Method and system for predicting product price by using dynamic setting change of plurality of indices

Abstract

According to one embodiment of the present disclosure, a method for predicting a product price by using a dynamic setting change of a plurality of indices comprises the steps of: collecting index data for a plurality of indices that affect a product price; performing, on the index data, pre-processing for model training; training a price prediction model by providing the pre-processed index data to the price prediction model; and setting a dynamic variable on the basis of a time delay and adjusting input data by reflecting the dynamic variable.

Description

Method and system for predicting product prices using dynamic setting changes of multiple indicators

[0001] The present disclosure relates to a product price prediction technology, and in particular, to a product price prediction method and system capable of predicting prices more accurately by dynamically changing the settings of multiple indicators.

[0002] Traditional methods for predicting product prices used a single indicator or multiple indicators with fixed settings to predict prices.

[0003] Conventional technologies have limitations, such as reduced prediction accuracy due to fixed settings for multiple indicators or an inability to reflect the importance of indicators over time. In particular, since the leading nature of indicators can change depending on macroscopic industrial shifts or global events (e.g., infectious diseases like COVID-19, international conflicts, etc.), failing to reflect these factors poses a problem where the reliability of predictions may be compromised.

[0004] A related prior art document is Korean Patent Publication No. 10-2022-0058806 (May 10, 2022).

[0005] One technical objective of the present disclosure is to provide a method and system for predicting product prices using dynamic setting changes of multiple indicators, which can provide more accurate product price predictions by reflecting the dynamic setting of multiple indicators.

[0006] Another technical objective of the present disclosure is to provide a method and system for predicting product prices using dynamic setting changes of multiple indicators, which can improve prediction performance by automatically reflecting the time difference between independent and dependent variables that change over time.

[0007] Another technical objective of the present disclosure is to provide a method and system for predicting product prices using dynamic setting changes of multiple indicators, which can analyze price fluctuations of various industries and raw materials in real time and provide a prediction system that considers complex economic factors.

[0008] Another technical objective of the present disclosure is to provide a method and system for predicting product prices using dynamic setting changes of multiple indicators, which can efficiently process data to prevent overfitting problems that may occur in large-scale data.

[0009] The technical problems of the present disclosure are not limited to the technical problems described above, and technical problems not mentioned will be clearly understood by those skilled in the art to which the present disclosure belongs from the present specification and the attached drawings.

[0010] A method for predicting the price of a product using a dynamic setting change of a plurality of indicators according to one embodiment of the present disclosure is performed in a product price prediction system and may include: a step of collecting indicator data for a plurality of indicators that affect the price of a product; a step of performing preliminary preprocessing for model training on the indicator data; a step of providing the preprocessed indicator data to a price prediction model to train the price prediction model; and a step of setting dynamic variables based on time delay and adjusting input data by reflecting the dynamic variables.

[0011] In one embodiment, the item price prediction method may further include the step of inputting adjusted input data into the price prediction model to predict the item price.

[0012] In one embodiment, the step of performing preliminary preprocessing for model training on the indicator data may include: a step of identifying and removing missing values ​​and noise from the indicator data; and a step of normalizing and processing the time series data of each variable of the indicator data.

[0013] In one embodiment, the step of providing the preprocessed indicator data to the price prediction model to train the price prediction model may include the step of training the price prediction model to learn a linear relationship between independent variables and dependent variables.

[0014] In one embodiment, the price prediction model may be a linear regression-based statistical learning model.

[0015] In one embodiment, the price prediction model performs training by having a specific ratio between the training data and the validation data, wherein the ratio of the training data may be set higher than the ratio of the validation data.

[0016] In one embodiment, the step of setting a dynamic variable based on the time delay and adjusting input data by reflecting the dynamic variable may include the step of setting the dynamic variable based on the time delay between the independent variable and the dependent variable.

[0017] In one embodiment, the item price prediction method may further include the step of resetting the importance of independent variables by comparing the predicted item price predicted by the price prediction model with the actual item price.

[0018] A product price prediction system using dynamic setting changes of multiple indicators according to one embodiment of the present disclosure includes: a price prediction device that collects indicator data for multiple indicators affecting product prices, performs preliminary preprocessing for model training on the indicator data, and then provides the preprocessed indicator data to a price prediction model to train the price prediction model; and a user terminal that checks product price prediction information through the price prediction device. The price prediction device may set dynamic variables based on time delays and adjust input data by reflecting the dynamic variables.

[0019] A non-transient computer-readable recording medium according to one embodiment of the present disclosure may have a computer program executed by a computer recorded thereon. The computer program is executed in a product price prediction system and may include the steps of: collecting indicator data for a plurality of indicators that affect the product price; performing preliminary preprocessing for model learning on the indicator data; providing the preprocessed indicator data to a price prediction model to train the price prediction model; and setting dynamic variables based on time delays and adjusting input data by reflecting the dynamic variables.

[0020] The technical solutions of the present disclosure are not limited to the technical solutions described above, and technical solutions not mentioned will be clearly understood by those skilled in the art to which the present disclosure pertains from the present specification and the attached drawings.

