JP2008021020A - Sales plan creation support system - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a sales plan creation support system for automatically and accurately selecting a more accurate demand prediction method. <P>SOLUTION: This support system is configured to store sales result data 100 as data configured of an OLAP database structure, and to review result data from various aspects, according to the combination of every hierarchy or sales period bucket. The resulting data, having a life cycle parameter which is the closest to a life cycle parameter corresponding to the result data, are retrieved, and the overall evaluation of a shipment rank, mean absolute error rate, cumulative error rate, and seasonal stability is carried out about the results of a plurality of demand prediction methods, preliminarily mounted based on the life cycle pattern and the result until proximity so that the optimal prediction result can be selected as a sales plan scheme. Accordingly, it is possible to select the proper prediction plan from among a plurality of demand prediction methods for each product, and even when the change of a demand environment is generated, it is possible to automatically correct the optimal prediction method, as necessary. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a sales plan creation support system for predicting demand for products based on past sales results, and more particularly to means for appropriately performing demand prediction for products having a short life cycle.

Conventionally, as a method of forecasting demand for products with a short life cycle, the parameters for each product are estimated by applying the sales volume results from the start of sales to the end of sales for each product to the growth model. The cumulative sales volume from the start of sales to that point is compared by comparing the actual sales volume from the start of sales of the products currently being sold to a certain point of time with the actual sales volumes of the various products from the start of sales to the same point in time. Select the product with the closest sales volume pattern and the estimated value for the product with the closest cumulative sales volume and the estimated value for the product with the closest sales volume pattern. There has been proposed a method for estimating a subsequent demand amount of a product being sold by estimating the parameter (see, for example, Patent Document 1).
Japanese Patent Laid-Open No. 10-228461

  According to the demand prediction method, it is possible to predict demand for a product having a short life cycle and insufficient sales performance.

  However, there are various methods for predicting demand for products being sold, not only applying sales volume results to growth models, but also applying statistical methods, for example, and product characteristics such as sales bases and product hierarchies. Since the optimum demand prediction method differs depending on the case, depending on the known technology, the optimum demand prediction cannot always be obtained.

  Even if various demand forecasting methods are used according to the product characteristics, it is necessary to rely on human judgment as to whether or not the optimum demand forecasting method according to the product characteristics is applied. A demand forecast cannot be obtained.

  The present invention has been made to solve the problems of the prior art, and the problem is to provide a sales plan support system capable of automatically and accurately selecting a more accurate demand forecasting method. There is to do.

  In order to solve the above-described problems, the present invention provides a means for periodically storing sales performance data from the start of sales to the end of sales for each product, and for each sales base level and each product level of each product, and the stored sales Means for calculating life cycle parameters corresponding to a combination of sales base hierarchy and product hierarchy from actual data, means for storing life cycle patterns for each product obtained based on the calculated life cycle parameters, and each accumulated product Multiple means using the means to search the life cycle pattern that is most similar to the life cycle pattern of the demand forecast target product from each life cycle pattern, and the sales result data up to the latest of the searched life cycle pattern and the demand forecast target product For the demand forecast calculation and multiple evaluation indicators shown in the demand forecast calculation results Means for setting the evaluation criterion limit value, means for selecting a demand forecast calculation result in which each evaluation index does not exceed the evaluation criterion limit value, and means for creating optimum demand forecast data based on the selected demand forecast calculation result And means for creating sales plan data based on the created optimum demand forecast data.

  Further, the present invention is characterized in that a sales policy registration function for assigning a sales policy such as a new product introduction plan or a sales suspension schedule is added to the means for creating the sales plan data.

  Further, according to the present invention, the sales base hierarchy includes a regional hierarchy in which a dealer is present and a sales form hierarchy in a direct sales system, a dealer system, an OEM sales, and the like, and the product hierarchy is a hierarchy for each category. It is a hierarchy to which each divided product belongs.

  Further, the present invention is characterized in that the means for accumulating the life cycle pattern for each product has an OLAP structure in which the database of accumulated data can perform cluster analysis, and the life cycle pattern can be recorded for each analysis unit. To do.

