JP2003346070A - Demand prediction method and demand prediction system - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a demand prediction method and a demand prediction system with improved demand prediction precision. <P>SOLUTION: This demand prediction system is provided with a demand environmental condition information database 5b accumulating demand environmental condition information affecting increase/decrease of a demand for a coming prediction time from a past time, in which an actual result is counted, a base data extraction means 4a extracting base data, from which a variation due to influence of the demand environmental condition information is removed from an actual value of a demand in the past time, a base forecast value extraction means 4c applying a forecast model representing a temporal variation tendency of the demand to the base data for finding prediction values of the base data in the coming prediction time, and a correction processing means 4d performing correction while taking the variation due to the influence of the demand environmental condition information to the prediction values of the base data into consideration for finding a forecast value of the demand in the coming forecast time. <P>COPYRIGHT: (C)2004,JPO

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a demand forecasting method and a demand forecasting system for forecasting the demand of a product.

[0002]

2. Description of the Related Art A demand forecasting system of this type includes:
For example, there has been proposed a method of predicting, using a statistical method, the demand of a commodity such as water, electric power, or seasonal commodity whose demand varies greatly depending on the season (for example, Japanese Patent Laid-Open No. -56890). In the system described in the above-mentioned publication, a moving average process is performed on the actual value of the demand amount of the target product to obtain a trend change in the demand amount, and a regression order applicable to the trend change is obtained. Seasonal adjustment is performed to obtain time-series data representing trends and irregular fluctuations, and to the obtained time-series data, the regression equation of the regression order with confidence limits obtained in the above process is applied to obtain future demand. Had predicted the amount.

[0003]

In the demand forecasting system described above, the demand is forecasted using a statistical method based on the actual demand. Changes due to steady-state factors such as gender, sales promotion and campaign activities, and changes due to sudden factors such as disasters are included.When forecasting based on actual values that include these changes, these changes There is a problem that the accuracy of the demand forecast is reduced due to the influence of the minute.

[0004] The present invention has been made in view of the above problems, and an object of the present invention is to provide a demand forecasting method and a demand forecasting system in which the demand forecasting accuracy is improved. .

[0005]

In order to achieve the above object, according to the first aspect of the present invention, there is provided a demand forecasting method for predicting a future demand based on an actual value of a past demand. ,
The process of inputting the influencing factor parameters that affect the increase or decrease in demand over the past period in which the performance was recorded from the past period to the future forecast period, and the change due to the influence of the influencing factor parameter from the actual value of demand in the past period The process of extracting the base data of the demand quantity that is not affected by the influencing factor parameters and applying the prediction model representing the temporal change tendency of the demand quantity to the base data to predict the base data in the future forecast period It is characterized by a process of obtaining a value and a process of performing a correction process on the predicted value of the base data in consideration of a change due to the influence factor parameter to obtain a predicted value of a demand amount in a future prediction period. And

According to a second aspect of the present invention, there is provided a demand forecasting system for forecasting a future demand based on a past demand actual value. Means for inputting the influencing factor parameter that affects the increase and decrease of the demand over a period of time, and the demand that is not affected by the influencing factor parameter by removing the change due to the influencing factor parameter from the actual amount of demand in the past A base data extracting means for extracting a base data of the amount, and a base prediction value extracting means for applying a prediction model representing a temporal change tendency of the demand amount to the base data to obtain a predicted value of the base data in a future prediction period. , The forecast value of the base data is corrected taking into account the change due to the influence factor parameter, and the demand And characterized in that it comprises a correcting means for obtaining a predicted value.

According to a third aspect of the present invention, the processing steps by the base data extracting means, the base predicted value extracting means, and the correction processing means are displayed to allow the person in charge to confirm the processing contents. A check means is provided.

According to a fourth aspect of the present invention, in the third aspect of the present invention, the influence factor parameter includes a parameter of a stationary factor whose occurrence time can be predicted and a parameter of a sudden factor that occurs suddenly. The parameters are
Past history information used to extract base data from actual values in the past period in which actual results were recorded, and future schedule information used to predict future demand from predicted values of base data in future prediction periods Characterized by comprising:

According to a fifth aspect of the present invention, in the fourth aspect of the present invention, when the daily demand is predicted, a day-of-week index indicating the influence of a change in the demand due to the day of the week is used as a stationary factor parameter. It is characterized by using.

According to a sixth aspect of the present invention, in the fourth aspect of the present invention, when predicting a future demand in a specific area, an area which indicates an event affecting the demand only in the area as an influencing factor parameter. It is characterized by using sex factor information.

