JP2000200260A - Commodity sales quantity prediction system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a commodity sales quantity prediction system which can predict how many unsold commodities will be sold more precisely. SOLUTION: This system has a customer classifying process part 4 which classifies customers by their features according to a purchase questionnaire data file 2 containing information generated in consideration of customer's purchase intentions and tastes such as interest, a vehicle classifying process part 10 which classifies vehicles from their features according to a vehicle specification element file 6 and a new vehicle specification element file 8 containing specifications and elements of existent vehicles which are being sold and new vehicles to be sold, a purchase actualization rate predicting process part 16 which find the purchase actualization rate of vehicles in the vehicle classifications in the customer classifications and the predicted quantities of vehicles to be sold by the vehicle classifications, and a sales quantity predicting process part 24 which predict the quantities of vehicles to be sold according to the evaluations of the existent vehicles and new vehicles and the purchase actualization rate, thereby predicting the quantity of new vehicles to be sold in consideration of the customer's tastes.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a system for predicting the number of products sold, and more particularly to a system for predicting the sales of new products that have not yet been sold from the characteristics of the products actually sold, the sales results, and the attributes of the customers who have purchased the products. Regarding improvement.

[0002]

2. Description of the Related Art It is extremely important to more accurately predict the expected sales volume of a newly planned product to be sold in the next period in determining a production plan, a sales strategy, and the like. As a method for estimating the number of new products sold, for example, there is a sales prediction method disclosed in Japanese Patent Application Laid-Open No. Hei 9-120395. According to this conventional technique, since a new product that has not been sold yet has no sales performance, the sales number of the new product is predicted with reference to the sales performance of the product that is actually sold. At that time, the products are classified by items such as a product name, a model number, and a color, and the purchasers (customers) are classified by items such as gender, age, and annual income. Then, the sales amount based on each factor is predicted by substituting the characteristics of the new product as coefficients into the linear model formula with each classification item as a factor, and the combination of the factors (addition of the number of sales obtained for each factor) is used for the new product. Of sales forecasts.

[0003]

Conventionally, however, sales forecasts have been made using a linear model formula as a prediction model on the assumption that each factor is independent. It is hard to say that they are independent. Further, in the linear model, there is no concept of time, so that the predicted number of sales is the same at any time.
That is, it is difficult to say that the prediction value obtained by such a method is highly accurate in any case.

[0004] In addition, conventionally, since the respective classifications of commodities and customers have already been given, analysis can be performed only within the range of known items. That is, since the sales prediction is performed based on the known classification, the accuracy is limited. Certainly, it is possible to read customer preference trends to some extent based on age and occupation.However, age and occupation etc. are closely related to customer preference and lead to uniform purchase motivation. Is hard to imagine. In other words, accurate sales forecasting cannot be performed by a conventional method.

The present invention has been made to solve the above problems, and an object of the present invention is to provide a product sales number prediction system capable of more accurately predicting the sales number of a product which has not been sold yet. To provide.

[0006]

In order to achieve the above object, a system for predicting the number of sold goods according to the present invention comprises:
A customer who purchased a current product in a product sales volume forecasting system that predicts the number of new products that have not yet been sold based on the characteristics of each current product actually sold, sales results, and attributes of the customer who purchased the product. Customer classification processing means for classifying customers from their characteristics based on the attributes of each customer and information indicating the taste of each customer, and classifying products from their characteristics based on specifications of current products, specifications and specifications of new products, and specifications Product classification processing means, and the customer classification processing means classifies the probability of being purchased in the future for each product classification classified by the product classification processing means based on the sales results of current products as the purchase realization rate by the customer classification processing means. Purchase realization rate prediction processing means for calculating for each of the customer classifications that have been performed, and for predicting the number of sales for each future product classification based on the calculated purchase realization rate;
Sales number calculation processing means for calculating the number of new products sold in the future according to the predicted evaluation of the new product and the evaluation of each current product included in the product classification including the new product. .

The purchase realization rate prediction processing means calculates a past purchase realization rate for each product category and each customer classification based on the sales results of the current product, and uses time series analysis based on the purchase realization rate. It is characterized by finding the purchase realization rate in the future.

Further, the number-of-sale-calculation processing means calculates the number of sales of the product category in the future predicted by the purchase realization rate prediction processing means for each product in accordance with the evaluation of each product included in the product category. The number of new products sold in the future is calculated by the distribution.

[0009] Further, information indicating the preference of each customer who has purchased the current product belonging to the same product category as the new product is displayed as an index for evaluating the new product, and the new product that receives the evaluation of the new product is received. There is provided product evaluation information receiving means, and the sales quantity calculation processing means calculates the sales number of the new product based on the evaluation of the new product received by the new product evaluation information receiving means.

[0010]

Preferred embodiments of the present invention will be described below with reference to the drawings. In the present embodiment, a case will be described as an example in which the sales volume of a newly developed vehicle is predicted.