  According to the present invention, means for searching for a life cycle pattern that is most similar to the life cycle pattern of the demand prediction target product from the accumulated life cycle patterns for each product is as a life cycle parameter and is in the middle of selling from the start of sales. In addition, a parameter indicating a trend and a parameter of a cumulative performance for each time series bucket of sales performance data for a period until the end of sales are used as search keys.

  Further, according to the present invention, the means for performing a plurality of demand prediction calculations using the searched life cycle pattern and the sales result data up to the latest of the demand prediction target product is sold during the period from the start of sales to the middle of sales and the end of sales. The past sales performance data is simulated using the parameters indicating the trend and the cumulative performance parameters for each time-series bucket of the performance data, and the demand prediction calculation is performed by using a plurality of statistical prediction models using the sales performance data. And

  Further, according to the present invention, the means for performing a plurality of demand prediction calculations using the searched life cycle pattern and the sales performance data up to the latest of the demand prediction target product is a period from the start of sales to the middle of sales and the end of sales. Assuming that the parameters indicating the trend for each time series bucket of the sales performance data are p, q, the cumulative performance parameter is m, the cumulative performance amount is Nt, and the sales performance acquisition timing from the sales start to the latest is t The demand prediction calculation is performed by applying the growth model based on Formula 1 as a basic formula.

  Further, in the present invention, the means for setting the evaluation reference limit value sets the evaluation reference limit value for the sales volume rank, average absolute accuracy, cumulative accuracy, and seasonal stability within the specified period indicated in the demand forecast calculation result. It is characterized by setting.

  According to the present invention, means for setting evaluation reference limit values for a plurality of evaluation indexes indicated in the demand prediction calculation results, and means for selecting demand prediction calculation results for which each evaluation index does not exceed the evaluation reference limit values. Thus, it is possible to automatically create optimum demand forecast data for various demand forecast target products, and to automatically correct to an optimum forecast method as needed even when the demand environment changes.

  Hereinafter, an embodiment of a sales plan creation support system according to the present invention will be described with reference to FIGS. 1 to 7. The scope of the present invention is not limited to this embodiment, and appropriate modifications can be made without departing from the scope of the invention.

  FIG. 2 is a hardware configuration diagram of the sales plan creation support system according to the embodiment. As is apparent from this figure, the sales plan creation support system of this example includes the data input device 10, the storage device 20, and the data processing. The device 30 includes a man-machine interface device 40 and a data output device 50.

  As the data input device 10, a barcode reader, a POS terminal, a wireless tag reader, or a combination thereof can be used. Further, a database server is used as the storage device 20, and a general-purpose computer or a PC server is used as the data processing device 30. Further, a PC terminal or a Web terminal is used as the man-machine interface device 40, and a printer, a communication device, or the like is used as the data output device 50. As the system configuration, various application examples such as a C / S system and a Web system can be given.

  FIG. 1 is a function related diagram of demand prediction processing executed using the system configuration shown in FIG. 2, a function 100 for storing sales performance data of the entire product, and sales performance data of a demand prediction target product. A function 110, a function 101 for calculating life cycle parameters for each sales base hierarchy and each product hierarchy from the sales result data of the entire product, a function 111 for calculating life cycle parameters from the sales result data of the demand forecast target product, and the sales life of the entire product A function 102 for accumulating a cycle pattern, a function 112 for accumulating a sales life cycle pattern of a demand forecast target product, and a life cycle pattern similar to the sales life cycle pattern of a demand forecast target product among the sales life cycle patterns of the entire product Function 120 to search, life cycle pattern searched And a function 121 for calculating a demand forecast using a growth model based on the sales performance data of the demand forecast target product, and a function 122 for generating simulated sales result data based on the growth model demand forecast calculation result and the sales performance data of the demand forecast target product 122. , A function 123 for accumulating the generated simulated sales performance data, a function 130 for performing demand prediction calculation by a statistical method based on the accumulated simulated sales performance data and the sales performance data for the entire product, and other functions for performing demand prediction calculation 131, a function 140 for registering evaluation criterion limit values, a function 141 for storing evaluation criterion limit values, a function 142 for evaluating an optimum demand prediction method for a demand forecast target product based on the accumulated evaluation criterion limit values, A function 143 for accumulating optimal demand forecast data, and a function for registering a sales policy that complements the evaluation based on the limit value of the evaluation standard 44 has a function 160 to create the material procurement plan and production planning based on the function 150, the accumulated sales plan data accumulated sales planning data created based on the optimum forecast data.