[0021] According to embodiments of the present disclosure, a more accurate prediction of product prices can be provided by reflecting the dynamic setting of multiple indicators.

[0022] According to embodiments of the present disclosure, prediction performance can be improved by automatically reflecting the time difference between independent and dependent variables that varies over time.

[0023] According to embodiments of the present disclosure, a prediction system can be provided that analyzes price fluctuations of various industries and raw materials in real time and considers complex economic factors.

[0024] According to the embodiments of the present disclosure, data can be processed efficiently to prevent overfitting problems that may occur in large-scale data.

[0025] The effects of the present disclosure are not limited to those described above, and unmentioned effects will become apparent to those skilled in the art from the present specification and the accompanying drawings.

[0026] FIG. 1 is a block diagram of a product price prediction system using dynamic setting changes of a plurality of indicators according to one embodiment of the present disclosure.

[0027] FIG. 2 is a flowchart of a method for predicting product prices using dynamic setting changes of multiple indicators according to one embodiment of the present disclosure.

[0028] FIG. 3 is a block diagram of a product price prediction device using dynamic setting changes of a plurality of indicators according to one embodiment of the present disclosure.

[0029] Figure 4 shows an example of indicators affecting HRC prices and graphs illustrating the trends of each indicator and HRC prices.

[0030] Figure 5 shows other examples of indicators affecting HRC prices and graphs illustrating the trends of each indicator and HRC prices.

[0031] Figure 6 is a graph exemplifying the influence of various indicators and events affecting the price of HRC.

[0032] Figure 7 is a graph illustrating the trends and long-term leading nature of major leading variables (slab prices, China transportation industry index).

[0033] Figure 8 is a graph illustrating the trends and long-term leading nature of major leading variables (China real estate construction start area, China wholesale and retail index).

[0034] Figure 9 is a graph to explain the trends and long-term leading nature of major leading variables (China transportation industry index, China construction industry index).

[0035] Figure 10 is a graph to explain the dynamic leading nature of a major leading variable (China Shanghai Composite Index 300). Figure 11 is a graph to explain the dynamic leading nature of a major leading variable (China Transportation Industry Index).

[0036] FIG. 12 is a graph showing the predicted and actual values ​​of the Chinese HRC price according to one embodiment of the present disclosure, and the main variables related to the prediction of the Chinese HRC price.

[0037] Specific structural or functional descriptions of embodiments according to the concept of the present disclosure are provided merely for the purpose of explaining embodiments according to the concept of the present disclosure, and embodiments according to the concept of the present disclosure may be implemented in various forms and are not limited to the embodiments described herein.

[0038] Embodiments according to the concept of the present disclosure may be subject to various modifications and may take various forms; therefore, embodiments are illustrated in the drawings and described in detail in this specification. However, this is not intended to limit the embodiments according to the concept of the present disclosure to specific disclosed forms, and includes modifications, equivalents, or substitutions that fall within the spirit and scope of the present disclosure.

[0039] Terms such as "first" or "second" may be used to describe various components, but said components shall not be limited by said terms. For the sole purpose of distinguishing one component from another, for example, without departing from the scope of rights according to the concept of the present disclosure, the first component may be named the second component, and similarly, the second component may be named the first component.

[0040] When it is stated that one component is "connected" or "connected" to another component, it should be understood that while it may be directly connected or connected to that other component, there may also be other components in between. Conversely, when it is stated that one component is "directly connected" or "directly connected" to another component, it should be understood that there are no other components in between. Expressions describing the relationships between components, such as "between," "exactly between," or "directly adjacent to," should be interpreted in the same way.

[0041] The terms used herein are used merely to describe specific embodiments and are not intended to limit the disclosure. Singular expressions include plural expressions unless the context clearly indicates otherwise. In this specification, terms such as “comprising” or “having” are intended to specify the existence of the described features, numbers, steps, actions, components, parts, or combinations thereof, and should be understood as not precluding the existence or addition of one or more other features, numbers, steps, actions, components, parts, or combinations thereof.

[0042] Unless otherwise defined, all terms used herein, including technical or scientific terms, have the same meaning as generally understood by those skilled in the art to which this disclosure pertains. Terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant technology, and should not be interpreted in an ideal or overly formal sense unless explicitly defined in this specification.

[0043] In this specification, the term "processor" may refer to hardware capable of performing functions and operations according to each name described in this specification, computer program code capable of performing specific functions and operations, or an electronic recording medium loaded with computer program code capable of performing specific functions and operations.

[0044] In other words, the term "processor" may mean hardware for carrying out the technical concept of the present disclosure, software for driving said hardware, a functional combination of said hardware and said software, and / or a structural combination of said hardware and said software.

[0045] Hereinafter, embodiments will be described in detail with reference to the attached drawings. However, the scope of the patent application is not limited or restricted by these embodiments. Identical reference numerals in each drawing denote identical components.

[0046] In the following description, a method and system for predicting product prices using dynamic setting changes of multiple indicators are described.