Embodiment 1 FIG. 1 is a configuration diagram showing a new vehicle sales volume prediction system which is an embodiment of a product sales volume prediction system according to the present invention. In this embodiment,
A customer classification processing unit 4 for classifying customers from their characteristics based on a purchase questionnaire data file 2 prepared in advance.
A vehicle classification processing unit 10 for classifying vehicles from their characteristics based on a vehicle specification data file 6 and a new vehicle specification data file 8 prepared in advance; and a customer classification file 12 which is an output of the customer classification processing unit 4. And a purchase realization rate prediction processing section 16 that calculates a purchase realization rate based on the vehicle classification file 14 output from the vehicle classification processing section 10 and obtains a predicted sales volume for each vehicle classification.
Is calculated based on the purchase realization rate prediction file 18 and the pre-test data file 20 prepared in advance.
2 and a sales number prediction processing unit 24 that outputs the data to the second unit.
The contents of each data file used in the present embodiment can be displayed on a display in response to an operation using a keyboard, a mouse, or the like. Note that the data configuration of each data file will be described together with the description of each process described below that uses the data file.

What is characteristic in the present embodiment is that merchandise (vehicles) in the market and customers who are consumers are aggregated based on their similarities, and the strength of preference between the aggregates is defined as a purchase realization rate. That is, the number of demands of new products (the number of units sold) is predicted using the purchase realization rate. In this way, by grouping in consideration of the customer's preference and the characteristics of the product, it is possible to perform a more accurate sales prediction in consideration of the customer's preference. Here, the purchase realization rate refers to a ratio of vehicles belonging to a certain vehicle classification in a certain customer classification. If the purchase realization rate is based on past performance,
It is calculated based on the actual number of units sold (actual value), and if it is for the future, it is a predicted value calculated based on the actual value.

Next, a process of estimating the number of new vehicles that have not yet been sold in the present embodiment will be described. In the present embodiment, the prediction process is a customer classification process,
Since it can be roughly classified into vehicle classification processing, purchase realization rate prediction processing, and sales volume prediction processing, the description will be made in this order. First, the customer classification process will be described. If this process is completed before the subsequent stage of the purchase realization rate prediction process,
This is a process that may be performed before or after the vehicle classification process.

FIG. 2 is a diagram showing an example of the data structure of the purchase questionnaire data file 2 in the present embodiment. The purchase questionnaire data file 2 stores monthly collection results of questionnaires conducted for customers who have purchased a new car in Japan. The questionnaire result is input using input means such as a keyboard (not shown). Note that a new vehicle here is a vehicle that has already been sold. The “purchaser” is a person who has purchased a new car, particularly a person who drives, and the “purchased car” is a vehicle purchased this time. "Purchase motivation" is 5 for each item such as style and acceleration performance
Each evaluation point was given a grade. The purchase motive may be a subjective evaluation of the purchased vehicle by the purchaser. Here, 5 is the best point. "Buyer Attributes"
Is customer information about the purchaser himself, and includes not only attribute information indicating objective characteristics such as gender and age but also information indicating subjective characteristics such as hobbies. The present embodiment is characterized in that such information taking into account the preference is also treated as customer information, and thus it is possible to consider the potential purchase consciousness of the purchaser.

Hereinafter, a customer classification process executed by the customer classification processing unit 4 using the purchase questionnaire data file 2 will be described with reference to a flowchart shown in FIG. In the customer classification process, customers are classified based on their characteristics.

First, the customer classification processing unit 4 performs principal component analysis based on the purchaser attributes stored in the purchase questionnaire data file 2 (step 101). Principal component analysis is an analysis method in which many variables are represented by a small number of indices (principal components) having a general meaning. In the present embodiment, this method generally used in statistical analysis is used. I do. Customers with close factors from the score obtained by the principal component analysis to the top two are grouped (step 102). This concept is illustrated in FIG. Then, for each of the groups divided in step 102, a classification symbol and a name that accurately represents the characteristics of each group are assigned as identification information of each group (step 10).
3). This assignment is performed using input means (not shown).
In the present embodiment, a code of alphanumeric characters starting with “C” is assigned as a classification symbol. As a name, for example, the group that seems to have a high usage rate for off-road and day leisure activities is "cross-country school", and the shopping rate for supermarkets and bicycles etc. Name groups that are likely to be expensive, such as "going out". When the customer can be classified into the abstract category by the above processing, the result is stored in the customer classification file 12 (step 1).
04). FIG. 5 is a diagram illustrating an example of the data configuration of the customer classification file 12 according to the present embodiment. As described above, in the customer classification process, the customers are classified according to their characteristics, so that It can be seen whether or not it has a characteristic tendency. Since customers who are to be included in the same customer class have similar tastes, it is considered that they often purchase similar vehicles, that is, vehicles included in the same vehicle class.

In the customer classification process, grouping and assignment of classification symbols and names are performed. Of these, designation of a customer constituting one group is automatically classified and grouped based on each score. 4 may be displayed, or a group may be designated by input means (not shown) by displaying the screen in FIG.
Further, the classification symbols may be automatically numbered. The same applies to the vehicle classification processing described next.