According to a seventh aspect of the present invention, in the second aspect of the present invention, the prediction model is at least one of a statistical model created by using a statistical method and a pattern model created from a change tendency of the actual value. And a prediction model selecting means for allowing an operator to select a prediction model to be used from a plurality of prediction models registered in advance.

According to an eighth aspect of the present invention, in the seventh aspect of the present invention, the predictive model selecting means sorts the models selected so far for a plurality of specified types of products in the order of decreasing frequency of selection. It is displayed as a selection candidate.

According to a ninth aspect of the present invention, in the invention of the seventh aspect, the prediction model selecting means sorts the models selected so far for the plurality of types of designated products in the order of the selected date and time in the new order. It is displayed as a selection candidate.

According to a tenth aspect of the present invention, in the seventh aspect of the present invention, the prediction model selecting means selects another product whose demand quantity change tendency in a past predetermined period is similar to the product to be predicted. The prediction model is displayed as a selection candidate.

According to an eleventh aspect of the present invention, in the second aspect, the base predicted value extracting means shifts the prediction model so that an index indicating a degree of deviation between the prediction model and the base data is equal to or less than a predetermined value. And apply it to base data.

According to a twelfth aspect of the present invention, in accordance with the eleventh aspect of the present invention, there is provided a notifying means for prompting a change of the selected prediction model when the index is larger than a predetermined upper limit.

According to a thirteenth aspect of the present invention, in the second aspect of the present invention, a database for storing correction rules created from the experience of the person in charge is provided, and the correction processing means includes the correction rules stored in the database and the influence factor parameter. And the correction processing is performed using

According to a fourteenth aspect of the present invention, in the second aspect of the present invention, the correction processing means performs the correction processing by taking into account the change due to the influence factor parameter with respect to the base data and its predicted value. And a comparison means for comparing correction data obtained by performing a correction process on the base data in the past period in which the performance was recorded with the performance value.

[0019]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS (Embodiment 1) FIG. 1 shows a block diagram of a system employing a demand forecasting method according to the present invention.

In this embodiment, a demand forecasting system is constructed by one computer 1.
It comprises an input device 2 composed of a pointing device such as a keyboard and a mouse, an output device 3 composed of a printer and a display device, an arithmetic device 4, and a storage device 5 such as a hard disk.

The storage device 5 stores three databases: an actual demand amount database 5a, a demand environment condition information database 5b, and a prediction model database 5c. The demand amount actual database 5a stores the actual value of the demand amount in the past period in which the actual result was recorded. In the demand environment condition information database 5b,
Influence factor parameters (hereinafter referred to as demand environment condition information) that affect the increase and decrease of the demand amount from the past period in which the results were recorded to the future prediction period are stored. In addition, the prediction model database 5c stores a plurality of prediction models that indicate a tendency of a change in the demand amount from a past period in which the results are recorded to a future period.

On the other hand, the arithmetic unit 4 calculates, based on the actual values in the past period stored in the actual demand amount database 5a,
Base data extracting means 4 for extracting the base data of the demand amount from which the change due to the influence factor parameter is removed.
a, and a prediction model selection unit 4b that allows a person in charge to select a prediction model that matches the change tendency of the base data extracted by the base data extraction unit 4a from among a plurality of prediction models registered in the prediction model database 5c. A base prediction value extracting means for calculating a prediction value of the base data in a future prediction period by applying the prediction model to the base data; and correcting the prediction value of the base data by taking into account a change due to the influence of the influence factor parameter. Do the processing,
Correction processing means 4d for obtaining a predicted value of a future demand.

FIG. 3 shows a procedure for estimating the demand of a certain product using the present system. First, while checking an interactive interface screen as shown in FIG. After performing the operation of selecting the prediction model (step S1), the arithmetic unit 4 uses the actual value of the demand amount, the demand environment condition information, and the selected prediction model.
Executes a demand amount prediction process, and outputs the obtained predicted value to the output device 3 (step S2).

FIG. 4 is an example of a screen displayed on the display device as the output device 3, and is a graph G1 showing a process and a result of a process of extracting base data from actual demand values.
The graph G2 showing the process and result of the process of obtaining the base predicted value by applying the prediction model to the base data is displayed on the left side of the screen, and the predicted value of the future demand is obtained by correcting the base predicted value in the center of the screen. A graph G3 and the like showing the process and the result of the processing are displayed on the right side of the screen.

Here, the procedure for estimating the demand amount by the present system will be described in detail with reference to the flowchart of FIG. In the present embodiment, actual demand values are recorded for six months from April to September, and the demand amounts for three months from October to December are predicted based on these actual values. The case will be described as an example.