[0047] FIG. 1 is a block diagram of a product price prediction system using dynamic setting changes of a plurality of indicators according to one embodiment of the present disclosure.

[0048] Referring to FIG. 1, a product price prediction system using dynamic setting changes of multiple indicators (hereinafter abbreviated as ‘product price prediction system’) may include a product price prediction device (100) and a user terminal (200).

[0049] The item price prediction device (100) is a computing device comprising a processor (101) and memory (102). Although not illustrated, the item price prediction device (100) may further include a communication module for being communicationally connected to a user terminal (200) via wireless or wired communication. The communication module may include communication circuitry. The item price prediction device (100) may be implemented as a service server, but is not limited thereto.

[0050] The processor (101) may include, as an example, at least one of a microprocessor, a central processing unit, a processor core, a multi-core processor, a multiprocessor, an Application-Specific Integrated Circuit (ASIC), or a Field Programmable Gate Array (FPGA), but is not limited thereto.

[0051] Memory (102) can store instructions (or programs) that can be executed by the processor (101). Memory (102) may include volatile memory or non-volatile memory.

[0052] Volatile memory can be implemented as DRAM (dynamic random access memory), SRAM (static random access memory), T-RAM (thyristor RAM), Z-RAM (zero capacitor RAM), or TTRAM (Twin Transistor RAM).

[0053] Non-volatile memory can be implemented as EEPROM (Electrically Erasable Programmable Read-Only Memory), flash memory, MRAM (Magnetic RAM), Spin-Transfer Torque (STT)-MRAM, Conductive Bridging RAM (CBRAM), FeRAM (Ferroelectric RAM), PRAM (Phase change RAM), Resistive RAM (RRAM), Nanotube RRAM, Polymer RAM, Polymer RAM (PoRAM), Nano Floating Gate Memory (NFGM), holographic memory, Molecular Electronic Memory Device, or Insulator Resistance Change Memory.

[0054] The item price prediction device (100) can utilize data on various multiple indicators to predict the price of an item. Here, each of the multiple indicators may be an indicator of a different type, and each indicator may be an indicator related to the item.

[0055] Hereinafter, with reference to FIGS. 2 to 12, the case in which the item price prediction device (100) predicts the price of a hot-rolled coil (hereinafter referred to as 'HRC') will be explained as an example. However, the subject of prediction is not limited to this, nor is the subject of prediction limited to items. That is, it will be easy to understand from the following explanation that the item price prediction device (100) and the item prediction method performed by the item price prediction device (100) can predict the value of items, products, markets, or assets.

[0056] FIG. 2 is a flowchart of a method for predicting the price of an item using dynamic setting changes of a plurality of indicators according to one embodiment of the present disclosure (hereinafter abbreviated as ‘item price prediction method’).

[0057] The item price prediction method of Fig. 2 can be performed in an item price prediction device (100).

[0058] The item price prediction device (100) can collect indicator data for multiple indicators that affect the item price (S210).

[0059] For example, a product price prediction device (100) can collect multiple indicators that affect HRC prices. Examples of multiple indicators include downstream industry indicators (e.g., construction industry index, transportation industry index, real estate index), supply and demand information, raw material prices, macroeconomic indicators, and inventory levels. Each of the multiple indicators is used as an independent variable in the price prediction model described below.

[0060] The item price prediction device (100) can perform preprocessing for model training on the collected indicator data (S220).

[0061] The item price prediction device (100) can preprocess collected indicator data to make it suitable for model training. For example, the item price prediction device (100) can perform missing value processing, correlation analysis between variables, and processing to remove unimportant variables.

[0062] The item price prediction device (100) can provide preprocessed indicator data to the price prediction model to train the price prediction model (S230).

[0063] In one embodiment, the price prediction model may be a linear regression-based statistical learning model. As an example, the price prediction model may be a Partial Least Squares (PLS) model. A PLS model is a linear regression-based statistical learning model. Because a PLS model is advantageous for finding the correlation between an independent variable 'X' and a dependent variable 'Y', it is useful in datasets where the correlation between variables is high.

[0064] The PLS model is a model based on the linear relationship of data, allowing for analysis by assuming a linear relationship between independent and dependent variables. When there are many independent variables, the PLS model can reduce them to a small number of components (major axes) to perform calculations while preserving important information. Therefore, it can reduce the influence of noise or unnecessary variables during the computation process. The PLS model has the advantage of being able to be trained with a small amount of data and clearly explaining the relationship between independent and dependent variables.

[0065] The item price prediction device (100) can set dynamic variables based on the time delay between independent variables and dependent variables (S240).

[0066] Afterwards, the item price prediction device (100) can adjust the input data by reflecting dynamic variables (S250).

[0067] For example, a product price prediction device (100) can analyze the time difference (time lag) between an independent variable and a dependent variable to reflect the effect of a change in the independent variable at a specific point in time on the dependent variable. Here, the independent variable may be an indicator, and the dependent variable may be an HRC price. Examples of indicators include, as previously mentioned, the construction industry index, the transportation industry index, the real estate index, supply and demand information, raw material prices, macroeconomic indicators, and inventory levels.