  Note that the sales base hierarchy is the regional hierarchy where the store is located (for example, the upper hierarchy in Asia, Europe, the Americas, Japan, etc., the middle hierarchy such as by country or prefecture, the lower level in the country, the province in the prefecture, etc. Hierarchy) and sales type hierarchies by business partners (for example, higher levels such as direct sales, dealers, OEM sales, directly managed stores and agents for direct sales, lower levels such as sales companies and wholesalers for sales) The product hierarchy is divided into categories, and each product belongs to a hierarchy to which each product belongs (for example, in the case of television, there are various inch sizes according to methods such as cathode ray tube, plasma, liquid crystal, etc.) Cases divided into different products 1, 2, and 3).

  The function 100 for storing the sales performance data of the entire product and the function 110 for storing the sales performance data of the demand forecast target product have an OLAP database structure, and sales of each sales result data is started for each sales base hierarchy and each product hierarchy. It accumulates as data in units of buckets from the end to the end, and the performance data can be reviewed from various perspectives depending on the combination of each hierarchy and each bucket.

  As shown in FIG. 3, the function 102 for accumulating the sales life cycle pattern of the entire product and the function 112 for accumulating the sales life cycle pattern of the demand forecast target product have the life cycle patterns corresponding to the sales bases and the cuts by hierarchy as shown in FIG. Individual performance data is recorded in the form of a data table having parameters pi, qi, mi and ID numbers for identification, and these are recorded as life patterns.

The function 121 for performing the demand prediction calculation using the growth model performs the demand prediction of the target product using a prediction method based on the Bass growth model described in Patent Document 1, for example. In this prediction method, the life cycle parameters p, q, and m are used, and the demand prediction is performed by calculating the accumulated performance amount Nt from the following formula 1, using the sales performance acquisition time from the start of sales to the latest as t.

  In addition, the function 130 for calculating the demand forecast using a general statistical method includes the life cycle parameters pi, qi, and mi of the demand forecast target product for the most recent sales performance according to the processing flow shown in FIG. To search for the most similar life cycle parameter ID from the past, and statistical prediction using the past sales result data simulated by the life cycle parameter for the entire period from the sales start to the end using the formula 1 and the latest result data Complementing the insufficient past results and enabling the application of general statistical prediction models can expand the scope of application of the prediction method. In addition, by having a function 131 for extending other prediction methods, the present invention can be applied to future prediction methods devised in the future.

  The function 142 for evaluating the optimum demand prediction method compares the plurality of prediction calculation results obtained as described above with the latest performance data and evaluates the prediction results according to the processing flow of FIG. That is, one or more evaluation indexes such as sales volume rank (shipment volume rank), average accuracy, cumulative accuracy, and seasonal stability within the specified period illustrated in FIG. 6 are extracted from each prediction calculation result. It is comprehensively evaluated whether or not the index satisfies a certain standard registered by the function 140 for registering the limit value of the evaluation standard.

Here, the designated period indicates an arbitrary period such as the last past n months, and the sales volume rank indicates the ratio of the cumulative sales volume of the product to the cumulative sales volume of the entire product hierarchy to which the product belongs. So, it is a so-called ABC rank. In this example, the A rank is simply 1, the B rank is 0.7, and the C rank is 0.3. The average accuracy is obtained by calculating the average absolute error rate (average value of absolute error rates in bucket units) from the predicted value Fi and the actual value Xi in the specified period as shown in Equation 2, and expressing it as accuracy Sa.