First, the person in charge inputs demand environment condition information using the input device 2 as demand environment condition information input means as a pre-stage of the process of estimating the demand (step S1).
1). In the demand environment condition information database 5b, FIG.
As shown in (a), the monthly value of the demand environment condition information that affects the increase and decrease of the demand amount is calculated from the past period in which the results were recorded to the future prediction period (that is, from April to December). A data table T1 for input is provided, and the demand environment condition information input in step S11 is written in a corresponding column of the data table T1.

Here, the demand environment condition information includes information relating to a stationary factor and information relating to a sudden factor. As the information relating to the stationary factor, a seasonality index indicating a seasonal variation in the demand, a variation in the demand according to a product life cycle. And a sales promotion / campaign index indicating fluctuations in demand due to sales promotions / campaigns, and as information on sudden factors, a sudden factor index indicating fluctuations in demand due to disasters or other sudden factors is used. Each index is used to extract base data from the actual value in the past period in which the actual performance was recorded, and used to predict future demand from the predicted value of the base data in the future forecast period Includes future schedule information.
As a seasonality index, the annual average value of demand is 100
The life cycle index is calculated by using the average value of the entire period from the start to the end of the sale of the product as 1
The relative value when 00 is used is used. In addition, as the sales promotion / campaign index, a value obtained by adding a positive or negative sign to the amount of change in the demand (estimated value) due to the influence of the sales promotion / campaign activity is used. A value obtained by adding a positive or negative sign to the fluctuation amount (estimated value) of the demand amount due to the influence of (burst factor) is used. still,
In the present embodiment, since the monthly demand is predicted, the seasonality index indicating the seasonal variation is used. However, when the demand is predicted on a daily basis, the demand based on the day of the week is used as a stationary parameter. In the case of forecasting demand in a specific region, events that affect demand only in that region (for example, local events) The locality factor information shown may be used.

When the input of the demand environment condition information is completed and the demand amount prediction processing is selected by the operation of the person in charge, the arithmetic unit 4 of the computer 1 displays the output device 3 on the screen of the display as shown in FIG. Interface screen I1
Is displayed, and input is made in the input window W1 and check boxes C1 and C2 displayed on the interface screen I1,
By selecting the selection buttons B1 to B6, a demand amount prediction process is performed. First, when the person in charge inputs the product number code of the target product to the product code input window W1 using the input device 2, the product category name to which the target product belongs is displayed in the display window W2. The arithmetic unit 4 also reads the actual value of the demand of the target product from the demand actual database 5a, and displays the screen I
In addition to the plotting on the graph G1 above, the past and future influence factor parameters of the target product are read from the demand environment condition information database 5b and displayed on the tables T2 and T3 on the screen I1.

When the target product is determined, the arithmetic unit 4 starts a prediction process. First, the base data extracting means 4a uses the demand environment condition input in step S11 based on the actual value registered in the actual demand database 5a. The change due to the influence of the information is removed, and the base data of the demand is extracted (step S12). FIGS. 5B to 5F are explanatory diagrams for explaining the procedure for extracting the base data step by step. FIG. 5B shows the actual value of the demand amount from April to September. As shown in FIG. 5C, the value of the sudden factor index is subtracted from the actual value, and the change due to the sudden factor is shown. After obtaining the data with the minutes removed, a predetermined correction value (for example, 2) is added to the value of the promotion / campaign index as shown in FIG.
Is subtracted to obtain data from which changes due to sales promotion / campaign factors have been removed. Here, when removing the change due to sales promotion / campaign factors, the correction value multiplied by the sales promotion / campaign index is a value obtained for each target product from the result of analyzing past performance values. By obtaining a correction value for each product based on the above, cleansing processing with high accuracy can be automatically executed, and the number of work steps can be reduced.

Next, the data obtained by the above-described processing is divided by a value obtained by dividing the life cycle index by 100 to remove a change due to a life cycle factor (see FIG. 5E). By dividing the seasonality index by a value obtained by dividing the seasonality index by 100, a change due to the seasonality index is removed, and base data of the demand is obtained (see FIG. 5F).

When the base data of the demand amount is obtained, the person in charge uses the prediction model selection means 4b to select a prediction model matching the change tendency of the base data from a plurality of prediction models registered in the prediction model database 5c. The selected prediction model is applied by the base prediction value extracting means 4c to the change tendency of the base data (step S13).

Here, a procedure for selecting a prediction model will be described with reference to FIGS. As the prediction model, for example, a plurality of statistical models created by using a statistical method such as an exponential smoothing method and a plurality of pattern models obtained from past performance values of a product are prepared. The person in charge can select a prediction model while watching I1. For example, when the person in charge checks the check box C1 or C2 on the interface screen I1 and selects either the statistical model or the pattern model, and then selects the pull-down key K1,
A selection support screen as shown in FIG. 6A is displayed. On this selection support screen, a table T4 showing a list of statistical models and a table T5 showing a list of pattern models are displayed up and down. When a desired model is selected and the operation button B9 is selected, a diagram is displayed. The shape of the selected model L is displayed as shown in FIG. When the person in charge selects the desired model on the screen of FIG. 6A and selects the operation button B8, the process of selecting the prediction model is completed, and the prediction selected in the display window W3 on the interface screen I1 is completed. The model name is displayed.