[0068] This time lag is an important factor in reflecting the prediction results as a leading period for long-term leadingness, that is, as a time lag. In other words, to reflect how late a change in the independent variable is reflected in the dependent variable in the relationship between the independent variable and the dependent variable, the product price prediction device (100) can set a dynamic variable for the time lag. For example, if the independent variable, the transportation industry index, rises, the dependent variable, the HRC price, does not rise immediately but rises after a certain period of time; the product price prediction device (100) can identify this time lag and set a dynamic variable. The dynamic variable based on this time lag can be set differently for each independent variable. In other words, the dynamic variable based on the time lag can be set differently for each indicator.

[0069] Meanwhile, the time delay may vary depending on the time point. The method of analyzing the time delay by the product price prediction device (100) is described in more detail as follows. As previously explained, the product price prediction device (100) analyzes time series data to determine how many days later the effect of a change in the independent variable at a specific time point appears on the dependent variable. However, the time delay determined in this way may vary depending on the time point. For example, the time delay at a past time point may be 90 days, and the time delay at a recent time point may be 180 days. As such, since the time delay at a specific time point may vary depending on economic conditions and external factors such as COVID-19, the product price prediction device (100) can dynamically analyze the impact of a single indicator rather than analyzing it as a fixed time difference. In other words, dynamically analyzing the impact of a single indicator means that the value of the time delay for the single indicator changes.

[0070] The item price prediction device (100) can predict the item price by inputting adjusted input data into a price prediction model (S250).

[0071] In one embodiment, the item price prediction device (100) can reset the importance of independent variables by comparing the item price predicted by the price prediction model with the actual item price.

[0072] FIG. 3 is a block diagram of a product price prediction device using dynamic setting changes of a plurality of indicators according to one embodiment of the present disclosure.

[0073] Referring to FIG. 3, the item price prediction device (100) may include an indicator collection unit (110), a preprocessing unit (120), a learning unit (130), a price prediction model (140), a dynamic variable setting unit (150), and an input adjustment unit (160).

[0074] The indicator collection unit (110) can collect indicator data for multiple indicators for a target price to be predicted (e.g., HRC price). Examples of multiple indicators include economic indicators such as the construction industry index, the transportation industry index, the real estate index, and steel inventory. Each of the multiple indicators is used as an independent variable in the price prediction model.

[0075] In one embodiment, the indicator collection unit (110) may collect downstream industry indicators, supply and demand information, raw material prices, macroeconomic indicators, and inventory levels as indicators for the HRC price. Downstream industry refers to an industry closely related to steel demand, and downstream industry indicators may include China real estate, China transportation industry index, and China construction industry index. Supply and demand information is an indicator that quantitatively represents steel supply or steel demand, and may include the HRC plant operating rate and steel consumption. Raw material prices are an indicator representing the price of raw materials that have a fluctuation trend similar to the HRC price, and may include the price of billets and the price of slabs. Macroeconomic indicators are indicators representing the overall situation of the industry and economy, and may include the China wholesale and retail index, the Shanghai stock price index, and the steel PMI (Purchasing Managers' Index). Inventory levels are an indicator showing the sentiment of market participants, and may include the inventory levels of the relevant HRC plant and the inventory levels of the relevant exchange.

[0076] The preprocessing unit (120) can perform preprocessing on the collected indicator data. For example, the preprocessing unit (120) can remove noise from the collected indicator data, supplement missing values, and analyze the correlation between independent variables and dependent variables.

[0077] In one embodiment, the preprocessing unit (120) can analyze the main leading variables among the collected indicator data and perform noise removal processing to retain only the main leading variables. For example, the preprocessing unit (120) can identify and remove missing values ​​and noise from the indicator data and normalize the time series data of each variable in the indicator data.

[0078] The learning unit (130) can train the price prediction model (140). The learning unit (130) can provide training data to the price prediction model (140) and train the price prediction model (140) to learn the linear relationship between independent variables and dependent variables.

[0079] The price prediction model (140) can be trained by the learning unit (130). The trained price prediction model (140) can predict a dependent variable (e.g., HRC price) based on input data including independent variables. The input data of the price prediction model (140) can be input by the input adjustment unit (160).

[0080] In one embodiment, the price prediction model (140) may be a PLS model. The PLS model may use PLS components, previously selected key variables, and variable data that changes over time (i.e., using variables from past [t1, t2, t3 …] time points, etc.) as parameters.

[0081] Meanwhile, the present disclosure has the characteristic that the number of independent variables is limited. Therefore, in order to increase the accuracy of learning, the training data and the verification data are each set to have a specific ratio, and the training of the price prediction model (140) can be performed such that the ratio of the training data to the total data is set higher than the ratio of the verification data to the total data. For example, 70% of the total data can be used as training data. Then, the remaining 30% of the total data can be used as verification data to evaluate how well the price prediction model (140) predicts.

[0082] The dynamic variable setting unit (150) can set dynamic variables for the indicator, and the input adjustment unit (160) can adjust input data by reflecting the dynamic variables set by the dynamic variable setting unit (150). In this embodiment, the dynamic variable setting unit (150) and the input adjustment unit (160) are described separately, but according to other embodiments, the dynamic variable setting unit (150) and the input adjustment unit (160) may be implemented as a single component.