Next, the vehicle classification processing executed by the vehicle classification processing unit 10 will be described. In the vehicle classification process, vehicles are classified based on their characteristics. The vehicle specification data file 6 and the new vehicle specification data file 8 used at that time will be described.

FIG. 6 is a diagram showing an example of the data structure of the vehicle specification data file 6 in the present embodiment. The vehicle specification file 6 stores specifications and specifications of vehicles sold in the past. Specifications and specifications are full length, full width,
Consists of objective information such as height, engine displacement, and weight. FIG. 7 is a diagram showing a data configuration example of the new vehicle specification file 8 in the present embodiment. This new car specification file 8 stores the specifications and specifications of the vehicle for which the current forecast sales volume is to be obtained. The information to be stored may be the same as the specifications and specifications stored in the vehicle specification data file 6. Even if the new vehicle to be predicted has not actually been commercialized yet, the specifications and specifications may be determined or provisionally determined. The contents of each of the files 6 and 8 are input using input means such as a keyboard (not shown). Hereinafter, the vehicle classification process will be described with reference to the flowchart shown in FIG.

The details of the processing in the vehicle classification processing are the same as those in the customer classification processing described above. That is, first, the vehicle classification processing unit 10 performs a principal component analysis based on the specifications and specifications set in the files 6 and 8 (step 111). The principal component analysis here may use the same method as the above-described customer classification processing. Vehicles with close factors from the score obtained by the principal component analysis to the top two are grouped (step 112). This concept is illustrated in FIG.
Shown in Then, a classification symbol and a name that accurately represents the characteristics of each group are assigned to each of the groups divided in step 112 (step 113). This assignment is performed using input means (not shown). In the present embodiment, codes of alphanumeric characters starting with "M" are assigned as classification symbols. As the name, for example, a name such as “medium sedan” from the viewpoint of the vehicle body shape and “economy 2BOX” taking into account features such as cost performance are given. When the vehicle can be classified into the abstract category by the above processing, the result is stored in the vehicle classification file 14 (step 114). FIG.
0 is a diagram showing a data configuration example of the vehicle classification file 14 in the present embodiment. In the vehicle classification process, multi-dimensional features are extracted from attributes (specifications and specifications) possessed by the vehicle, so that each vehicle has objectively what features as shown in the contents of the vehicle classification file 14. You know if you are. Classification based on objective vehicle information,
When a customer selects (purchases or plans to purchase) a vehicle included in a certain vehicle category, a customer included in the customer class including the customer selects any vehicle included in the vehicle class including the vehicle It is thought that there are many. In particular, in the present embodiment, principal component analysis is performed based on the specifications and specifications of the new vehicle “new X vehicle” to be predicted, and is included in any of the vehicle classifications.

When the customers and vehicles are classified as described above, the purchase realization rate prediction processing unit 16 calculates the purchase realization rate based on the classification result. FIG. 11 is a diagram showing an example of the data configuration of the purchase realization rate prediction file 18 in the present embodiment. In this purchase realization rate prediction processing, each customer classification in the future (prediction date) finally specified. The purpose is to calculate the purchase realization rate for each vehicle category and the predicted sales volume for each vehicle category to generate the purchase realization rate prediction file 18 shown in FIG. The purchase realization rate prediction file 18 holds two-dimensional data (matrix table) of the customer classification and the vehicle classification for each month, and the sales volume of each classification item and the ratio of the sales volume in the customer classification. That is, the purchase realization rate is set. As the sales volume and the purchase realization rate, an actual value is set for the past, and the predicted sales volume and the predicted purchase realization rate calculated by calculation are set for the future. In the present embodiment, by predicting the purchase realization rate for each classification that can be expressed in two dimensions, customer needs that could not be obtained so far can be reflected in the classification, and a more accurate prediction can be made. It is characterized by realizing. This purchase realization rate prediction process will be described with reference to the flowchart shown in FIG.

The purchase realization rate prediction processing unit 16 first calculates the number of units sold and the purchase realization rate, which will be described later, only for the past month. First, in order to generate two-dimensional data, information (classification symbols and names) regarding all customer classifications is uniquely extracted from the customer classification file 12 and set as two-dimensional data rows (step 121). Similarly, all the classifications (classification symbols and names) are uniquely extracted from the vehicle classification file 14 and set as two-dimensional data columns (step 122). Next, each classification is searched from the customer classification file 12 and the vehicle classification file 14, and the number of units sold for each vehicle classification for each customer classification is totaled and set in the corresponding fields (step 123). For example, according to FIG.
In January 2000, outdoor groups like C and D (C1
The number of sold sports (M3) type vehicles such as car B to customer 0) is 980. Subsequently, the total value of the sales volume is calculated for each customer classification (step 12).
4). For example, according to FIG. 11, in January 1990, 15,000 vehicles were sold to a customer of the outdoor group (C10). When the total value for each customer class is obtained in this way, the purchase realization rate is calculated for each vehicle class for each customer class (step 125). For example, according to FIG. 11, in January 1990, the purchase realization rate of a sports (M3) type vehicle in an outdoor group (C10) is:
980 units / 15,000 units ≒ 7%.
The purchase realization rate can be said to be a value expressing the degree of preference for each customer classification. According to the example shown in FIG. 11, it is possible to obtain an analysis result that a customer of the outdoor group (C10) prefers to purchase a minivan (M9) type vehicle.

