not generated. In the data aggregation process, sales data from each store is aggregated for each product, and the hard disk 2 is used as the number of sales on the day.
Record in 2. Similarly, warehousing data and warehousing data are totaled for each product and recorded in the database 8 of the database server 9. As a result, daily sales, warehousing, and warehousing data are accumulated.

Further, for each product, the cumulative sales for the week (cumulative sales from Monday), the cumulative number of purchases so far, the cumulative number of sales, and the number of stocks are calculated. These data are recorded in the database 8 of the database server 9 (step S4).

FIG. 5 shows an example of recorded data. As shown in the figure, each product is specified by a combination of a brand and a product number. In this embodiment,
One brand contains multiple products, and
The case where multiple brands exist will be described.

FIG. 5 shows the sales accumulated in this way.
An example of inventory data is shown. The number of sales in each week is the total number of sales from Monday to Sunday of the week for each product. In the figure, the product number "62421" is "0 sheets" for the first week, ... "31 sheets" for the 30th week, and 31st.
It is shown that the number of sales is “40” for the week and “48” for the 32nd week. Note that the cumulative number of sales for the week (here, the 32nd week) will be fixed on Sunday night.

The cumulative number of purchases is the total number of arrivals of the product. The cumulative number of sales is the total number of sales up to the present. The stock quantity is the stock calculated by the cumulative purchase quantity-sales quantity-adjusted quantity. This stock quantity is also updated daily. The adjusted number refers to the number of items in which the inventory has decreased due to factors other than sales, such as the number of damaged products and lost products. Data of such products are also sent from each store.

The sales start week is a week in which the number of sales is not 0 at first (that is, the first sale week).

Further, in the database 8, not only the sales / inventory data for this year as described above, but also the sales performance data for the previous year or earlier are recorded. Figure 6 shows an example of sales performance data for the previous year. Although not shown, sales performance data for the years prior to last year is also recorded.

4. Hardware Configuration of Sales Forecasting Computer FIG. 7 shows the hardware configuration of the sales forecasting computer. The basic configuration is similar to that of the host computer.
However, a program for sales forecast is recorded on the hard disk 54. In this embodiment,
At the beginning of the year (or the end of the previous year), a week index (sales index) for forecasting sales is calculated in advance and recorded in the hard disk 52.

5. Week Index Calculation Processing FIG. 8 shows a flowchart of the week index calculation program executed at the beginning of the year. First, the CPU 50 accesses the database 8 and selects from last year's sales result data,
Data of products with a large number of sales is extracted (step S1)
1). In this embodiment, when the products are arranged in the descending order of the number of sold products, the product having the cumulative sales composition ratio of 70% or more is selected. In this way, by extracting and using the sales performance data of the products with a large number of sales, it is possible to grasp the accurate sales transition. Figure 9
A shows an example of the extracted sales performance data for the previous year.

Next, the merchandise of the sales performance data is grouped by item and by sales start week (step S1).
2). Here, items are skirts, jackets,
A group of products such as pants. For example, in the data of FIG. 9A, only the skirt of item number “01” is shown, but similar data exists for other items such as jackets and pants. The sales start week is the week when the product starts to sell. In other words, it's the week when the sales of more than one piece is the first time.

In the data of FIG. 9A, it is common that the sales start week of the products in the first and second rows is "01" (that is, the first week). Therefore, the products in the first and second rows are grouped as one group and their sales results are totaled. The data obtained by such grouping is shown in FIG. 9B. In this way, by item,
You can get the results of last year by grouping by the week when sales start.

Next, the CPU 50 calculates the average S _avg (Item, Ws, Wf) together with the group sales data in which the sales start week is one week before and after the sales result data of each week (step S13). That is, the average of sales results is calculated by the following formula.

S _avg (Item, Ws, Wf) = (S (Item, Ws-1, Wf) + S (I
tem, Ws, Wf) + S (Item, Ws + 1, Wf)) / 3 where Item is the item number and Ws is the sales start week. That is, Item and
The group is specified by Ws and. Wf is the number of weeks when the sales start week is the “first week”. S _avg (Item, Ws, W
f) is Wf in the group specified by Item, Ws
It is the average of sales results for the week. S (X, Y, Z) is the sales record in the Zth week of the group in which the sales start week of the item X is Y.

FIG. 10 shows the average sales record including the preceding and following weeks. In addition, in this figure, only the data up to the 7th week is shown, but if there is a sales record after the 8th week,
These data are also recorded. In this way, by averaging the sales results for the preceding and following weeks, even if there is a sudden change in the sales results, this can be averaged to reduce the effect.