[0083] The dynamic variable setting unit (150) can analyze the time difference (time delay) between the independent variable and the dependent variable, and set the dynamic variable based on the analyzed time difference (time delay).

[0084] The dynamic variable setting unit (150) can set a dynamic variable for time delay that reflects how late a change in the independent variable is reflected in the dependent variable in the relationship between the independent variable and the dependent variable. The dynamic variable for time delay is not specific and can be dynamically changed depending on the time point of the input data. For example, at a time point one month ago, the independent variable X1 may have a dynamic variable of a time delay of 90 days with respect to the dependent variable. On the other hand, at the current time point, the independent variable 'X1' may have a dynamic variable of a time delay of 60 days with respect to the dependent variable 'Y'. In this way, the dynamic variable is a variable that is set differently depending on the time delay that a specific independent variable has with respect to the dependent variable. The dynamic variable can be dynamically set differently at each time point.

[0085] The input adjustment unit (160) can adjust the input data (i.e., independent variables) by reflecting a time delay according to the dynamic variables set by the dynamic variable setting unit (150). Specifically, the input adjustment unit (160) can set independent variables that have a time delay. Then, the input adjustment unit (160) can shift the data by the delayed time during which the independent variable affects the dependent variable by reflecting the dynamic variables at a specific point in time. For example, if the transportation industry index of January 2023 affects the HRC price in April 2023, which is 90 days later, the input adjustment unit (160) can adjust the input data to reflect the transportation industry index of January in the forecast for April 2023.

[0086] In this way, when the dependent variable is predicted by reflecting the time lag of the independent variable, the time when the independent variable actually exerts influence is reflected, so the price prediction model (140) can provide more accurate prediction results. That is, after the time when past indicators actually exert influence is identified, the data reflecting the time lag is input into the price prediction model (140). Therefore, the predicted value of the HRC price derived based on the input data has characteristics that match the actual market situation more closely than the predicted value of the HRC price derived based on data without reflecting the time lag. As a result, the inflection point prediction becomes accurate, and the volatility of the dependent variable is better reflected.

[0087] Hereinafter, with reference to FIGS. 4 to 12, an exemplary environment for predicting HRC prices in China in a method and system for predicting product prices using dynamic setting changes of multiple indicators according to the present disclosure will be described in more detail.

[0088] Figure 4 shows an example of indicators affecting the price of HRC and graphs illustrating the trends of each indicator and the price of HRC. Figure 5 shows another example of indicators affecting the price of HRC and graphs illustrating the trends of each indicator and the price of HRC.

[0089] In the graphs shown in Fig. 4 and Fig. 5, the red line (R) represents the trend of the value for the corresponding indicator (i.e., independent variable), and the blue line (B) represents the trend of the HRC price.

[0090] Figure 4 shows the 'China Shanghai Composite Stock Price Index 300' and the 'China CHZ Cargo Volume Index' as indicators, and Figure 5 shows the 'China Wholesale and Retail Index', the 'China Real Estate Price Index', and the 'CHZ Construction Index' as indicators.

[0091] As can be seen in Figures 4 and 5, it can be observed that downstream industries and HRC prices influence each other, and in particular, looking at the trends in the graphs, it can be confirmed that downstream industry-related indices lead HRC prices. That is, qualitative analysis of downstream industry-related indices reveals that rising points, falling points, upward trends, downward trends, and inflection points have characteristics that lead HRC.

[0092] However, it can be observed that the leading indicators fluctuate depending on the time period. For example, indices related to downstream industries led by about 90 days during the COVID-19 period, but in recent times, they have been characterized by leading by about 180 days. This may be due to factors such as the fact that during the COVID-19 period, the degree of leading between macroeconomic indicators and HRC prices was not significant due to high price volatility, whereas in recent times, monotonous price fluctuations have influenced the relatively slow speed of price reflection.

[0093] As such, it can be seen that more accurate price prediction is possible when leading indicators are used as independent variables.

[0094] Figure 6 is a graph exemplifying the influence of various indicators and events affecting the price of HRC.

[0095] Referring to Figure 6, various factors influencing HRC prices are exemplified. Upon examination, it can be seen that after HRC prices skyrocketed due to COVID-19, they recovered as distribution networks were restored following the easing of COVID-19 restrictions. Furthermore, it can be observed that HRC prices exhibited a continuous downward trend due to the economic deterioration of the steel industry as well as the industrial and economic downturn in China. Subsequently, after a short-term rise in HRC prices following the announcement of China's stimulus policies, prices reversed to a decline due to the crisis in foreign banks. Although a trend of HRC prices reversing to an upward trend appeared whenever China implemented repeated domestic economic stimulus policies, it can be seen that price reversals to a downward trend continued to occur due to external factors and the failure of economic recovery.

[0096] In such an environment, the input variables for HRC can be set as shown in Table 1 below.

