KR20240054860A - Method and system for providing trade trend index for overseas export support of distribution products - Google Patents

Abstract

An embodiment of the present invention includes a data management module acquiring trade data, economic trend data, and customer preference data, a learning module generating a prediction model using the collected data, and a prediction module using the prediction model. A method of providing information is disclosed, including the step of predicting the amount of change in volume over a predetermined period and the step of the trend index calculation module calculating the trend index.

Description

Method and system for providing trade trend index to support overseas export of distributed products {METHOD AND SYSTEM FOR PROVIDING TRADE TREND INDEX FOR OVERSEAS EXPORT SUPPORT OF DISTRIBUTION PRODUCTS}

The present invention relates to a method and system for providing trade trend indices by country and item to support overseas exports of domestic distributors.

The existing organization providing information for export trade is the Korea Trade Statistics Promotion Agency. The Trade Statistics Institute provides the trade economy diffusion index and the import/export unit price terms of trade index.

However, this is limited to providing information on the overall trade economy and does not meet the market demand for product-specific trends.

In addition, the Korea Trade Statistics Promotion Agency provides import and export statistics, cargo statistics, FTA statistics, customized statistics for small and medium-sized businesses, and corporate trade activity statistics through the operation of a trade statistics information portal service. The Korea Trade Statistics Promotion Agency's statistical information uses export and import data from the Korea Customs Service to provide export and import amount, weight, and quantity information for each item based on HScode, making it difficult for domestic distributors who use other product classification systems to utilize it.

Meanwhile, the Korea Trade-Investment Promotion Agency provides trade investment statistics such as Korean export status, global export status, foreign direct investment statistics, and country-specific market information such as World Bank indicators, KOTRA indicators, and media news, as well as trade rankings and market share. It provides promising market information for each item based on statistical data such as:

In the case of distributed products that are sensitive to trends, customer preferences through online sales mall reviews, etc. affect sales volume, so it was necessary to provide a trend index that reflects this information.

The purpose of the present invention is to collect information related to export trade, such as trade statistics and customer preference information based on unstructured data, and learn the data to provide a trend index based on predicted sales volume by item country.

The present invention includes the steps of a data management module acquiring trade data, economic trend data, and customer preference data, a learning module generating a prediction model using the collected data, and a prediction module using the prediction model to determine a predetermined Discloses a method of providing information, including the step of predicting the amount of change in volume for a period and the step of the trend index calculation module calculating the trend index.

The present invention can easily obtain quantified numerical information on export trends by target country and item by collecting trade statistics and economic trend information.

In addition, by converting HScode information, which is the classification standard for trade products, into a Kan-code based system, which is a distribution product classification system, and providing information, it is possible to make it easier for domestic distribution businesses to obtain information for overseas export of operated products.

In addition, through the trend information visualization module, it is possible to explain not only the result information of the trend change but also the influencing factors that brought about the trend change, thereby providing more meaningful information to users.

Figure 1 shows an information providing device 100 according to an embodiment of the present invention.
Figure 2 shows an example of a data utilization set when creating a model for each country/product group according to an embodiment of the present invention.
Figure 3 is a flowchart showing a method for calculating a trade trend index according to an embodiment of the present invention.

The technology described below may be subject to various changes and may have various embodiments, and specific embodiments will be illustrated in the drawings and described in detail. However, this is not intended to limit the technology described below to specific embodiments, and should be understood to include all changes, equivalents, and substitutes included in the spirit and scope of the technology described below.

Throughout the specification, when a part is said to be "connected" to another part, this includes not only the case where it is "directly connected," but also the case where it is "electrically connected" with another element in between. .

Terms such as first, second, A, B, etc. may be used to describe various components, but the components are not limited by the terms, and are only used for the purpose of distinguishing one component from other components. It is used only as For example, a first component may be named a second component without departing from the scope of the technology described below, and similarly, the second component may also be named a first component. The term and/or includes any of a plurality of related stated items or a combination of a plurality of related stated items.

In terms used in this specification, singular expressions should be understood to include plural expressions, unless clearly interpreted differently from the context, and terms such as “including” refer to the described features, numbers, steps, operations, and components. , it means the existence of parts or a combination thereof, but should be understood as not excluding the possibility of the presence or addition of one or more other features, numbers, step operation components, parts, or combinations thereof.