Further, in this embodiment, as will be described later, the product sales forecast is made based on the sales start week. Therefore, it is preferable that all the weeks are the sales start weeks of any of the product groups. However, for an item for which a sufficient number of sales record data cannot be obtained, there may be a case where data regarding a specific week as the sales start week cannot be obtained. For example, as shown in FIGS. 9A and 9B, there is no product whose start week is the fourth week. However, by calculating the average sales performance including the group starting the week before and after,
Data can be obtained as shown in FIG. 10A. Note that the sum of S _avg (Item, Ws, Wf) for all weeks is T (Item, W
s).

After calculating the average sales record as shown in FIG. 10A as described above, the weekly index (sales index) is calculated for each group (step S1).
4). The week index Index (Item, Ws, Wf) is calculated by the following formula.

Index (Item, Ws, Wf) = S _avg (Item, Ws, Wf) / T (Item, Ws) An example of the index Index (Item, Ws, Wf) calculated by the above,
Shown in FIG. 10B.

Further, the CPU 50 performs a weighted average of the week index of 2 years ago and the week index of 3 years ago calculated in the same manner as above together with the week index of the previous year (step S15). In this embodiment, the week indexes of the previous year, two years ago, and three years ago are weighted at 5: 3: 2, respectively, and the average value is calculated. FIG. 11 shows a weekly index obtained by weighted averaging the weekly indices for three years. The CPU 50 records this weighted average weekly index on the hard disk 52.

By weighted averaging the weekly indexes for three years, it is possible to create an index that reflects the latest sales trend while eliminating the fluctuations in sales due to special factors.

As described above, the weighted average week index for sales forecast is recorded in the hard disk 52 of the sales forecast computer 10.

6. Sales Forecast / Order Quantity Calculation Process 6.1 Condition Setting FIG. 12 shows a flowchart of the sales forecast / order quantity calculation process program recorded in the hard disk 52 of the sales forecast computer 10.

First, the CPU 50 displays a condition setting screen on the display 54 and prompts for input (step S).
21). FIG. 13 shows a screen for setting conditions. On this screen, the lead time (period required from ordering to warehousing) can be entered as the number of weeks for each material of the product.

"K" is a product using a knit material, "C"
Is a product that uses materials that are mainly cut and used, "J"
Indicates products that use materials used for trainers, etc., and “Miscellaneous goods” indicate products such as shoes, mufflers, and hats. In the example in the figure, "K" has a lead time of 4 weeks,
A lead time of 3 weeks is entered for “C”, a lead time of 3 weeks for “J”, and a lead time of 3 weeks for “miscellaneous goods”. These lead times are used when determining the order quantity.

Also, in the column of the missing item correction, "valid"
You can select either "disable" or "disable". here,
Out-of-stock correction refers to correcting the actual sales on the assumption that the sales could not be performed due to a shortage at the store when the sales results of the previous week were calculated for sales forecast. If "valid" is set, the stockout correction is performed. If "invalid" is set, the stockout correction will not be performed.

Further, the column of "expected inventory week number" is a numerical value indicating how many weeks of sales have inventory. In the figure, it is specified to determine the inventory quantity so as to have the inventory corresponding to the sales for four weeks.

Each of the above information is recorded in the hard disk 52 and used in the sales forecast / order quantity determination process.

6.2 Sales Forecasting After setting the conditions as described above, sales are forecasted (step S22). Flow charts of the sales forecasting process are shown in FIGS. 14 and 15. In this embodiment, the sales forecast in the state where there is no sales record for the product (first sales forecast) and the sales forecast after the sales record for the product is obtained (additional sales forecast) are processed separately. is doing. A flowchart of the initial sales forecast is shown in FIG. 14, and a flowchart of the additional sales forecast is shown in FIG.

In the initial sales forecast, first, the CPU 5
0 acquires a predetermined sales plan for this year for the product for which the sales forecast is made from the database 8 (step S31). This sales schedule is determined based on the sales target of the brand, the characteristics of the product, etc.
It is recorded in the database 8.

Next, the planned sales start date of the product is acquired from the database 8. The planned sales start date is also determined in advance based on the characteristics of the product and recorded. Based on the planned start date of sale and the item classification of the product, it is determined which group the product belongs to, and the weighted average week index of the group is acquired from the hard disk 52 (step S32).