In the present embodiment, when displaying the list of prediction models, the prediction models are displayed in the new order of the selected date and time. However, the prediction models are displayed in the order of high frequency of selecting the prediction models. May be. In addition, a list of prediction models used so far for a product in the same product category as the target product may be displayed, or the trend of base data in a predetermined period (for example, the latest 6 months) may be similar. A list of prediction models selected for the product may be displayed. In addition, for a plurality of prediction models, an index indicating the freshness of the selected date and time, an index indicating the selected frequency, an index indicating the similarity of the trend tendency of the base data in the predetermined period to the target product, and the like are used. It is also possible to create a comprehensive evaluation index value from the index and to display the prediction model in the order of the evaluation index values.

When the operator selects the operation button B4 on the interface screen I1 after the selection process of the prediction model is completed, a fitting process for matching the prediction model selected in the above-described process to the base data by the arithmetic unit 4 is performed. Be executed. FIG. 7A shows the base data of the target product for each month, and FIG. 7B shows the selected prediction model L, and the offset amount of the prediction model as shown by L1 to L3 in FIG. Is adjusted, and the offset amount is adjusted so that the deviation between the selected prediction model and the base data is reduced. The offset amount is determined by the differences d1, d2 between the actual value of each month and the model pattern as shown in FIG.
d3 ... of the square sum ^{(d1 2 + d2 2 + d3} 2 + d4 2 +
The offset amount may be determined so that d5 ² ) is minimized. If the sum of squares of the differences d1, d2, d3,... Between the actual value of each month and the model pattern becomes larger than the preset upper limit as a result of the adjustment of the offset amount, the selected pattern is Since it is determined that they do not match the base data, the arithmetic unit 4 as a notifying unit performs an alarm display on the display unit as the output unit 3,
It prompts the person in charge to change the model pattern.

The base predicted value extracting means 4c matches the model pattern L with the base data as described above (see FIG. 9 (a)), displays the result on the graph G2 of the interface screen I1, and sets the prediction period. Then, a predicted value (base predicted value) of the base data is extracted from the locus of the model pattern L in (step S14).
In FIG. 9A, □ indicates base data, and △ indicates a predicted base value. FIG. 9B shows the base data after the extraction processing, and FIG. 9C shows the base predicted value.

When the base predicted value is obtained, the correction processing means 4
d adds a correction process to the base predicted value to obtain a predicted value of the demand amount. As shown in FIG. 10A, monthly demand environment condition information in a future forecast period is registered in a data table T1 in a demand environment condition information database 5b. A correction process is performed on the base predicted value obtained by the extracting means 4c taking into account the change due to the influence of the demand environment condition information, and a predicted value of the demand amount is obtained (step S15). FIG.
(F) is an explanatory view explaining step by step the procedure of obtaining the predicted value of the demand. FIG. 10B shows the base predicted value from October to December. After multiplying the base predicted value by a value obtained by dividing the seasonality index by 100, and taking into account the influence of the seasonality factor (FIG. 10) (C), and further multiply by a value obtained by dividing the life cycle index by 100 to obtain data in which the influence of the life cycle factor is added (FIG. 10 (d)).
reference). Next, the correction processing unit 4d adds a value obtained by multiplying the value of the sales promotion / campaign index by 2 to the data obtained by the above processing, and takes into account the change due to the sales promotion / campaign factor (FIG. 10 (e)). )), The value of the sudden cause index is added, and the predicted value of the demand amount in the future forecast period is obtained in consideration of the change due to the sudden cause (FIG. 10).
(F)). Then, the arithmetic unit 4 displays the obtained predicted value of the demand amount and the base predicted value on the graph G3 of the interface screen I1. FIG. 11 is an example of a screen for displaying the predicted value of the demand amount on the display device,
On this screen, the predicted value of each product in the future prediction period is displayed for each product number.

As described above, the base data extracting means 4a
Extracts the base data by removing the change due to the known demand environment condition information from the actual value of the demand, and base prediction value extracting means 4c applies a prediction model to the obtained base data to predict the base data. Value, and the correction processing unit 4d performs a correction process by taking into account the change due to the influence of the demand environment condition information on the predicted value of the base data, thereby obtaining the predicted value of the demand amount. Since forecast processing is performed taking into account the change due to the influence of demand environment condition information, errors due to the influence of demand environment condition information can be reduced, realizing a demand forecast system with improved forecast accuracy. it can. Even if the forecast value of the demand volume includes the change due to the unknown demand environment condition information, the change due to the unknown demand environment condition information is also included in the prediction model. And the predicted value of the base data is obtained, the change due to the unknown demand environment condition information can be canceled, and the error included in the predicted value can be further reduced.