As described above, when the past purchase realization rate and the like are calculated based on the actual value, and the past matrix table is completed, the future purchase realization rate and the like are obtained based on the past actual value.

When the forecast target date is specified from the outside, the purchase realization rate prediction processing unit 16 generates a two-dimensional data in the same manner as in steps 121 and 122 to generate two-dimensional data of the specified date. Data rows and columns are set (steps 126 and 127). In the present embodiment, an example is given in which the number of units sold in January 2005 is predicted as the future date. In the present embodiment, the matrix table is generated based on the hypothesis that the classification of customers and vehicles will not change in the future. Then, the total value of the sales volume for each customer class and the purchase realization rate for each vehicle class for each customer class are calculated (step 128). This is calculated using a time series analysis that is a general technique. When the total value and the purchase realization rate are obtained, the number of vehicles for each vehicle class in each customer class is calculated (step 129). This can be calculated by total number x purchase realization rate. For example, the predicted sales volume of sports (M3) type vehicles in the outdoor group (C10) is 25,000 units × 7% = 175.
You can ask for zero. After calculating the predicted number of sales for each vehicle classification for each customer classification, the total number for each vehicle classification is calculated (step 130). For example, a minivan (M
If the number of vehicles of type 9) other than the customers of the outdoor group (C10) and the outing group (C13) is 0, the total number is 18020.

As described above, when the purchase realization rate prediction processing section 16 generates the purchase realization rate prediction file 18, finally, the sales quantity prediction processing section 24 acquires the unsold vehicle "new X car". Make a prediction of the shares that will be made. In this sales volume prediction processing, the purchase realization rate prediction file 18 and the pretest data file 2 shown in FIG.
The share is predicted from the attractiveness of the new car calculated using 0. The purpose is to finally obtain the predicted number of new vehicles by completing the predicted number file 22 shown in FIG. In the present embodiment, the matrix table for the past and each forecasted year and month is completed. Only the data required for the calculation, that is, the total sales volume for each customer category and the purchase realization rate of the product category that includes the forecasted vehicle are calculated, as long as it is possible to obtain the expected sales volume for each rate and vehicle category. You may make it.

FIG. 13 is a diagram showing an example of the data structure of the pretest data file 20 used in the sales volume prediction processing. Pretest data file 20
Is a file that stores data that has been evaluated in advance before launching a new car to be marketed. The evaluation items registered in the file 20 are the same as the purchase motivation of the purchase questionnaire so that the evaluation can be easily compared, and the evaluation method for each evaluation item is the same as the evaluation method of the purchase questionnaire. The performer of the pretest is usually a person who can evaluate a pre-sale new vehicle in the manufacturer.
This pretest data is input using input means such as a keyboard (not shown). Hereinafter, the sales volume prediction processing will be described with reference to the flowchart shown in FIG.

First, the sales volume prediction processing unit 24 calculates the degree of attractiveness for the corresponding vehicle (step 1).
31). The applicable vehicle refers to a new vehicle to be predicted (“new X vehicle” in this example) and vehicles included in the same vehicle classification as the new vehicle. According to the example shown in the present embodiment, the new type X vehicle is classified into the vehicle class M9 “minivan” as registered in the vehicle classification file 14, so that the C vehicle belonging to the same vehicle class is
The attractiveness of the car and D car is calculated together with the new X car. As the attractiveness, an average of the evaluation points in each vehicle is used. Therefore, in the new X model, the pretest data file 2
The average value of the N evaluation points for each item of 0 is the average value of all the evaluation points for each item of the purchase motive registered in the purchase questionnaire data file 2 for each of the vehicles C and D. It becomes attractiveness. In other words, attractiveness is
It is the evaluation itself, and the higher the score, the higher the evaluation in that item.

After the attractiveness of each item for each vehicle is calculated, the overall attractiveness of each vehicle is determined based on the attractiveness (step 132). In this method, a principal component analysis is performed using the attractiveness of each vehicle as an input, and a factor point for each vehicle is obtained.
For example, assuming that the factors shown in FIG. 16 are obtained by the principal component analysis, in the present embodiment, only the first factor that most contributes to the obtained result is used, and this first factor is The overall attractiveness of each vehicle. The overall attractiveness is the overall evaluation of the vehicle itself, and the higher the score, the higher the evaluation.