Before providing a detailed description of the drawings, it would be clarified that the division of components in this specification is merely a division according to the main function each component is responsible for. That is, two or more components, which will be described below, may be combined into one component, or one component may be divided into two or more components for more detailed functions. In addition to the main functions it is responsible for, each of the components described below may additionally perform some or all of the functions handled by other components, and some of the main functions handled by each component may be performed by other components. Of course, it can also be carried out exclusively by .

In addition, when performing a method or operation method, each process that makes up the method may occur in a different order from the specified order unless a specific order is clearly stated in the context. That is, each process may occur in the same order as specified, may be performed substantially simultaneously, or may be performed in the opposite order.

Hereinafter, it will be described that the information providing device 100 performs an operation according to an embodiment of the present invention. The information provision device 100 is a device that consistently processes input data and performs necessary calculations according to a specific model or algorithm. For example, an information provision device may be implemented in the form of a PC, a server on a network, a smart device, or a chipset with an embedded design program.

Figure 1 shows an information providing device 100 according to an embodiment of the present invention.

1 shows a block diagram of an information provision device related to an embodiment of the present disclosure. The components of the information providing device 100 shown in FIG. 1 are exemplary. Only some of the components shown in FIG. 1 may constitute the information providing device 100, and additional component(s) other than the components shown in FIG. 1 may be included in the information providing device 100.

First, the information providing device 100 may include a processor 110, a memory 120, and a communication unit 130.

The communication unit 130 can transmit and receive data with external devices such as other electronic devices or servers using wired or wireless communication technology. For example, the communication unit 130 may transmit and receive sensor information, user input, learning models, and control signals with external devices.

The memory 120 may store data supporting various functions of the information providing device 100.

The processor 110 may determine at least one executable operation of the information providing device 100. Additionally, the processor may control the components of the information providing device 100 to perform the determined operation.

For example, the information providing device 100 may be a TV, projector, mobile phone, smartphone, desktop computer, laptop, digital broadcasting terminal, PDA (personal digital assistants), PMP (portable multimedia player), navigation, tablet PC, or wearable device. , it can be implemented as a fixed or movable device, such as a set-top box (STB), DMB receiver, radio, washing machine, refrigerator, desktop computer, digital signage, robot, vehicle, etc.

Referring to FIG. 1, the information providing device 100 may further include an operating unit 140 for providing a trade trend index, and the operating unit 140 includes a data management module 141, a data mart 142, and a learning module. (143), a prediction module (144), a trend index calculation module (145), and a trade trend information visualization module (146) may be included.

Additionally, external data may be included to provide the trade trend index of the information provision device.

At this time, external data may include a trade statistics database, economic trend database, and customer preference database.

Specifically, trade statistics collection data included in the trade statistics database may include information on the weight, export amount, and number of exports by country and item.

Economic trend data included in the economic trend database may include information such as GDP information by country, producer price index, consumer price index, and market capitalization by country industry.

In addition, customer preference information included in the customer preference database includes customer review information for each item in shopping malls in each country, trend information from overseas information search portals (e.g. Google, Bing, Safari, etc.), and domestic information search portals (e.g. Naver). , the following) trend information, etc. may be included.

The external data may be provided as public data by public institutions or specific database collection agencies, and the data may be obtained through linking methods such as API. Additionally, the acquired data is stored in the memory 120 of the information provision device or provided to the information provision device so that it can be processed.

Hereinafter, each configuration of the information provision device 100 will be described in detail.

The data linkage management module 141 of the operation unit 140 can acquire data from the external database described above through a linkage method such as API and extract information necessary for visualizing information and generating learning data.

At this time, in order to extract data for information visualization, the processor performs functions such as converting monthly statistical information from daily data to visualize monthly trend information, converting data sets for learning time series data, deleting missing data, or supplementing data by estimating approximate values. can be performed.

Additionally, the data management module 141 may periodically retrieve external data.

The data mart 142 according to an embodiment of the present invention can store data extracted through data conversion in the data management module 141. This can also be performed by memory 120.

The learning module 143 according to an embodiment of the present invention can extract information for each country and item stored in the data mart 142 and derive characteristics that affect the conditions of the country and item.