Subsequently, the CPU 50 calculates the sales SP (good, weekly) based on the sales schedule and the weighted average week index of the product.
Wf) is predicted (step S33).

SP (good, Wf) = ST × Index (Item, Ws, Wf) Here, SP (good, Wf) is the sales forecast quantity of the product good in the Wf week from the scheduled sales start week. . ST is the total sales forecast for this product in the current term, and Index (Item, Ws, Wf) is the weighted average weekly index. Item Item is an item classification to which the product good belongs, and the planned start week Ws is the planned sales start week of the product good. In other words, the total sales schedule is distributed to each week according to the weighted average week index, and the sales forecast for each week is performed.

The CPU 50 records this sales forecast in the hard disk 52 and outputs it in the form of printout or data (step S34).

These processes are executed for all products for which sales are predicted. This process is usually
It is executed at any time after the planned sales quantity of each product is determined. As a result, sales forecast data for each product as shown in FIG. 15 is recorded in the hard disk 52. In the figure, SKU
Is a code that identifies each product.

The sales forecast of each product thus calculated is recalculated based on the sales record when the sales record for the product occurs. The process is shown in the flowchart of FIG. The sales forecast process of FIG. 16 is normally executed on the first day of each week (Monday in this embodiment).

First, the CPU 50 obtains the performance data of each product from the database 8 (step S41).
Further, the digestion rate and the like are calculated based on the acquired data. FIG. 17A shows an example of the acquired result data. These data are obtained by receiving data from each store or distribution center, totaling them by the server computer, and recording them in the database 8.

The "product number", "SZ", and "COL" indicate the product number, size, and color of each product, and each product can be specified by these. The above SKU is obtained by combining these three. In the “material name”, the categories of “K”, “C”, “J”, and “miscellaneous goods” described above are recorded. In the sale start week, the sale start week of the product is recorded. “Upper” is the standard selling price at the store.

In the "5 weeks ago" column, the number of sales for the week 5 weeks ago is recorded. "4 weeks ago""3 weeks ago""2 weeks before"
The “previous week” is the number of sales in four weeks ago, three weeks ago, two weeks ago, and the previous week, respectively. The “cumulative purchase amount” is the cumulative number of purchased products up to the previous week for the product. "Total sales"
Is a cumulative total of the number of sales of the product up to the previous week. “Weekend inventory” is the inventory as of the previous weekend.
"This week's arrival" is the quantity that has already been ordered and is scheduled to arrive this week. Similarly, "arrival next week""arrival three weeks later"
The “arrival after 4 weeks” is the quantity to be received next 3 weeks, 4 weeks later.

The CPU 50 calculates the proportion of the purchased number of sheets sold (digestion rate) based on the cumulative purchase and the cumulative sales, and records it in the "cumulative digestion rate" column. In addition, based on the purchased number of the previous week and the sold number of the previous week, calculate the percentage of the purchased number (single week digestion rate) and record it in the "Single week digestion" column To do. In this embodiment, the weekend inventory is added to the previous week's sales, and the previous week's sales are divided by this to obtain the single-week digestibility.

Next, the CPU 50 determines whether the out-of-stock correction set in FIG. 13 is valid (step S42). If it is not valid, the stockout correction is not performed, and the process proceeds to step S45. If the out-of-stock correction is set to be valid, step S43 and the subsequent steps are executed to perform out-of-stock correction.

In some stores, the product sells more than expected, and there may be a situation (stockout) in which the product cannot be sold due to lack of inventory. When such a situation occurs, the sales record does not reflect what it should be, so if sales forecasts are made using this, there is a possibility that erroneous forecasts will be made. Therefore, in such a case,
Modify the sales results used for sales forecast. In this embodiment, this is called out-of-stock correction.

First, in step S43, the CPU 50 estimates whether such a stockout has occurred.
In this embodiment, the presence or absence of a missing item is estimated based on the cumulative digestibility and the single-week digestibility. In other words, if the cumulative digestibility exceeds 60% and the single-week digestibility also exceeds 60%, it is determined that a shortage has occurred at any of the stores.

If it is determined that no shortage has occurred, the process proceeds to step S45. When it is determined that the product is out of stock, the sales performance of the previous week is corrected according to the following formula (step S44).

Last week's sales = ((previous week's sales / 0.6)-(weekend inventory + previous week's sales)) × 0.6 + previous week's sales As a result, the previous week's sales of the products in the first row of FIG. 17A are as follows:
Corrected to 386.4 (see the column for sales of corrected shortages). Further, the previous week's sales of the products in the second and third rows that do not satisfy the above conditions are not corrected and are left as 277 and 241.