[0097] Serial Number Category Input Variable Description 1. Downstream Industries China Real Estate Index Real estate, transportation, and construction are downstream industries where steel materials are primarily used. The downstream industry index is closely related to future steel demand. China Transportation Industry Index China Construction Industry Index Crude Steel Production Volume 2. Supply and Demand Information Factory Utilization Rate An indicator that allows for the quantitative verification of steel supply and demand. Supply and demand are major factors in the formation of future steel prices. Number of Production Lines Steel Consumption Volume 3. Raw Materials Hot-rolled / Wall-rolled / Cold-rolled Prices HRC prices and steel raw material prices show similar trends. Billlet Prices Slab Prices 4. Macro Indicators China Wholesale and Retail Index Provides insight into the overall market status of Chinese industries. The economic boom and bust situations of the industry are highly correlated with price increases or decreases. China Shanghai Stock Price Index China Steel PMI Index Steel Earnings 5. Inventory Volume Factory Inventory Volume Inventory volumes at the exchange and factories are variables that clearly reflect the sentiment of market participants. Exchange Inventory Volume

[0098] Subsequently, among the input variables shown in Table 1, key antecedent variables can be analyzed. That is, although various categories of data related to HRC price determination were collected, the analysis of the collected data revealed that downstream industry-related indices are the most important factor. This analysis can be performed by filtering and selecting only independent variables that have a certain degree of similarity to the dependent variable.

[0099] Indicators related to downstream industries can be established as the most important group of variables for long-term forecasting. This is because, for secondary processed raw materials such as HRC, supply and demand are crucial in determining prices, and these factors have a high correlation with the needs of downstream industries. Specifically, demand may be most closely linked to the needs of downstream industries. In the case of supply, steel mills may exhibit a lagging trend, following demand (a boom in downstream industries). Therefore, for example, if production volume increased during a downstream industry boom, it can be concluded that supply volume would not be increased based on predictions of market conditions.

[0100] Supply and demand indicators are metrics that can quantitatively represent supply and demand. If information related to supply has been primarily collected, supply and demand indicators may be evaluated as relatively less important.

[0101] When examining raw material-related indicators, raw material prices are conceptually important because HRC prices are determined by base materials (e.g., iron ore, scrap, sintered ore), heat sources (e.g., electricity rates, coal, coke), external factors (e.g., supply, war), and margins (e.g., labor costs, corporate profits). However, from a long-term forecasting perspective, raw material prices are characterized by low importance due to their low leading correlation with HRC prices. Therefore, raw material prices can be filtered out as noise.

[0102] Macroeconomic indicators are metrics that allow for an understanding of the economic conditions of each industry and country, such as the PMI index, the corporate confidence index, and the economic confidence index. However, since macroeconomic indicators do not exhibit a significant leading relationship with HRC prices from a long-term forecasting perspective, they may be filtered out at this point in time. Nevertheless, there may be additional macroeconomic indicators that do have a significant relationship with HRC prices.

[0103] Inventory levels can be identified through exchange inventory and factory inventory. However, inventory levels may be filtered out as no significant leading relationship with HRC prices is identified from a long-term forecasting perspective.

[0104] After the preprocessing process of selecting major leading variables as described above is performed, a process of identifying a significant leading relationship between the selected major leading variables and the HRC price can be performed. This process can be performed in the learning unit (130) or the dynamic variable setting unit (150) of FIG. 3.

[0105] Figures 7 to 9 are graphs illustrating the trends and long-term leading nature of major leading variables.

[0106] Figure 7 illustrates a graph showing the relationship between slab prices and Chinese HRC prices, and a graph showing the relationship between the Chinese transportation industry index and Chinese HRC prices. In the graphs shown in Figure 7, the blue dotted line represents the point in time of 'today', and the red dotted line represents the point in time of 'future'.

[0107] Referring to the graph in Figure 7 showing the relationship between slab prices and Chinese HRC prices, it can be seen that while slab prices have a similar trend to Chinese HRC prices, they do not exhibit long-term leading properties with respect to Chinese HRC prices. In other words, slab prices have a similar trend to Chinese HRC prices but show a trend that is almost synchronized with Chinese HRC prices in time, without long-term leading properties.

[0108] On the other hand, referring to the graph in FIG. 7 showing the relationship between the Chinese transportation industry index and the Chinese HRC price, it can be seen that the Chinese transportation industry index exhibits long-term leading characteristics for the Chinese HRC price for a certain period (e.g., about 3 months). Therefore, the goods price prediction device (100) can re-select an independent variable that exhibits such long-term leading characteristics among the major leading variables and determine the time lag associated with long-term leading characteristics.

[0109] Figure 8 illustrates a graph showing the relationship between the area of ​​real estate starts in China and the price of Chinese HRC, as well as a graph showing the relationship between the wholesale and retail index and the price of Chinese HRC. Referring to the graph in Figure 8 showing the relationship between the area of ​​real estate starts in China and the price of Chinese HRC, it can be seen that the area of ​​real estate starts in China has a long-term leading tendency for the price of Chinese HRC. Therefore, the area of ​​real estate starts in China can be selected as an independent variable exhibiting a long-term leading tendency.

