Civil aviation passenger ticket quantity prediction method and device
By performing cluster analysis and time series prediction on the sub-cabin information of passenger tickets carried on historical flight segments, the error problem of ticket volume prediction in civil aviation passenger transport has been solved, achieving higher accuracy and cost-effectiveness.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-11-20
- Publication Date
- 2026-04-10
AI Technical Summary
In the field of civil aviation passenger transport, existing technologies have significant errors in cost calculation based on ticket volume, and the cost of estimating each use case separately is too high and difficult to explain the prediction results.
By collecting sub-cabin information of passenger tickets carried on historical flight segments, cluster analysis is used to divide them into different price markets, and time series forecasts are performed for each market to output the predicted passenger ticket volume.
This improved the accuracy of passenger ticket volume forecasting, reduced forecasting costs, and ensured the precision of market trends and the accuracy of cost calculations.
Smart Images
Figure CN121836778A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the technical field of aviation data, and in particular to a civil aviation ticket volume prediction method and device. BACKGROUND
[0002] In the field of civil aviation passenger transport, airlines (referred to as airlines) pay more and more attention to the accuracy of estimated accounts and prediction analysis. The calculation of incentive fees and other contents in the estimated accounts, as well as the development of multi-dimensional prediction analysis, all need to be based on passenger transport detailed transaction estimation data with high accuracy in both total volume and structure. This detailed transaction data should cover the information of ticket surface and product, passenger, booking, sales, carriage, settlement and other dimensions, and also include income amount information and ticket volume information.
[0003] In the prior art, unstructured time series prediction is usually used, whether it is a SARIMA (seasonal autoregressive integrated moving average model) or an LSTM (long short-term memory network) model, or a daily, weekly or monthly estimation dimension, which only carries out time series estimation on income. However, in actual data usage scenarios, fee calculation needs to be based on multiple dimensions such as ticket price, product, flight segment, advance sales period, passenger type, and sales channel, and the measurement of fee calculation can choose ticket volume or income. Since the existing scheme only estimates income, it will produce a large total volume error in the ticket volume-based fee calculation scenario, that is, due to the difference in the sales trend of different ticket prices, systematic errors will be formed. In addition, the fee calculation rules are clearly directed to specific products, cabins and flight segments, corresponding to specific markets and passenger groups. In this case, whether the fee calculation is based on ticket volume or income, the deviation will be further magnified. If separate estimation is made for each usage scenario, the cost will be too high, and the prediction results will be difficult to interpret. SUMMARY
[0004] The embodiment of the present application provides a civil aviation ticket volume prediction method to improve the accuracy of civil aviation ticket volume prediction while considering the prediction cost. The method comprises: Collecting sub-cabin information of historical flight segment carried passenger tickets; the sub-cabin information includes ticket price and ticket volume; Performing cluster analysis on the sub-cabin information of the historical flight segment carried passenger tickets to obtain a plurality of clusters; the center point of the cluster includes the average ticket price of the sub-cabin; According to the average ticket price of the sub-cabin of each cluster, the sub-cabin of the historical flight segment carried passenger tickets is divided into different price market; Respectively for different price markets, collect the historical ticket volume of each period, and perform ticket volume prediction based on time series prediction analysis method, and output the ticket volume prediction value of each price market.
[0005] This invention also provides a civil aviation passenger ticket volume prediction device to improve the accuracy of civil aviation passenger ticket volume prediction while taking into account prediction costs. The device includes: The data acquisition module is used to collect sub-cabin information of passenger tickets carried on historical flight segments; the sub-cabin information includes ticket price and ticket quantity. The market segmentation module is used to perform cluster analysis on the sub-cabin information of passenger tickets carried on historical flight segments to obtain multiple clusters; the centroid of each cluster includes the average fare of the sub-cabin; and the sub-cabins of passenger tickets carried on historical flight segments are divided into different price markets based on the average fare of the sub-cabins in each cluster. The ticket volume prediction calculation module is used to collect historical ticket volumes for different price market segments at different time periods, perform ticket volume prediction based on time series forecast analysis, and output the predicted ticket volume value for each price market segment.
[0006] This invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the above-described civil aviation passenger ticket volume prediction method.
[0007] This invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described civil aviation passenger ticket volume prediction method.
[0008] This invention also provides a computer program product, which includes a computer program that, when executed by a processor, implements the above-described civil aviation passenger ticket volume prediction method.