(Embodiment 2) FIG. 12 is a block diagram of a system adopting the demand forecasting method of the present invention.

In the present embodiment, in the demand forecasting system of the first embodiment, a correction rule database 5d for storing correction rules created from empirical knowledge (know-how) of each person is stored in the storage device 5. are doing.
Then, the correction processing means 4d includes the base predicted value extracting means 4
Correction processing is performed on the base predicted value obtained by c using the demand environment condition information stored in the demand environment condition information database 5b and the correction rule stored in the correction rule database 5d. We want to estimate the amount.
Note that, except that the correction rule database 5d is provided, the configuration is the same as that of the first embodiment. Therefore, the same components are denoted by the same reference numerals, and description thereof will be omitted.

FIGS. 13 to 17 show a case where the correction process is performed by applying the correction rule obtained from the rule of thumb of the person in charge.
It is explanatory drawing at the time of predicting the demand amount of this month (X month).

For example, as shown in FIG. 13, when the sales policy of the target product (product number is AAA01) is the same as the previous month, and especially when there is no plan for promotion / campaign, the predicted value is output as a value of about 80%. Since the correction rule has been obtained, the correction processing unit 4d outputs a value of about 80% of the predicted value before applying the correction rule as a final predicted value (person-in-charge predicted value).

Further, as shown in FIG. 14, when the sales policy of the target product (product number code is BBB01) is switched from normal to sales promotion, a correction rule is obtained in which the predicted value of the demand is added by a certain amount. Therefore, the correction processing unit 4d adds, for example, 2 to the predicted value before the correction rule is applied.
The value obtained by adding 000 is output as a final predicted value.

As shown in FIG. 15, in a product belonging to a certain product category (product number series is CCC),
When the sales policy of the target product (product code is CCC02) is switched from the diffusion stage to the normal stage, a correction rule is obtained that the forecast value for the current month will be approximately equal to the previous month's forecast value for products of the same series issued earlier. Therefore, the correction processing unit 4d outputs the predicted value of the previous month of the same series of products (product code is CCC01) as the predicted value of the current month of the target product by using the correction rule.

As shown in FIG. 16, when the sales policy of both the target product (product number code is DDD01) and another product belonging to the same product category (product number code is DDD02) is switched from normal to sales promotion. Is a correction rule that the demand amount of the target product is about 80% of the demand amount of the product having the part number code DDD02 based on the balance between the sales of the two products. Is used to calculate the forecast value for the current month of the target product, which is approximately 80% of the predicted value of the product with the product code of DDD02.
Is output as the value of

As described above, in the present system, the correction rules obtained based on the experience of the person in charge are stored in the correction rule database 5d, and the correction processing means 4d performs the correction processing in consideration of the correction rules. In addition, the experience (know-how) of the person in charge can be reflected in the demand amount prediction processing, and more accurate prediction can be performed.

(Embodiment 3) FIG. 17 shows a block diagram of a system adopting the demand forecasting method of the present invention.

In the present embodiment, in the demand forecast system of the second embodiment described above, the correction data obtained by the correction processing means 4d performing the correction process on the base forecast value in the past predetermined period, and the demand demand actual database 5a And a predicted value / actual value comparison means 4e for comparing the actual value of the demand amount registered in the storage device and outputting the comparison result to the output device 3.
Since the configuration other than the predicted value / actual value comparison means 4e is the same as that of the demand amount prediction system of the first or second embodiment, the same components are denoted by the same reference numerals and description thereof will be omitted.

In this system, the base data extracting means 4
After extracting the base data by removing the change due to the influence factor parameter from the actual value of the demand amount, the base predicted value extracting means 4c sets the offset amount so that the selected prediction model matches the base data. The base prediction value is obtained by adjustment, and the correction processing unit 4d adds the change due to the influence factor parameter to obtain the predicted value of the demand. However, the correction processing unit 4d records the actual result. Even in the past period, correction data is obtained by adding correction processing to the base predicted value. The predicted value / actual value comparison unit 4e compares the correction data obtained by performing the correction process on the base predicted value in the past period with the actual value of the demand registered in the demand actual database 5a. , The comparison result is displayed on the display as the output device 3.