Next, the CPU 50 calculates the sales forecast value SG (good, n) after n weeks for each product according to the following formula.

SG (good, n) = (SR (good) / Index (Item, W
s, W _c-1 )) × Index (Item, Ws, W _{c + n} ) where SR (good) is
This is the sales record for the previous week (if there is a correction for missing items, it is after correction) _{Index (Item, Ws, W c} -1) is, goods Item, a weekly index of the previous week W _c-1 of the start of the week the group Ws. W _c means this week and W _{c + n} means the week n weeks after this week. In this embodiment, this week is the 39th week. In addition, n is 0
It changes to ~ 6 and makes sales forecast for each week from this week to 6 weeks ahead. This is done for each product.

An example of the week index (three-year weighted average) used for the prediction is shown in FIG. 17B. The calculated sales forecast for each week is shown in FIG. From "39 weeks" (this week),
Sales forecasts for "45 weeks" have been calculated.

6.3 Calculation of Required Order Quantity After initial or additional calculation of the sales forecast for each week as described above, the CPU 50 calculates the necessary order quantity (step S23 in FIG. 12).

FIG. 19 schematically shows the concept for calculating the required order quantity in this embodiment. This week (39 weeks)
The products to be ordered will arrive in 42 weeks (leading week that can be controlled) after the lead time LT. This lead time LT is set in FIG.

In this embodiment, from this week (39th week),
When the week (42 weeks) ahead of the lead time LT of the product starts, the order quantity of this week is determined so that the inventory has the forecast sales for the last 4 weeks. The number of weeks before which the inventory has the forecast sales is called the forecast inventory week number. That is, in this embodiment, the forecast inventory week number is set to "4". In other words, based on the sales forecast, the stock is secured in an amount that can cover the sales up to four weeks ahead.

FIG. 20 shows a flowchart of this required order quantity calculation program. CPU50 is this week (39th week)
To the controllable first week (42th week), the sales forecast values for the product are summed to be "sales forecast 1" (step S51). For example, as shown in FIG. 18, for the products in the first row, the sales forecasts for the 39th to 41st weeks are totaled, and “773” is obtained and recorded.

Next, the planned arrival numbers for the same period are totaled (step S52). In other words, the quantity of products already ordered and not yet received is summed up to be “work in process 1”. This is recorded in the database 8,
It can be obtained based on the order record for each product and the arrival record. In FIG. 18, “Work in progress 1” is “110”.
It is calculated and recorded as "0".

Next, the stock quantity at the time when the controllable first week (42th week) starts is calculated as "stock 1" (step S53). This "stock 1" can be calculated by adding "work in process 1" to "stock of last weekend" and subtracting "sales forecast 1".

Next, the sales forecasts of the controllable first week (42th week) to the forecast inventory week period are totaled (step S5).
4). That is, the sales forecasts for 42 weeks to 45 weeks are totaled. Then, "work in process 1" is subtracted from "total work in process inventory" to calculate "work in process 2" (step S55). Here, the “total work-in-process inventory” refers to the quantity currently in work (ordered and not yet received) for the product.
That is, “work in progress 2” indicates a work in progress scheduled to arrive after the controllable first week.

If the "stock 1" is greater than 0, the CPU 50 calculates the "order frame" by the following formula (step S5).
6, S57).

"Order frame" = "Sales forecast 2"-("Inventory 1" + "Work in process 2") If "Inventory 1" is 0 or less, "Order frame" is calculated by the following formula ( Step S56,
S58).

“Ordering frame” = “sales forecast 2” − “work in process 2” If the “ordering frame” obtained as a result of the above calculation is larger than 1, the number of orders at this time is set to this “ordering frame” (step S
59, S60). Also, if the "ordering frame" is 0 or less,
This time, no order is placed (steps S59 and S61).

FIG. 18 shows an example of the calculated "order frame" and "order quantity". The products on the first line and the products on the second line are
There is no order this time. It is indicated that the products in the third row should be ordered with 270 pieces.

The calculated order quantity and the like are output to the database, recorded, and printed out.
Further, it is transmitted to the maker 3 via the host computer 6. The maker 3 receives this and performs production (see FIG. 1).

[0112] In the above, "ordering frame" = "number of orders"
However, the number of orders may be determined by a human with the "order limit" as the upper limit.