Similarly, as shown in Equation 3, the total accuracy is obtained by calculating the absolute error rate for the total of the predicted value Fi and the total of the actual value Xi in the specified period as the accuracy Ra.

The seasonal stability Hs is an index that can be applied to a product that has actual data of two seasons or more and has a remarkable seasonal characteristic. In Equation 4, σ represents the standard deviation of the seasonal index, M represents the number of seasons, and j represents the number of σ for 12 months.

  For each of these evaluation indexes, the prediction results for each sales base and each product hierarchy are repeated in time series, and evaluation is performed so that the prediction result with the maximum average coefficient is regarded as the optimum value. If the sales start / end times of individual products are known in advance, information should be entered artificially. Based on these evaluation results and known information, the optimum prediction method is selected from the group prediction for each sales base hierarchy and the product hierarchy and the product single prediction result. In addition, about what performed group prediction, ratio distribution is carried out after prediction calculation, and it is set as the predicted value according to product. Even if the evaluation result is an optimum value, the result of the demand forecast calculation cannot be applied if it is below the minimum evaluation standard artificially set by the function 140 for registering the limit value of the evaluation standard. Then, the planned value from the outside is applied by the function 144 for registering the sales policy.

  Hereinafter, a method for predicting demand for a target product using the sales plan creation support system according to the embodiment will be described.

  The entire product sales record data 100 previously fetched from the input device 10 is stored in the OLAP structure database, and has information for each sales base hierarchy and each product hierarchy, and dicing and slicing by a cluster analysis method based on the information. Multi-dimensional data analysis such as drilling is possible. Accordingly, as shown in the life cycle pattern table 102 of FIG. 3, the overall product sales performance data 100 includes the product hierarchy 1, the product hierarchy 2, and the product hierarchy 3 for each sales base A. Sales base A, sales base B, sales base C by group, sales base A + sales base B total, sales base C + sales base D total, sales base E + sales base F total, sales performance data for each product level 1 In the case of summing up, life cycle parameters pi, qi, mi are calculated using the life cycle parameter calculation function 101 for each base for each tabulated data. Each of these parameters can be obtained by a method according to the method of Japanese Patent Laid-Open No. 10-228461 based on a basic growth model (see the above-mentioned Equation 1).

  The life cycle pattern table 102 is searched from the similar life cycle pattern search function 120 by assigning an ID number for keeping the uniqueness of the life cycle parameter to the aggregate unit of the sales result data thus obtained. The life cycle parameter value can be called later.

  On the other hand, for the demand forecast target product, the demand forecast target product sales performance data 110 is also a database having an OLAP structure similar to the overall product sales performance data 100, and the life cycle parameter calculation function 101 for each sales base hierarchy and each product hierarchy. The life cycle parameter (ID: mn ***) from the sales start time to the latest result acquisition period t1 is obtained using the same prediction target product life cycle parameter calculation function 111, and then stored in the life cycle pattern table 102 For the data from the stored life cycle parameters to the period of t1, the similar life cycle pattern search function 120 searches for the parameter most similar to the parameter of ID: mn ***. In the example shown in FIG. 3, the similar life cycle pattern ID: 12345 is the most similar pattern, and this data is a parameter when tabulated at the sales base A and the product hierarchy 3.

  Next, the growth model demand prediction calculation function 121 is used to calculate the demand prediction after t1 from the life cycle pattern ID: 12345 obtained from the search result and the sales performance data from the start of sales of the demand prediction target product to t1. This demand prediction calculation is based on the method disclosed in Japanese Patent Laid-Open No. 10-228461, and can be applied when the life cycle is short and the actual data is not sufficient.