FIG. 18 shows an example of a screen for displaying the comparison result. In FIG. 18, A shows the predicted value obtained without applying the correction rule,
B in the figure indicates a predicted value obtained by applying the correction rule, and C in the figure indicates an actual value. The number of the correction rule used is displayed below the graph, and the correction process is performed in consideration of the correction rule during the period from February to April.

As described above, the predicted value / actual value comparison means 4e
Displays the result of comparing the correction data obtained from the base prediction value with the actual demand value in the past period on the display, so that the correction processing by the correction processing means 4d is appropriately performed from the display on the display. Can be confirmed, and the result can be reflected in demand environment condition information, a correction rule, and a prediction model.

[0051]

As described above, the invention of claim 1 is a demand forecasting method for predicting a future demand based on a past demand of a product, and the performance is recorded. Inputting the influencing factor parameters that affect the increase or decrease in demand over the past period from the past period to the future forecasting period, and removing the change due to the effect of the influencing factor parameter from the actual value of demand in the past period A process of extracting base data of the demand amount that is not affected by the factor parameter, and a process of applying a prediction model representing a temporal change tendency of the demand amount to the base data to obtain a predicted value of the base data in a future forecast period And a process of correcting the forecast value of the base data by taking into account the change due to the influence of the influence factor parameter to obtain a forecast value of the demand amount in a future forecast period. After removing the change due to the known influencing parameter from the actual demand value, extracting the base data, applying the prediction model to the obtained base data to obtain the predicted value of the base data In addition, the forecast value of the base data is corrected by taking into account the change due to the influence factor parameter to calculate the demand value forecast value. Since the prediction process is performed in consideration of the change due to the above, errors due to the influence factor parameters can be reduced, and the prediction accuracy can be improved. Even if the forecast value of the demand volume includes the change due to the unknown influence factor parameter, the change due to the unknown influence factor parameter is also included in the prediction model, so this prediction model is applied to the base data. Therefore, since the predicted value of the base data is obtained, there is an effect that the amount of change due to the unknown influencing factor parameter is canceled and the error included in the predicted value can be reduced.

A second aspect of the present invention is a demand forecasting system for predicting a future demand based on a past demand of a product. An input means for inputting an influencing factor parameter that influences an increase or decrease in demand over the forecast period, and an influence of the influencing factor parameter by removing a change due to the influencing factor parameter from the actual value of demand in the past period. Base data extraction means for extracting base data of a demand amount that does not exist, and base prediction value extraction for obtaining a prediction value of the base data in a future forecast period by applying a prediction model representing a temporal change tendency of the demand amount to the base data Means and a correction process taking into account the change due to the influence factor parameter to the predicted value of the base data, Correction processing means for obtaining a predicted value of the quantity, wherein the base data extracting means removes the change due to the known influence factor parameter from the actual value of the demand quantity, extracts the base data, The predictive value extracting means applies a predictive model to the obtained base data to obtain a predictive value of the base data, and the correction processing means corrects the predictive value of the base data by taking into account the change due to the influence factor parameter. By performing the processing, the forecast value of the demand amount is obtained, and the forecast processing is performed in consideration of the change due to the influence factor parameter with respect to the actual value and the forecast value. There is an effect that an error can be reduced and a demand amount prediction system with improved prediction accuracy can be realized. Even if the forecast value of the demand volume includes the change due to the unknown influence factor parameter, the change due to the unknown influence factor parameter is also included in the prediction model, so this prediction model is applied to the base data. Since the predicted value of the base data is obtained, there is an effect that a change in the unknown influence factor parameter is canceled, and a demand amount prediction system in which the error included in the predicted value is further reduced can be realized.

According to a third aspect of the present invention, in the second aspect of the invention, the processing steps by the base data extracting means, the prediction model selecting means, the base predicted value extracting means and the correction processing means are displayed, and the contents of the processing are given to the person in charge. It is characterized in that a confirmation means capable of making the confirmation is provided, and the prediction processing can be executed while confirming the processing contents by the confirmation means, and there is an effect that the usability of the system is improved.

According to a fourth aspect of the present invention, in the third aspect of the present invention, the influence factor parameter includes a parameter of a steady factor whose occurrence time can be predicted and a parameter of a sudden factor that occurs suddenly. The parameters are used to extract the past history information used to extract the base data from the actual value in the past period in which the results were recorded, and to predict the future demand from the predicted value of the base data in the future forecast period, respectively. It is characterized by including future schedule information to be used, and has an effect that the demand amount can be predicted in consideration of both the stationary factor and the sudden factor.

According to a fifth aspect of the present invention, in the invention of the fourth aspect, when predicting the daily demand, a day-of-the-week index indicating the influence of the change in the demand due to the day of the week is used as a stationary factor parameter. It is characterized in that the demand amount can be predicted on a daily basis in consideration of the change due to the day of the week.