In the above, the ordering instruction is transmitted online to the maker 3, but the ordering instruction may be transmitted by a recording medium or printout paper.

7. Other Embodiments (1) In the above embodiment, the apparel product has been described as an example, but the present invention is also applied to other products for which management from production to sales is effective, for example, industrial products, processed food products, etc. be able to. Further, it can be applied to order management of products for which purchase management is important, for example, products placed in convenience stores.

(2) In the above embodiment, the order quantity is calculated based on the sales forecast, but the production quantity, the purchase quantity, the acquisition quantity, etc. may be determined based on the sales forecast.

(3) In the above embodiment, the sales forecast number is calculated, but the sales amount, the sales amount, etc. may be calculated.

(4) In the above embodiment, the order quantity is calculated also based on the sales forecast for the initial order of the product, but the initial order quantity is determined by another method (for example,
Only the additional order quantity and the breakdown from the overall plan) may be determined based on the sales forecast.

[Brief description of drawings]

FIG. 1 is a diagram showing an entire system.

FIG. 2 is a diagram showing an outline of a business model.

FIG. 3 is a diagram showing a hardware configuration of a host computer.

FIG. 4 is a diagram showing a flowchart of a data totaling process.

FIG. 5 is a diagram showing sales / inventory data.

FIG. 6 is a diagram showing sales / inventory data.

FIG. 7 is a diagram showing a hardware configuration of a sales forecasting computer.

FIG. 8 is a diagram showing a flowchart of a sales index creation process.

FIG. 9 is a diagram showing sales record data.

FIG. 10 is a diagram showing a calculation process of a week index.

FIG. 11 is a diagram showing an example of a week index.

FIG. 12 is a diagram showing a flowchart of sales forecast / order quantity calculation processing.

FIG. 13 is a diagram showing the contents of condition setting.

FIG. 14 is a diagram showing a flowchart of a first-time sales forecasting process.

FIG. 15 is a diagram showing a sales forecast for each product.

FIG. 16 is a diagram showing a flowchart of sales forecasting processing at the time of addition.

FIG. 17 is a diagram showing actual data and weekly index data.

FIG. 18 is a diagram showing sales forecast and order data.

FIG. 19 is a diagram schematically showing an order quantity calculation method.

FIG. 20 is a diagram showing a flowchart for calculating a required order quantity.

[Explanation of symbols] 2 ... Logistics center computer 3 ... Manufacturer 4 ... Store cashier 6 ... Host computer 9 ... Database server 10 ... Sales forecasting computer

Claims (26)