When the total attractiveness of each vehicle is calculated, a predicted share for each vehicle is obtained based on the total attractiveness (step 133). In the present embodiment, the ratio of the total attractiveness of each vehicle to the total total attractiveness is set as the predicted share. For example, based on the example shown in FIG. 14, the total of the total attractiveness of the vehicle class M9 to be predicted is 1.98.
Therefore, the predicted share of the new X model is 0.89 / 1.98
≒ 0.45 or about 45%. By calculating in the same way, the predicted share of car C is 0.50 / 1.
98 ≒ 0.25, the forecast share of D car is 0.29 / 1.
98 / 0.15. As is clear from this calculation method, the expected share is calculated on the assumption that the higher the overall attractiveness, the more likely it is to sell.

When the predicted share of each vehicle is obtained, the predicted sales volume in the target year and month is calculated for each vehicle based on the predicted share (step 134). Since the predicted sales volume of the vehicle class M9 to be predicted has been obtained in the purchase realization rate prediction process in the previous stage, the predicted sales volume may be distributed according to the predicted share. Therefore,
The predicted sales volume of the vehicle category M9 to be predicted is 1
As the number of new models is expected to be 1,020 units / month,
8020 × 0.45 = 8109 units. By the same calculation, the predicted sales volume of car C is 450
The predicted sales volume of five vehicles and the D car can be obtained as 2703 vehicles. The result calculated as described above is stored in the predicted number file 22.

In the present embodiment, from the viewpoint that vehicles with higher evaluations will be more likely to be sold, the predicted sales volume of the vehicle class including the new vehicle is determined, and the predicted sales volume is calculated for the vehicles included in the vehicle classification. By allocating according to the relative relationship of evaluation (overall attractiveness), it is possible to predict sales even for new vehicles that have not been sold yet.

As described above, according to this embodiment, vehicles and customers, including new vehicles that have not been sold yet, are grouped, and the characteristics (purchase realization rate) for each vehicle class are time-series for each customer class. In this way, the sales volume of new vehicles is predicted based on this characteristic. That is, in the present embodiment, the number of vehicles sold is predicted using the purchase realization rate expressing the strength of the preference between the sets, so that each item is compared with the conventional evaluation in which each item is independently evaluated. It is possible to more accurately predict the sales volume of new cars that have not yet been sold.

Embodiment 2 FIG . In the sales volume prediction process according to the first embodiment, principal component analysis is performed using the average (attractive level) of each evaluation point of each vehicle as an input, and the first factor obtained as a result is used as the overall attractiveness. The expected share of each vehicle was calculated based on the attractiveness. In the present embodiment, the principal component analysis is performed in the same manner as in the first embodiment, but the similarity to the new vehicle is obtained using not only the first factor but also the second factor obtained as a result, and the past sales are calculated. The feature is that the forecast share of each vehicle is calculated by referring to the actual results. In other words, it is possible to evaluate not only the degree of attractiveness of each vehicle but also the similarity between a vehicle already sold and a vehicle to be sold in the future and an evaluation in consideration of past sales performance, and more accuracy. Prediction of high market share.

The system configuration diagram in the present embodiment may be the same as that in FIG. 1, and only the contents of the sales volume prediction processing executed in the sales volume prediction processing section 24 and the contents of the predicted volume file 22 obtained as a result of this processing are shown. different.

In the following, the process of estimating the share of a new vehicle that has not been sold in the present embodiment will be described. The customer classification process, the vehicle classification process, the purchase realization rate prediction process, and the sales volume prediction process, other than the sales volume prediction process, that constitute this prediction process are the same as those in the first embodiment, and a description thereof will be omitted. Only the sales volume prediction processing which is a feature of the embodiment will be described with reference to the flowchart shown in FIG. The purpose of the sales volume prediction process is to generate a predicted volume file 22 shown in FIG. 18 for storing a predicted share value and the like, as in the first embodiment.

First, the sales volume prediction processing section 24 evaluates the evaluation points of the new vehicle to be predicted (the "new X vehicle") and the already sold vehicles (current vehicles) included in the same vehicle classification as this new vehicle. Each is calculated (step 231). This evaluation point may be calculated using the same calculation method as the degree of attraction in the first embodiment.
That is, the average value of the N evaluation points for each item of the pretest data file 20 for the new X vehicle is determined by the purchase of each of the current vehicles C and D which belong to the same vehicle classification as the new X vehicle. The average value of all the evaluation points for each item of the purchase motive registered in the questionnaire data file 2 is set as the evaluation point.