According to a sixth aspect of the present invention, in the invention of the fourth aspect, when a future demand in a specific area is predicted, an area which indicates an event affecting the demand only in the area as an influencing factor parameter. It is characterized by using sexual factor information, and has an effect that the demand amount in the target area can be predicted in consideration of a change due to an event occurring only in the target area.

According to a seventh aspect of the present invention, in the second aspect of the present invention, the prediction model is at least one of a statistical model created by using a statistical method and a pattern model created from a change tendency of the actual value. Including a prediction model selecting means for allowing the operator to select a prediction model to be used from among a plurality of prediction models registered in advance,
Using the prediction model selection means, a prediction model suitable for the prediction target can be selected, and the prediction process can be performed.

According to an eighth aspect of the present invention, in the invention of the seventh aspect, the predictive model selecting means sorts the models selected so far for a plurality of types of designated products in the order of decreasing frequency of selection. It is characterized in that it is displayed as a selection candidate, and has the effect of reducing the work of selecting a prediction model.

According to a ninth aspect of the present invention, in the invention of the seventh aspect, the predictive model selecting means sorts the models selected so far for the plurality of types of designated products in the order of the selected date and time in the new order. It is characterized in that it is displayed as a selection candidate, and has the effect of reducing the work of selecting a prediction model.

In a tenth aspect based on the seventh aspect, the prediction model selecting means selects another product whose demand quantity change tendency in a past predetermined period is similar to the product to be predicted. The present invention is characterized in that a prediction model is displayed as a selection candidate, and has an effect that labor for selecting a prediction model can be reduced.

According to an eleventh aspect of the present invention, in the second aspect, the base predicted value extracting means shifts the prediction model so that an index indicating a degree of deviation between the prediction model and the base data is equal to or less than a predetermined value. In this case, the degree of divergence between the prediction model and the base data can be reduced.

According to a twelfth aspect of the present invention, in the invention of the eleventh aspect, when the index is larger than a predetermined upper limit, a notifying means for prompting a change of the selected prediction model is provided. This has the effect that the person in charge can be informed that the degree of deviation between the prediction model and the base data is large.

According to a thirteenth aspect of the present invention, in the second aspect of the present invention, there is provided a database for storing correction rules created from the experience of the person in charge, and the correction processing means includes the correction rules stored in the database and the influence factor parameter. The correction processing is performed by using the correction rule, and the correction rule created from the experience of the person in charge can be reflected in the correction processing by the correction processing means, and there is an effect that the prediction accuracy of the demand amount is improved.

According to a fourteenth aspect of the present invention, in the second aspect of the invention, the correction processing means performs a correction process by taking into account a change due to the influence factor parameter with respect to the base data and its predicted value. A correction means for comparing the correction data obtained by performing the correction processing on the base data in the past period in which the performance was recorded, and the performance value is provided, and from the comparison result of the comparison means, the prediction processing is performed with high accuracy. There is an effect that it can be confirmed whether or not the operation is being performed.

[Brief description of the drawings]

FIG. 1 is a block diagram of a demand amount prediction system according to a first embodiment.

FIG. 2 is a flowchart illustrating the operation of the above.

FIG. 3 is a flowchart illustrating a business flow of the above.

FIG. 4 is an explanatory diagram of an interface screen of the above.

FIGS. 5A to 5F are explanatory diagrams of a process of extracting base data according to the embodiment.

FIGS. 6A and 6B are explanatory diagrams of processing for selecting a prediction model according to the first embodiment.

FIGS. 7 (a) to 7 (c) are explanatory diagrams of a process of adjusting a prediction model according to the first embodiment to base data.

FIG. 8 is an explanatory diagram of a process of matching the prediction model with base data.

FIGS. 9A to 9C are explanatory diagrams of a process of extracting a base prediction value according to the embodiment.

FIGS. 10 (a) to (f) are explanatory diagrams of processing for correcting a base predicted value in the above and obtaining a predicted value of a demand amount.

FIG. 11 is an explanatory diagram of a screen displaying a predicted value of the demand amount according to the embodiment.

FIG. 12 is a block diagram of a demand forecasting system according to a second embodiment.

FIG. 13 is an explanatory diagram of a correction process using the above correction rule.

FIG. 14 is an explanatory diagram of a correction process using another correction rule of the above.

FIG. 15 is an explanatory diagram of a correction process using another correction rule according to the first embodiment.

FIG. 16 is an explanatory diagram of a correction process using still another correction rule according to the first embodiment.

FIG. 17 is a block diagram of a demand amount prediction system according to a third embodiment.

FIG. 18 is an explanatory diagram of a screen for comparing an actual value and a predicted value according to the first embodiment.