The method of deriving the above features derives the upper value of 'feature importance' of 'XGboost Regression' or selects the upper value of the absolute weight value in a multivariate linear regression model.

At this time, XGBoost (eXtreme Gradient Boosting) may be a machine learning library based on a boosting algorithm. Additionally, deriving the top value of "feature importance" in "XGBoost Regression" may mean identifying the most important features in the XGBoost model. Through this, you will be able to identify key characteristics that affect model predictions.

According to an embodiment of the present invention, in order to consider the time series impact of economic trend data, the learning module replaces the economic trend element values of a predetermined period (for example, 1 to 6 months ago) as features. You can add

The learning module 143 of the operation department can generate a prediction model for each combination of cases according to each condition and store a prediction model for each item that minimizes the mean square error or the average absolute error.

According to an embodiment of the present invention, the learning module 143 can utilize data suitable for country and item characteristics to select the optimal prediction model for each item.

For example, if the target country is the United States and the product group is clothing, the U.S. clothing industry GDP data can be used, and in the case of electrical and electronic products, the U.S. electricity sector GDP data can be used.

Additionally, the accuracy of the forecast model can be improved by utilizing economic trend data according to the selected country.

Figure 2 shows an example of a data utilization set when creating a model for each country/product group according to an embodiment of the present invention.

Referring to Figure 2, each data attribute records whether or not it is utilized by product group or country (e.g., 1 if used, 0 if not used) so that it can be used to utilize the data when creating a model for each country/product group. You can.

According to an embodiment of the present invention, a data set for learning is formed according to a combination of data attributes for each country/item, and the learning module 143 can select data characteristics to minimize prediction error.

The learning module 143 can select data characteristics to minimize prediction errors by recording combinations of data characteristics that minimize prediction errors for all possible combinations for each country/item and storing the corresponding learning model.

For example, when the number of data features in the learning module increases, it may take a lot of time to create the optimal learning model, so it is also possible to quickly extract a combination of data features using Feature Importance.

In addition, extraction of data feature combinations using Feature Importance creates a prediction model for all feature combinations for the target country and item, calculates the importance of the data features used to create the model, and selects and utilizes the features with high importance. You can.

The prediction module 144 can derive predicted values for volume change (or export amount change) according to the user's search conditions (country and item).

To do this, you can load a learning model that minimizes the forecast error by country/product and derive the forecast value for the next forecast period (for example, Month+1) from the model.

The prediction module 144 can store periodically calculated prediction results in a data mart and visualize the information in order to shorten the search time.

Additionally, in order to calculate the trend index, data on change compared to previous export volume (value) by country/product can be saved.

Another function of the prediction module 144 according to an embodiment of the present invention can predict customer preference for each country item from a customer preference learning model or predict the amount of trend change from a trend prediction model.

At this time, product ratings and review data from online shopping malls in that country can be used to predict customer preference for each country item.

Specifically, product numbers of online shopping malls for each country can be mapped to product group data used in the market information system, and product rating information for each product group can be stored in the data mart. Additionally, product preference can be calculated based on the number of positive and negative words in review data for each product group.

The trend prediction model according to an embodiment of the present invention can search for product keywords belonging to the corresponding country/product group on a country-specific trend search portal site (e.g., Google Trend Search, Naver Trend Search) and store them in the data mart.

In the prediction model, you can create a prediction model that uses data characteristics such as changes in search volume by country/product group and search volume, and store the predicted values in the data mart.

The trend index calculation module 145 according to an embodiment of the present invention calculates the trend index by assigning a weight vector to each forecast value based on the quantity change forecast, customer preference forecast, and trend change forecast value in the forecast module 144. can do.

For example, the trend index calculation module 145 sets the export volume, customer preference, and change forecast value to 100 at the time of data collection, and converts the change amount in the forecast period into a relative value based on 100 to generate the change value. You can. The trend index calculation module 145 can calculate the trend index by assigning weights to each element to the change value.

The trade trend information visualization module 146 according to an embodiment of the present invention can visualize monthly statistics related to trade statistics and economic trends organized in a data mart based on KANCODE, or visualize information on items with high transaction volume.

One of the main features of the trade trend information visualization module 146 is to provide contribution values of characteristic elements that influence trend changes.