When the evaluation score of each item for each vehicle is calculated, the similarity with the new X vehicle is calculated based on the evaluation score (step 232). First, a principal component analysis is performed using each evaluation point as an input to obtain a factor point for each vehicle.
In the present embodiment, among the obtained factors, the first factor is stored in the predicted number file 22 as the X value in the two-dimensional coordinates, and the second factor is stored as the Y value. Then, the distance between the current vehicle and the new X vehicle is obtained. For example, as shown in FIG.
The coordinate value of the car is (0.7, 0.9), the coordinate value of the car C is (0.2, 0.7), and the coordinate value of the car D is (0.8, 0.
1), the distance between the C car and the new X car is (0.7-
The square root of (0.2) ² + (0.9−0.7) ² is about 0.54. The distance between the D car and the new X car is calculated in the same way, and is about 0.81. The calculation result is stored in the predicted number file 22. The smaller the value obtained by this calculation is, the shorter the distance is, the closer to the evaluation of the new X vehicle, and it can be said that it is similar to the new X vehicle. When calculating the distance (similarity) from the new X vehicle, the principal component analysis is used in the present embodiment, but another method such as a neural network may be used.

Next, the attractiveness ratio of the current vehicle to the new X vehicle is calculated (step 233). In the present embodiment, it is assumed that the attractiveness ratio is inversely proportional to the square of the similarity. According to the above example, the distance between the new car X and the car C is 0.54 and 0.81, respectively.
4: 0.81 ≒ 1: 1.5, so the attractiveness ratio is 1
/ 1 ² : 1 / 1.5 ² ≒ 1: 0.44 ≒ 2.2: 1. The calculation result is stored in the purchase realization rate prediction file 18. As is clear from this calculation method, the higher the similarity, the higher the attractiveness ratio. Here, the degree of attractiveness refers to the degree of attractiveness of the purchaser to the new X car. In other words, a purchaser of a vehicle with a high degree of attraction can be said to be strongly interested in a new X vehicle similar to the new X vehicle when the new X vehicle is newly introduced to the market. The degree of attractiveness of each vehicle can be determined based on the level of the evaluation points.

Next, the sales volume of the current vehicle in a certain month before the sales of the new X vehicle is read from the purchase realization rate prediction file 18,
It is stored in the predicted number file 22. In the example shown in FIG. 18, the actual sales volume of the car C is 1230, and the actual sales volume of the car D is 400. At this time, the share within the same vehicle category is further calculated and stored in the predicted number file 22 (step 234).

Here, if it is assumed that a new X car has been put on the market in the month, how many people who will purchase the current car of the same vehicle classification will transition to the new X car is calculated (step 235). . This is obtained as follows based on the attractiveness ratio. That is, for example, the actual sales volume of the C car is 1230 units, so that 1230 × (2.2 / (2.
2 + 1.0)) ≒ 845 units, and the actual sales volume of the D car is 400 units, so 400 x (1.0 / (2.2 + 1.
0)) It will be $ 125. That is, if the new X car was sold in the month, it is estimated that 845 of the 1230 people who purchased the C car would have purchased the new X car instead of the C car. Similarly, 4000 who purchased D car
It is speculated that 125 of the people would have purchased the new X car instead of the D car. As a result, if a new X car was put on the market in the month, 970 (= 845 + 1)
25) units are on sale, while the C car has 385 (= 1
230-845) units, D car 275 (= 400-12)
5) It is estimated that units were sold. When the share and the increase / decrease of the share are further obtained based on the calculated number of sold vehicles, the obtained value is stored in the predicted number file 22. From the calculation method using the attractiveness ratio in the present embodiment, when a new X vehicle is put on the market, the new X
It can be seen that this is based on the assumption that more customers will flow from vehicles similar to cars. The present embodiment is characterized in that a predicted share in the same vehicle classification is obtained by incorporating past sales results as described above.

When the predicted share of each vehicle in the same vehicle classification is determined as described above, the predicted sales volume in the target year and month is calculated for each vehicle based on the predicted share (step 236). This processing may be the same as in the first embodiment. That is, since the predicted sales volume of the vehicle classification M9 to be predicted has been obtained in the purchase realization rate prediction process in the preceding stage, the predicted sales volume may be distributed according to the predicted share. For example, if the predicted sales volume of the vehicle class M9 in the prediction target month is 10000 units / month, the new X
The projected sales volume of the car is 10,000 × 0.595 ≒ 595
One can be requested. By the same calculation, the predicted sales volume of car C is 2362, and the predicted sales volume of car D is 1
687 units can be obtained. The result calculated as described above is stored in the predicted number file 22.

As described above, according to the present embodiment, when estimating the share of vehicles that have not been sold yet, not only the evaluation points of each vehicle but also the number of vehicles sold in the past are taken into account. It is possible to obtain the predicted share supported by. This makes it possible to more accurately predict the number of new vehicles that have not been sold. Also,
When a new vehicle is introduced, the degree of influence on a current vehicle similar to the new vehicle can be estimated at the same time.

In each of the above embodiments, the case where the product sales number prediction system is applied to a system for predicting the sales volume of a new vehicle as a product has been described as an example. It goes without saying that it can be applied.