[Explanation of symbols]

1 computer 2 Input device 3 Output device 4 arithmetic unit 4a Base data extraction means 4b Prediction model selection means 4c Base predicted value extracting means 4d correction processing means 4e Actual value comparison means 5 Storage device 5a Actual demand database 5b Demand environment condition information database 5c Prediction model database 5d correction rule database

   ────────────────────────────────────────────────── ─── Continuation of front page    (72) Inventor Yoshiaki Morisawa             1048 Kadoma Kadoma, Kadoma City, Osaka Prefecture Matsushita Electric Works, Ltd.             In the formula company

Claims (14)

[Claims] 1. A demand forecasting method for predicting a future demand based on an actual value of a past demand, wherein the demand is increased or decreased from a past period in which the performance is recorded to a future forecast period. The process of inputting the influencing factor parameter that affects the, and removing the change due to the influencing factor parameter from the actual value of the demand in the past period, the base data of the demand that is not affected by the influencing factor parameter Extracting, and applying a prediction model representing a temporal change tendency of the demand amount to the base data to obtain a prediction value of the base data in a future prediction period; Performing a correction process in consideration of a change due to the influence of the parameter to obtain a predicted value of the demand amount in a future prediction period. Way measurement. 2. A demand forecasting system for predicting a future demand based on an actual value of a past demand, wherein the demand is forecasted from a past period in which the performance is recorded to a future forecast period. Input means for inputting an influencing factor parameter that influences, and a demand amount base data that is not affected by the influencing factor parameter by removing a change due to the influencing factor parameter from the actual value of the demand amount in the past period. Base data extracting means for extracting the base data, a prediction model representing a temporal change tendency of the demand amount is applied to the base data, and a base predicted value extracting means for obtaining a predicted value of the base data in a future prediction period; A correction process is performed on the predicted value taking into account the change due to the influence of the influence factor parameter to obtain a predicted value of the demand amount in a future prediction period. Demand forecasting system, characterized by comprising a correction processing unit that. 3. A checking means for displaying the processing steps by said base data extracting means, said base predicted value extracting means and said correction processing means and allowing a person in charge to check the processing contents. The demand forecasting system according to claim 2. 4. The influential factor parameter includes a parameter of a stationary factor whose occurrence time can be predicted and a parameter of an unexpected factor that occurs suddenly.
The past history information used to extract the base data from the actual value in the past period in which the results were recorded, and the future plan used to predict the future demand from the predicted value of the base data in the future forecast period, respectively 4. The demand forecasting system according to claim 3, comprising information. 5. The demand according to claim 4, wherein, when predicting the demand on a daily basis, a day-of-week index indicating the influence of the change in the demand on the day of the week is used as a parameter of the stationary factor. Quantity forecasting system. 6. A method for predicting a future demand in a specific area, wherein locality factor information indicating an event affecting the demand only in the area is used as the influence factor parameter. Item 4. The demand amount prediction system according to Item 4. 7. The prediction model includes at least one of a statistical model created by using a statistical method and a pattern model created from a change tendency of an actual value, and includes a plurality of prediction models registered in advance. 3. The demand amount prediction system according to claim 2, further comprising a prediction model selecting means for allowing a worker to select a prediction model to be used from the inside. 8. The predictive model selecting means includes a model selected so far for a plurality of types of designated products,
The demand forecasting system according to claim 7, wherein the demand is displayed as a selection candidate in the order in which the selection frequency is high. 9. The prediction model selection means according to claim 1, further comprising:
The demand forecasting system according to claim 7, wherein the selected date and time are displayed as selection candidates in a new order. 10. The prediction model selecting means displays, as a selection candidate, a prediction model selected for another product in which a change in demand in a past predetermined period is similar to the product to be predicted. The demand forecasting system according to claim 7, wherein 11. The base predictive value extracting means shifts a predictive model and applies it to base data such that an index indicating a degree of deviation between the predictive model and base data is equal to or less than a predetermined value. Item 2. The demand forecasting system according to Item 2. 12. The demand forecasting system according to claim 11, further comprising a notifying means for prompting a change of a selected prediction model when said index is larger than a predetermined upper limit. 13. A database for storing correction rules created from the experience of a person in charge, wherein the correction processing means performs correction processing using the correction rules stored in the database and the above-mentioned influence factor parameters. The demand forecasting system according to claim 2, wherein 14. The correction processing means according to claim 1, wherein said correction processing means performs correction processing on the base data and its predicted value by taking into account a change due to the influence of the influence factor parameter. 3. A demand forecasting system according to claim 2, further comprising a comparing means for comparing the correction data obtained by performing the correction process with the actual value.