[0110] Referring to the graph in Figure 8 showing the relationship between the Chinese wholesale and retail index and the Chinese HRC price, it can be seen that the Chinese wholesale and retail index has long-term leading properties for the Chinese HRC price. Therefore, the Chinese wholesale and retail index can be selected as an independent variable that exhibits long-term leading properties.

[0111] Figure 9 shows a graph showing the relationship between the Chinese transportation industry index and the Chinese HRC price, and a graph showing the relationship between the Chinese construction industry index and the Chinese HRC price.

[0112] Referring to the graph in Figure 9 showing the relationship between the Chinese transportation industry index and Chinese HRC prices, it can be seen that the Chinese transportation industry index has a long-term leading tendency for Chinese HRC prices. Therefore, the Chinese wholesale and retail index can be selected as an independent variable exhibiting a long-term leading tendency.

[0113] Referring to the graph in Figure 9 showing the relationship between the Chinese construction industry index and the Chinese HRC price, it can be seen that the Chinese construction industry index has long-term leading properties for the Chinese HRC price. Therefore, the Chinese construction industry index can be selected as an independent variable exhibiting long-term leading properties.

[0114] Meanwhile, the time lag for independent variables exhibiting long-term leading characteristics can be dynamically changed. For a more specific explanation of this, refer to Figures 10 and 11.

[0115] Figures 10 and 11 are graphs illustrating the dynamic leading nature of major leading variables.

[0116] The graphs shown in Fig. 10 are intended to explain the case where the Shanghai Composite Index 300 of China has a leading tendency for the price of China HRC, but the leading tendency (time lag) changes dynamically depending on the time point.

[0117] Referring to the graph in Figure 10, where the data for the Shanghai Composite Index 300 is shifted by a time lag of 90 days, it can be seen that the overall trends of the Shanghai Composite Index 300 and the Chinese HRC price coincide during the period from March 2021 to July 2022. However, it can be seen that the leading indicators do not match during the period after 2023.

[0118] On the other hand, if we refer to the graph in Figure 10 where the data for the Shanghai Composite Index 300 is shifted by a time lag of 180 days, it can be seen that the trends of the Shanghai Composite Index 300 and the Chinese HRC price coincide in the period after 2023.

[0119] In this way, the time delay associated with long-term leadingness can be set as a dynamic variable. In other words, the time delay associated with long-term leadingness can be dynamically set differently depending on the time point of the input data.

[0120] The graphs shown in Figure 11 are intended to explain the case where the Chinese transportation industry index has a leading nature with respect to the Chinese HRC price, but the leading nature (time lag) changes dynamically depending on the time point.

[0121] Referring to the graph in Figure 11, where the data for the China Transport Industry Index is shifted by a time lag of 90 days, it can be seen that the overall trends of the China Transport Industry Index and the China HRC price coincide during the period from 2021 to 2022. On the other hand, it can be seen that the leading indicators do not match during the period from 2023 onwards.

[0122] Referring to the graph in Figure 11 where the data for the Chinese transportation industry index is shifted by a time lag of 180 days, it can be seen that the trends of the Chinese transportation industry index and the Chinese HRC price are similar.

[0123] FIG. 12 is a graph showing the predicted and actual values ​​of the Chinese HRC price and the key variables related to the Chinese HRC price according to one embodiment of the present disclosure.

[0124] In the graph shown at the top of Fig. 12, orange represents the predicted value of the Chinese HRC price, and blue represents the actual value of the Chinese HRC price. The graph shown at the bottom of Fig. 12 illustrates the main variables used in the prediction.

[0125] Referring to Figure 12, although there is some noise, the predicted and actual values ​​of the Chinese HRC price show a nearly similar trend over the entire period from 2021 to 2025. Although there are sections with some errors, these errors are at an acceptable level considering that it is a 150-day forecast. Furthermore, it can be seen that the inflection points of the rise and fall are predicted without any lag.

[0126] As such, even though all major variables were used, it can be confirmed that the downstream influence is significant, as indicated by the data analysis results, and that variable importance varies considerably over time. In other words, the baseline order of variable importance does not change significantly and exhibits a monotonic change in influence; furthermore, it can be observed that the results are predicted by a combination of derived variables such as the transportation industry, Shanghai stock price index, and construction industry index.

[0127] The device described above may be implemented as a hardware component, a software component, and / or a combination of a hardware component and a software component. For example, the device and components described in the embodiments may be implemented using one or more general-purpose or special-purpose computers, such as, for example, a processor, a controller, an arithmetic logic unit (ALU), a digital signal processor, a microcomputer, a field programmable gate array (FPGA), a programmable logic unit (PLU), a microprocessor, or any other device capable of executing and responding to instructions. The processing unit may execute an operating system (OS) and one or more software applications executed on said operating system. Additionally, the processing unit may access, store, manipulate, process, and generate data in response to the execution of the software. For ease of understanding, the processing unit may be described as being used as a single unit, but those skilled in the art will understand that the processing unit may include a plurality of processing elements and / or a plurality of types of processing elements. For example, the processing unit may include multiple processors or one processor and one controller. Additionally, other processing configurations, such as parallel processors, are also possible.