[Claims] 1. A sales forecasting device for forecasting future sales based on past sales results, which obtains the sales results of each transaction target, and based on the sales results, the sales time of each transaction target. As at least one indicator, each transaction object is grouped, and based on the sales time of the transaction object that makes the prediction, it is determined which group the transaction object that makes the prediction belongs to, and the transaction object that makes the prediction, A sales forecasting apparatus, which calculates a sales forecast of a transaction target based on a sales record of the transaction target group at a past response time having a similarity with the time of performing the forecast. 2. A sales forecasting program for realizing a sales forecasting apparatus using a computer, which obtains the sales record of each transaction target, and based on the sales record, at least one sales period of each transaction target. As one index, each trading object is grouped, based on the sales time of the trading object that makes the prediction, it is determined which group the trading object that makes the prediction belongs to, and the prediction is made for the trading object that makes the prediction. A program for causing a computer to perform a process of calculating a sales forecast of a transaction target based on the sales record of the transaction target group at a past response time that is similar to the time when the transaction is performed. 3. A recording medium on which the program according to claim 2 is recorded. 4. A sales forecasting device for forecasting future sales based on past sales results, wherein transaction objects grouped with the sales time of each transaction object as at least one index, for each group. It is configured to be accessible to the recording unit that records the sales results, and based on the sales time of the transaction target for which the forecast is made, it is determined which group the transaction target for the forecast belongs to, and the transaction for which the forecast is made. For the target, obtain the sales record of the transaction target group at the past response time that is similar to the forecast time from the recording unit, and calculate the sales forecast of the transaction target based on the sales record. A featured sales forecasting device. 5. A sales forecasting program for realizing a sales forecasting device using a computer, to which group the trading subject making the forecasting belongs based on the sales time of the trading subject making the forecasting. For each transaction object grouped with the sales time of each transaction object as at least one index, access the recording unit that records the sales performance of each group, and forecast the transaction object for which the forecast is made. A program for obtaining the sales record of the transaction target group from the recording unit in the past corresponding time having a similarity with the above, and causing the computer to perform the process of calculating the sales forecast of the transaction target based on the sales record . 6. A recording medium on which the program according to claim 5 is recorded. 7. The apparatus, program or recording medium according to claim 1, wherein the forecasting period is for each predetermined unit period, and the sales forecast for a plurality of unit periods is for each predetermined unit period. What is characterized by calculating. 8. The apparatus, program or recording medium according to claim 7, wherein the sales forecast calculates at least a relative sales ratio in each unit period. 9. The apparatus, program or recording medium according to claim 8, wherein the relative sales ratio and the past sales result and the sales forecast for the elapsed unit time for which the sales result has already been obtained among the plurality of unit times are calculated. It is characterized by making a sales forecast for the non-elapsed unit period based on the relative sales ratio for the non-elapsed unit period. 10. The apparatus, program or recording medium according to claim 9, wherein the immediately preceding elapsed unit period is used as the elapsed unit period. 11. The apparatus, program or recording medium according to claim 9, wherein a plurality of unit periods are used as the elapsed unit period. 12. The apparatus, program or recording medium according to any one of claims 9 to 11, wherein it is estimated whether there is a stock shortage based on the arrival and sales of the sales record of the elapsed unit time, and the stock shortage occurs. If it is determined that there is, the sales forecast is made by using the corrected sales record obtained by correcting the sales record. 13. The apparatus, program or recording medium according to claim 7, wherein the predetermined unit time is time, time zone, day, week, month or year. 14. The apparatus, program or recording medium according to claim 1, wherein the corresponding time is before the previous year corresponding to the time of making a prediction,
Characterized by being before the previous month or before the previous day. 15. The apparatus, program or recording medium according to any one of claims 1 to 14, wherein the similarity of the corresponding times includes commonality of day, month, week, day of the week, time or climate. What to do. 16. The apparatus, the program, or the recording medium according to claim 14 or 15, wherein sales forecast is performed by using sales records at a plurality of corresponding periods. 17. The apparatus, program or recording medium according to any one of claims 1 to 16, wherein the sales period is before or after the sales period of the group to which the transaction object performing the prediction belongs, or both of them. It is characterized by making sales forecasts in consideration of the sales results of the group. 18. The apparatus, program or recording medium according to any one of claims 7 to 17, wherein an overall sales schedule of a plurality of periods for the transaction target and a corresponding period of a transaction target group to which the transaction target belongs. It is characterized by calculating the sales forecast of the transaction target in each unit period based on the transition of sales results. 19. The apparatus, program or recording medium according to any one of claims 1 to 18, wherein the required delivery amount of the transaction target is calculated based on the calculated sales forecast, and delivery order data is output. What to do. 20. The apparatus, program or recording medium according to any one of claims 1 to 19, wherein the predetermined unit period is set to be shorter than a shift period required from ordering to delivery of the transaction target. What to do. 21. The apparatus, program or recording medium according to claim 1, wherein the required delivery amount is calculated by calculating a delivery amount required in a future unit time in consideration of the deviation period. Yes, the required amount of delivery in the future unit period is based on the sales forecast. 22. The apparatus, program or recording medium according to claim 21, wherein the required amount of delivery in the future unit time is:
Determining the forecasted inventory amount in the relevant unit period so as to be equal to or have a prescribed relationship with the forecasted total sales for the prescribed unit period prior to the relevant unit period, and determine the delivery amount based on the forecasted inventory amount. Characterized by. 23. The apparatus, program or recording medium according to any one of claims 1 to 22, wherein the transaction target has a fluctuation in sales due to seasonal factors. 24. The apparatus, program or recording medium according to any one of claims 1 to 23, wherein the transaction objects are classified by their attributes, and are grouped by a combination of the attributes and the sales period. Characterized by 25. The apparatus, program or recording medium according to any one of claims 1 to 24, wherein the sales time used for the grouping is a sales start time. 26. A sales forecasting method for predicting future sales based on past sales records by using a computer, wherein the sales record of each transaction target is acquired, and the forecast is made at the sales timing of the trading target. Based on this, it is determined which of the existing trading objects corresponds to the trading object for which the forecast is made, and for the trading object for which the forecast is made, the sales of the trading object at a past time that is similar to the time of making the forecast A sales forecasting method, which comprises calculating a sales forecast of the transaction target based on a track record.