  In this way, the demand forecast result obtained by using the growth model forecast function 121 can be applied as another statistical forecast model by the statistical technique demand forecast calculation function 130 by positioning the actual data as an extension. There is sex. In general statistical prediction models, demand forecast calculation is performed using actual data from one season to two or more seasons in consideration of seasonality. For example, when only 11 months have passed since the start of sales, or 12 months have passed. However, there are many cases in which statistical prediction cannot be applied because it is less than 12 months of one season, such as when the acquisition of performance data is delayed and there are only 11 months of results because the base is located overseas. Therefore, if it is possible to generate the performance data for one season or more by supplementing the insufficient performance data by using the prediction results of the growth model of the simulated sales performance data generation function 122, the application range of the general statistical prediction model can be increased. Can be spread. The demand prediction model based on a plurality of other prediction methods can be applied to a portion where the prediction accuracy cannot be increased only by the demand prediction based on the growth model based on Equation 1. In addition, when there is sufficient performance data for one season or more, it is not necessary to supplement in particular, and statistical prediction can be performed by directly using the data of the entire product sales performance 100. In addition, the other demand forecast calculation function 130 makes it possible to add a new forecast model to be developed in the future, and in combination with the growth forecast model demand forecast calculation function 121 and the statistical method demand forecast function 130, it is optimal among these forecast methods. Can be selected.

  The plurality of demand prediction results are evaluated by the optimum prediction method evaluation function 142, and the processing flow is shown in FIG.

  First, simulated sales performance data based on the life pattern ID: 12345 generated by the simulated sales performance data generation function 122 is acquired by processing 331 so that various statistical prediction methods can be applied, and thereafter, the growth model demand prediction calculation function 122 and statistics are acquired. A plurality of demand prediction methods 1 to n (processes 332 to 335) by the method demand prediction calculation function 130 and other demand prediction calculation functions are performed, and the sales volume rank, average accuracy, cumulative accuracy, For the seasonal stability, comprehensive evaluation is performed using a radar chart as shown in FIG. 6, and the best result is adopted.

  Since the demand prediction calculation is performed using sales performance data for each sales base and each product hierarchy, the optimal prediction result may be optimal for group prediction for each hierarchy or optimal for individual product prediction. Further, when the evaluation standard limit value 140 and the evaluation standard limit data 141 are registered in advance and fall below the limit value of the evaluation standard, it is considered that the demand forecast is not applicable, so the sales policy registration function 144 Instructing external inputs such as new product launch plans and sales suspension schedules.

  Depending on the result of such evaluation, in processing 350, processing 353, and processing 355, the state of group prediction by hierarchy, single product prediction, or non-evaluation criteria is indicated in the optimum demand prediction data 143 for each demand forecast target product. The data is recorded together with the application prediction algorithm i351 and the application prediction algorithm j354, and is used for comparative evaluation with the prediction results from the next time. When the group prediction by hierarchy is adopted, since the prediction value is the total amount of the same hierarchy product at the same sales base, it is automatically distributed to the individual product prediction value by the ratio distribution value defined in advance by the processing 352. The FIG. 7 is a diagram showing an example of automatic allocation of predicted values. The demand forecast of a certain product is changed into a demand forecast value at a directly managed store, a demand forecast value at an agency, a demand forecast value at a sales company, and a demand forecast value at an OEM. An example of automatic distribution is shown.

  The sales plan data 150 created by the optimum demand forecasting method and sales policy registration becomes sales plan data for the material procurement plan / production plan 160, and every time new sales result data is acquired as shown in the flow of FIG. By executing the series of processes as needed, sales plan data is always provided by an optimal prediction method.

It is a function related figure of the sales plan creation support system concerning an embodiment. It is a hardware block diagram of the sales plan creation assistance system which concerns on embodiment. It is a related figure of the life cycle pattern table which concerns on embodiment, a demand forecast object life cycle pattern, and the simulation sales performance data generation function. It is a processing flow figure of the sales plan creation support system concerning an embodiment. It is a flowchart which shows the procedure of the evaluation process based on the some demand prediction method calculation process and evaluation index which concern on embodiment. It is explanatory drawing which shows an example of the evaluation parameter | index which concerns on embodiment. It is a graph which shows the automatic distribution of the demand forecast which concerns on embodiment.