[0128] Software may include computer programs, code, instructions, or a combination of one or more of these, and may configure a processing unit to operate as desired or command the processing unit independently or collectively. Software and / or data may be permanently or temporarily embodied in any type of machine, component, physical device, virtual equipment, computer storage medium or device, or transmitted signal wave so as to be interpreted by the processing unit or to provide instructions or data to the processing unit. Software may be distributed over networked computer systems and may be stored or executed in a distributed manner. Software and data may be stored on one or more computer-readable recording media.

[0129] The method according to the embodiment may be implemented in the form of program instructions that can be executed through various computer means and recorded on a computer-readable medium. The computer-readable medium may include program instructions, data files, data structures, etc., either alone or in combination. The program instructions recorded on the medium may be those specifically designed and configured for the embodiment, or may be those known and available to those skilled in the art of computer software. Examples of computer-readable recording media include magnetic media such as hard disks, floppy disks, and magnetic tapes; optical recording media such as CD-ROMs and DVDs; magneto-optical media such as floptical disks; and hardware devices specifically configured to store and execute program instructions, such as ROM, RAM, and flash memory. Examples of program instructions include machine code, such as that generated by a compiler, as well as high-level language code that can be executed by a computer using an interpreter, etc. The hardware device may be configured to operate as one or more software modules to perform the operation of the embodiment, and vice versa.

[0130] Although the embodiments have been described above with reference to limited examples and drawings, those skilled in the art can make various modifications and variations from the description above. For example, suitable results can be achieved even if the described techniques are performed in a different order than described, and / or the components of the described system, structure, device, circuit, etc. are combined or assembled in a form different from described, or replaced or substituted by other components or equivalents.

[0131] Therefore, other implementations, other embodiments, and equivalents to the claims also fall within the scope of the claims set forth below.

[0132] A method and system for predicting product prices using dynamic setting changes of multiple indicators as described above can be applied to the field of predicting product prices.

Claims

1. As a method for predicting product prices performed in a product price prediction system, A step of collecting indicator data for multiple indicators that affect the price of goods; A step of performing preliminary preprocessing for model training on the above indicator data; A step of providing preprocessed indicator data to a price prediction model to train the price prediction model; and A step including setting dynamic variables based on time delays and adjusting input data by reflecting dynamic variables, A method for predicting product prices using dynamic setting changes of multiple indicators.

2. In Paragraph 1, A further step of inputting adjusted input data into the price prediction model to predict the price of the item, A method for predicting product prices using dynamic setting changes of multiple indicators.

3. In Paragraph 1, The step of performing preliminary preprocessing for model training on the above indicator data is, A step of identifying and removing missing values ​​and noise from the above indicator data; and A step comprising normalizing and processing the time series data of each variable of the above indicator data, A method for predicting product prices using dynamic setting changes of multiple indicators.

4. In Paragraph 2, The step of providing the above-mentioned preprocessed indicator data to the price prediction model to train the price prediction model is, A step comprising training to learn the linear relationship between independent variables and dependent variables using the above price prediction model, A method for predicting product prices using dynamic setting changes of multiple indicators.

5. In Paragraph 4, The above price prediction model is a linear regression-based statistical learning model, A method for predicting product prices using dynamic setting changes of multiple indicators.

6. In Paragraph 4, The above price prediction model is, Proceed with training by ensuring a specific ratio between the training data and the validation data, To ensure that the proportion of training data is set higher than the proportion of validation data, A method for predicting product prices using dynamic setting changes of multiple indicators.

7. In Paragraph 2, The step of setting dynamic variables based on the above time delay and adjusting input data by reflecting the dynamic variables is, A step comprising setting the dynamic variable based on the time delay between the independent variable and the dependent variable, A method for predicting product prices using dynamic setting changes of multiple indicators.

8. In Paragraph 1, A method further comprising the step of re-establishing the importance of independent variables by comparing the predicted item price predicted by the above price prediction model with the actual item price. A method for predicting product prices using dynamic setting changes of multiple indicators.

9. A price prediction device that collects indicator data for multiple indicators affecting the price of goods, performs preliminary preprocessing on the indicator data for model training, and then provides the preprocessed indicator data to a price prediction model to train the price prediction model; and It includes a user terminal that checks product price prediction information through the above-mentioned price prediction device, and The above price prediction device is, Setting dynamic variables based on time delays and adjusting input data by reflecting dynamic variables, Product price prediction system using dynamic setting changes of multiple indicators.

10. A non-transient computer-readable recording medium on which a computer program executed by a computer is recorded, The above computer program is, A step of collecting indicator data for multiple indicators that affect the price of goods, performed in a goods price prediction system; A step of performing preliminary preprocessing for model training on the above indicator data; A step of providing preprocessed indicator data to a price prediction model to train the price prediction model; and A step including setting dynamic variables based on time delays and adjusting input data by reflecting dynamic variables, Non-transient computer-readable recording medium.

Images