Embodiment 3 FIG . Sales volume prediction processing unit 24
Calculates the attractiveness of a vehicle (new vehicle) not yet sold by referring to the pretest data file 20, and predicts the share of the new vehicle and each vehicle belonging to the same vehicle classification as the new vehicle based on the attractiveness. . In the pretest data file 20 used for this prediction, the evaluation points for the new vehicle are registered. In each of the above embodiments,
The evaluation points are input and set based on the subjectivity of those who can evaluate the new car before sale in the manufacturer. When evaluating a new car in this way, it does not refer to any objective data such as market trends. Therefore, in the present embodiment,
The new vehicle can be evaluated with reference to the actual market evaluation of vehicles belonging to the same class as the new vehicle, thereby providing a support function that enables highly accurate evaluation of the new vehicle.

FIG. 19 is a configuration diagram showing a new vehicle sales volume prediction system according to the present embodiment. This embodiment has a configuration in which a pretest data generation processing unit 26 is added to the system shown in FIG. The pretest data generation processing unit 26 generates and displays an index for evaluating a new vehicle based on the vehicle classification file 14 and the purchase questionnaire data file 2, receives the evaluation of the input new product, and stores the index in the pretest data file 20. This is a new product evaluation information receiving means to be registered. Other configurations may be the same as those in the first embodiment. Since the prediction processing in the present embodiment may be the same as in each of the above embodiments, only the data registration processing to the pretest data file 20 which is a feature of the present embodiment will be described using the flowchart shown in FIG. I do. This processing is performed by the vehicle classification processing unit 10
Is executed before the sales number prediction processing by the sales number prediction processing unit 24.

First, the pretest data generation processing section 26
Extracts a vehicle having the same classification as the new vehicle "new X vehicle" from the vehicle classification file 14 (step 301), and searches the purchase questionnaire data file 2 for each of the extracted vehicles to obtain a score for each item (step 301). Step 302). Based on FIG. 10, C car and D belonging to the classification “M9” of the new X car
The purchase questionnaire data file 2 is searched for the car, and first, the score of each item of Mr. C who purchased the car C is obtained.
Similarly, the score of each item of Mr. D who purchased the car D is obtained. Normally, a plurality of people purchase each model, and the average is taken to obtain a score for each item.

When the evaluation points for each item are obtained for each vehicle, the obtained evaluation points are displayed on the screen as an index for inputting the evaluation points of the new vehicle (step 303). Is performed as follows.

First, the maximum value and the minimum value of the evaluation points are obtained for each item such as style and acceleration performance. Then, the distance between the maximum value and the minimum value is divided into n equal parts. Basically, it is desirable to equally divide the number of vehicles belonging to the same classification as the new vehicle extracted in step 301. Next, after assigning the vehicle closest to each equally divided value, this is displayed on the screen. This display example is shown in FIG. As shown in this figure, since the evaluations of other vehicles belonging to the same classification are displayed, the evaluator makes a relative and more objective evaluation of the new vehicle to be evaluated as compared with the evaluations of the other vehicles. be able to. The evaluator determines the evaluation point for the item by moving the pointer 28 up and down to the position of the determined evaluation point. In the example shown in FIG. 21, only the style is shown as an item.
Evaluation points are similarly input for other items.

The pretest data generation processing unit 26 generates pretest data by the evaluator by receiving evaluation points of all items indicated by the pointer 28 (step 304). Then, the pretest data is written into the pretest data file 20 (step 305).

According to the present embodiment, the evaluation of a vehicle belonging to the same class as a new vehicle to be evaluated in the actual market is obtained and its evaluation point is displayed. The evaluation of the new car can be made relatively and objective while taking the evaluation into consideration. In other words, it is possible to perform an objective evaluation on the new vehicle in accordance with the market trend as compared with the subjective evaluation by the evaluator in each of the above embodiments. The sales volume prediction processing unit 24 predicts the share of the new vehicle and each vehicle belonging to the same vehicle classification as the new vehicle based on the evaluation of the new vehicle obtained as described above, so that the accuracy of the predicted share is further improved. Can be done.

Note that step 30 in the present embodiment
In No. 3, evaluations of vehicles belonging to the same classification as a new vehicle based on past performance are displayed at equal intervals, and the pointer 28 is used to input an evaluation point of the new vehicle. However, the present embodiment is characterized in that the evaluation of a vehicle belonging to the same classification as a new vehicle is displayed as an index in order to enable an objective evaluation of the new vehicle in accordance with market trends. The present invention is not limited to the user interface as shown in FIG. For example, instead of displaying evaluations of vehicles belonging to the same classification as a new vehicle at equal intervals, the evaluation points themselves may be displayed. Further, the evaluation points may be displayed in a table format. Further, the evaluation point of the new car may be input by a numerical value instead of being indicated by the pointer 28.

[0052]

According to the present invention, merchandise and customers including merchandise not yet sold are grouped, and the number of merchandise sales is predicted using a purchase realization rate expressing the strength of preference between the groups. I did it. That is, since the preferences of individually different customers are taken into account, it is possible to more accurately predict future sales of new products. As a result, it can be linked to more realistic new product development, production planning, sales planning, and the like.