The influence value of each characteristic element can be calculated by forming a combination of several characteristics and calculating the average change according to the presence or absence of the characteristic, similar to the SHAP value calculation method used in cooperative game theory.

Figure 3 is a flowchart showing a method for calculating a trade trend index according to an embodiment of the present invention.

Referring to FIG. 3, the information providing device 100 may collect external data through the data management module 141 of the operating unit 140.

The data management module 141 can convert monthly statistical data into monthly statistical data and store it in the data mart (S320).

The learning module 143 divides part of the collected data into training data and part into test data to generate a prediction model that minimizes the mean absolute error or mean square error (S330, S331).

The prediction model may be a univariate linear regression model, a multivariate linear regression model, or a neural network model with a multilayer perceptron.

The prediction module 144 uses the prediction model generated from the learning model to predict the quantity change in the next prediction period (1 month) (S340).

The trend index calculation module 145 can calculate the trend index (S350). As an example of trend index calculation, the initial start period of collected data can be converted to 100, the predicted value can be converted to match the conversion value, and then the value can be provided as a trend index.

To provide a trend index according to an embodiment of the present invention, the trend index can be calculated complexly by utilizing statistical changes such as the previously described forecast value, 3-month moving average of volume change, and 6-month moving average.

Specifically, the method of calculating the trend index, which includes changes in customer preferences and trends, is as follows.

The trend index calculation module 145 extracts customer preferences from external data that can indicate customer preferences by country and product group. Examples of external data representing customer preferences may be product reviews and product evaluation scores of online shopping malls in the relevant country.

If there is no numerically quantified product evaluation score, the trend index calculation module 145 can predict customer preferences, etc. through an ML model that predicts evaluation scores from customer evaluation data with both reviews and evaluation scores.

Additionally, the trend index including trend changes can be calculated based on changes in information search volume provided by information portal sites such as Naver Google. Based on the total search volume of the base year and month, the change in trends for each product by country is calculated through the change in search volume in each subsequent unit period.

In addition, the trend index calculation method, which includes customer preferences and trend changes, predicts the volume change using customer rating information based on the aforementioned product reviews, trend change information based on product search volume, industrial trends, and economic trend index as influential factors. You can create an ML model or regression model and perform predictions based on that model.

The trade trend information visualization module 146 according to an embodiment of the present invention can present representative characteristic elements that have a major influence on the prediction model for each country and item (S360, S370).

Specifically, the SHAP value can be used to calculate representative characteristic elements.

More specifically, the trade trend information visualization module 146 can derive the influence of variables using SHAP (Shapley Additive exPlanations) for characteristic factors derived through the XGBoost algorithm.

At this time, SHAP value may refer to a method of explaining the prediction results of machine learning by applying the principles of game theory.

Specifically, by sorting the SHAP value in descending order and providing a specific number of information at the top, it is possible to derive what characteristic factors are reflected in the information on promising products with a high rate of increase in volume change.

The trade trend information visualization module recommends promising products that include changes in customer preferences and trends by using customer rating information based on the aforementioned product reviews, trend change information based on product search volume, volume change forecasts, and 3-month/6-month changes. Based on descriptive statistics such as moving averages, the increase in each change can be sorted and ranked in descending order, and a comprehensive ranking can be calculated based on each ranking information, and the top information of the ranking can be provided (S360, S380).

Those skilled in the art will understand that the various illustrative logical blocks, modules, processors, means, circuits and algorithm steps described in connection with the embodiments disclosed herein may be used in electronic hardware, (for convenience) It will be understood that it may be implemented by various forms of program or design code (referred to herein as software) or a combination of both.

The present invention described above can be implemented as computer-readable code on a program-recorded medium. Computer-readable media includes all types of recording devices that store data that can be read by a computer system. Examples of computer-readable media include HDD (Hard Disk Drive), SSD (Solid State Disk), SDD (Silicon Disk Drive), ROM, RAM, CD-ROM, magnetic tape, floppy disk, optical data storage device, etc. There is.

Claims (1)

A data management module acquiring trade data, economic trend data, and customer preference data;
A learning module generating a prediction model using the collected data;
A prediction module predicting the quantity change in a predetermined period using the prediction model, and
Including a step where the trend index calculation module calculates a trend index,
How to provide information.

Images