Explanation of symbols

100 Product-wide sales performance data 101 Life cycle parameter calculation function for each product hierarchy by site 102 Product-wide life cycle pattern table 110 Demand forecast target sales result data 111 Forecast target product life cycle parameter calculation function 112 Forecast target product life cycle pattern table 120 Similar Life cycle pattern search function 121 Growth model demand forecast calculation function 122 Simulated sales performance data generation function 123 Simulated sales performance data 130 Statistical method Demand forecast calculation function 131 Other demand forecast calculation function 140 Sales policy registration function 141 Sales plan data 142 Optimal forecast method Evaluation function 143 Optimal demand forecast data 160 Material procurement planning and production planning system

Claims (8)

A means for storing sales performance data from the start of sales to the end of sales for each product on a regular basis and for each sales base level and each product level,
Means for calculating life cycle parameters corresponding to the combination of the sales base hierarchy and the product hierarchy from the accumulated sales performance data;
Means for accumulating a life cycle pattern for each product determined based on the calculated life cycle parameters;
Means for searching for a life cycle pattern most similar to the life cycle pattern of the demand forecast target product from the accumulated life cycle patterns of each product;
A means for performing multiple demand prediction calculations using the searched life cycle pattern and the sales performance data up to the latest of the target product for demand prediction;
Means for setting evaluation standard limit values for a plurality of evaluation indexes shown in the demand forecast calculation results;
Means for selecting a demand forecast calculation result in which each evaluation index does not exceed the evaluation standard limit value;
Means for creating optimal demand forecast data based on the selected demand forecast calculation results;
A sales plan creation support system comprising means for creating sales plan data based on the created optimum demand forecast data.   2. The sales plan creation support system according to claim 1, wherein a sales policy registration function for giving a sales policy such as a new product introduction plan and a sales stoppage schedule is added to the means for creating the sales plan data.   The sales base hierarchy is composed of a regional hierarchy in which dealers exist and sales form hierarchies in business partners such as direct sales, dealers, and OEM sales, and the product hierarchy is divided into categories for each product. 2. The sales plan creation support system according to claim 1, wherein the sales plan creation support system is a hierarchy to which it belongs.   The means for accumulating a life cycle pattern for each product has an OLAP structure in which a database of accumulated data can be subjected to cluster analysis, and a life cycle pattern can be recorded for each analysis unit. Sales plan creation support system.   The means for searching the life cycle pattern most similar to the life cycle pattern of the demand prediction target product among the accumulated life cycle patterns for each product is a period from the start of sales to the middle of sales and the end of sales as a life cycle parameter. The sales plan creation support system according to claim 1, wherein a parameter indicating a trend and a parameter of a cumulative performance for each time series bucket of sales performance data are used as search keys.   The means for performing a plurality of demand prediction calculations using the searched life cycle pattern and the sales performance data up to the latest of the demand forecast target product is a time series bucket of sales performance data for the period from the start of sales to the middle of sales and the end of sales. Using the sales performance data to simulate the shortage of performance data when the data required for a specific demand forecasting method cannot be obtained using the trend parameter and cumulative performance parameter each time. The sales plan creation support system according to claim 1, wherein demand forecast calculation is performed using a plurality of statistical prediction models. The means for performing a plurality of demand prediction calculations using the searched life cycle pattern and the sales performance data up to the latest of the demand forecast target product is a time series of sales performance data for the period from the start of sales to the middle of sales and the end of sales. When the parameters indicating the trend for each bucket are p, q, the cumulative performance parameter is m, the cumulative performance amount is Nt, and the sales performance acquisition time from the start of sales to the latest is t, 7. The sales plan creation support system according to claim 5, wherein the demand forecast calculation is performed by applying to the growth model.

  The means for setting the evaluation standard limit value sets the evaluation standard limit value for the sales volume rank, the average absolute accuracy, the cumulative accuracy, and the seasonal stability within the designated period indicated in the demand forecast calculation result. The sales plan creation support system according to claim 1.

Images