[Brief description of the drawings]

FIG. 1 is a configuration diagram showing a new vehicle sales volume prediction system which is an embodiment of a product sales volume prediction system according to the present invention.

FIG. 2 is a diagram illustrating a data configuration example of a purchase questionnaire data file 2 according to the first embodiment.

FIG. 3 is a flowchart showing a customer classification process according to the first embodiment.

FIG. 4 is a conceptual diagram used to describe customer classification using principal component analysis in the first embodiment.

FIG. 5 is a diagram showing a data configuration example of a customer classification file according to the first embodiment.

FIG. 6 is a diagram showing an example of a data configuration of a vehicle specification data file according to the first embodiment.

FIG. 7 is a diagram showing an example of a data configuration of a new vehicle specification file in the first embodiment.

FIG. 8 is a flowchart showing a vehicle classification process according to the first embodiment.

FIG. 9 is a conceptual diagram used to explain classification of vehicles using principal component analysis in the first embodiment.

FIG. 10 is a diagram showing a data configuration example of a vehicle classification file according to the first embodiment.

FIG. 11 is a diagram showing a data configuration example of a purchase realization rate prediction file according to the first embodiment.

FIG. 12 is a flowchart showing a purchase realization rate prediction process according to the first embodiment.

FIG. 13 is a diagram showing a data configuration example of a pretest data file according to the first embodiment.

FIG. 14 is a diagram illustrating an example of a data configuration of a predicted number file according to the first embodiment.

FIG. 15 is a flowchart showing a sales volume prediction process in the first embodiment.

FIG. 16 is a diagram showing an example of a factor obtained by principal component analysis performed in the sales volume prediction processing according to the first embodiment.

FIG. 17 is a flowchart illustrating a sales volume prediction process according to the second embodiment.

FIG. 18 is a diagram showing an example of a data configuration of a predicted number file according to the second embodiment.

FIG. 19 is a block diagram showing a new vehicle sales volume prediction system as another embodiment of the product sales volume prediction system according to the present invention.

FIG. 20 is a flowchart showing a process of registering data in a pretest data file according to the third embodiment.

FIG. 21 is a diagram showing an example of a method for inputting evaluation points of a new vehicle in the third embodiment.

[Explanation of symbols]

2 Purchase questionnaire data file, 4 customer classification processing unit, 6 vehicle specification data file, 8 new vehicle specification data file, 10 vehicle classification processing unit, 12 customer classification file, 14 vehicle classification file, 16 purchase realization rate prediction processing unit, 18 purchase realization rate prediction file, 20 pretest data file, 22 predicted number file, 24 sales number prediction processing unit, 26 pretest data generation processing unit, 28 pointer.

Continued on the front page. (72) Inventor, Motohisa Iizuka 1 Toyota Town, Toyota City, Aichi Prefecture Inside Toyota Motor Corporation (72) Inventor Yuji 1 Toyota Town, Toyota City, Aichi Prefecture Toyota Motor Corporation F-term (Reference) 5B049 BB11 BB16 CC05 DD01 EE01 EE12 FF03 FF04

Claims (4)

[Claims] 1. A product sales number forecasting system for predicting the number of new products that have not yet been sold based on the characteristics of each current product actually sold, the sales results, and the attributes of customers who have purchased the product. Customer classification processing means for classifying customers from their characteristics based on the attributes of the customers who purchased the products and the information indicating the preferences of each customer, and products based on the specifications of current products, specifications and specifications of new products, and specifications Product classification processing means for classifying based on the characteristics thereof, and the customer will be determined as the purchase realization rate by using the probability of being purchased in the future for each product classification classified by the product classification processing means based on the sales performance of the current product. A purchase realization rate prediction processing means for calculating for each customer classification classified by the classification processing means and for predicting the number of sales for each future product classification based on the calculated purchase realization rate; Sales number calculation processing means for calculating the number of new products sold in the future in accordance with the measured evaluation of the new product and the evaluation of each current product included in the product classification including the new product. Product sales volume forecasting system. 2. The purchase realization rate prediction processing means calculates a past purchase realization rate for each product category and each customer category based on the sales performance of the current product, and uses time series analysis based on the purchase realization rate. 2. The system according to claim 1, wherein a purchase realization rate in the future is obtained. 3. The number-of-sale-calculation processing means assigns the number of sales of the product category in the future predicted by the purchase realization rate prediction processing means to each product in accordance with the evaluation of each product included in the product category. 2. The system according to claim 1, wherein the number of new products sold in the future is calculated by distribution. 4. A new product for displaying information indicating the preference of each customer who has purchased a current product belonging to the same product category as a new product as an index for evaluating the new product, and for receiving an evaluation of the input new product. 2. The apparatus according to claim 1, further comprising a product evaluation information receiving unit, wherein the sales number calculation processing unit calculates a sales number of the new product based on the evaluation of the new product received by the new product evaluation information receiving unit. 3. Product sales forecast system.