[0009] This invention avoids the conventional method of forecasting based on first class, business class, and economy class, as is often done in existing technologies. Instead, it divides the market into segments based on sub-cabin information of historical flight segments and actual market conditions. Then, it forecasts ticket volume separately for each price segment, achieving ticket volume forecasting for each segment. This invention categorizes markets based on ticket prices and sales volume for each sub-cabin, fully reflecting the actual market acceptance. Compared to segmenting the market according to airline-preset product types such as first class, business class, and economy class, it is more accurate in distinguishing the trends of different price segments, such as high, medium, and low-price markets, and is more conducive to market trend prediction. In addition, the sub-cabin classes are highly aligned with the discount tiers of civil aviation tickets; each airline has more than 20 sub-cabin classes, covering all fares from free tickets to first class, with moderate differentiation of discounts; the number and identification of sub-cabin classes remain stable year after year, making them easier to debug and operate in the long term compared to fields such as product codes, fare bases, and route segment value columns, which need to be constantly added or removed; at the same time, sub-cabin classes are part of the basic ticket information and do not require additional processing, which can reduce operating costs while predicting ticket volume. Attached Figure Description
[0010] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. In the drawings: Figure 1 This is a flowchart illustrating the civil aviation passenger ticket volume prediction method in an embodiment of the present invention; Figure 2 This is a specific example diagram of the civil aviation passenger ticket volume prediction method in the embodiments of the present invention; Figure 3 This is another specific example of the civil aviation passenger ticket volume prediction method in the embodiments of the present invention; Figure 4 This is another specific example of the civil aviation passenger ticket volume prediction method in the embodiments of the present invention; Figure 5 This is a schematic diagram illustrating the long-term sales trend forecast for the high, medium, and low-end markets in this embodiment of the invention. Figure 6 This is a schematic diagram of the civil aviation passenger ticket volume prediction device in an embodiment of the present invention; Figure 7 This is a schematic diagram of the server in an embodiment of the present invention. Detailed Implementation
[0011] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings. Here, the illustrative embodiments of the present invention and their descriptions are used to explain the present invention, but are not intended to limit the present invention.
[0012] First, the technical terms involved in this invention will be explained.
[0013] Time series: A time series is a series of data points arranged in chronological order, such as daily temperature or monthly sales figures. Its core value lies in predicting future changes by analyzing patterns in historical data, such as trends, seasonality, or cyclicality. In practical applications, time series analysis is commonly used in economic forecasting, inventory management, and weather forecasting.
[0014] Flight segment: A flight segment refers to an independently billable flight unit operating under a "single transport contract and single flight number" from one airport to another (including stopovers). 1. Defined primarily by the transport contract, i.e., the ticket and flight number, stopover segments under the same ticket may be split into multiple billing segments. 2. It is the basic unit for ticket pricing, revenue sharing between airlines, and other settlements.
[0015] Ticket Quantity: In the context of air ticketing, ticket quantity is essentially equivalent to the number of flight segment tickets, that is, the total number of ticket segments corresponding to the independent flight segments contained in a single ticket. For example, a connecting ticket for "Airport 1-Airport 2-Airport 3" contains 2 flight segments, so its ticket quantity is 2; a one-way ticket contains 1 flight segment, so its ticket quantity is 1. It is a basic indicator for airlines to calculate capacity and track ticket sales.
[0016] K-means clustering: K-means clustering is an unsupervised machine learning algorithm that divides samples in a dataset into K pre-defined clusters, ensuring high similarity within each cluster and low similarity between clusters. It iteratively selects K initial cluster centers, adjusts their positions, and reassigns clusters until the sum of squared errors within each cluster is minimized. It is commonly used for customer segmentation, image segmentation, and other similar scenarios.
[0017] Significance test: A significance test is a statistical method used to determine whether differences in sample data are caused by random factors or whether a real effect exists. Its core purpose is to verify the credibility of a hypothesis. It calculates a p-value and compares it to a pre-set significance level (e.g., 0.05). If the p-value is smaller, the null hypothesis is rejected, and the difference is considered statistically significant. It is commonly used in scenarios such as medical experiments and market research to verify results.
[0018] Adjusted R 2 (Adjusted R-squared, coefficient of determination): Adjusted R-squared 2 For R 2 The correction index, which measures the regression model's ability to explain the variation in the dependent variable while taking into account the number of independent variables in the model, is related to R. 2 Unlike other variables, it does not automatically increase when the independent variable increases. It only increases when the new variable truly enhances the explanatory power of the model, and can more objectively reflect the model's fit in practical applications.
[0019] The acquisition, storage, use, and processing of data in this application all comply with the relevant provisions of national laws and regulations.
[0020] Current solutions collect airlines' historical monthly or weekly revenue data over many years, input the data into a mathematical model, and estimate total revenue for the next month or week by analyzing historical trends. Revenue is a commonly used indicator, and its time series data is not only easy to collect but also relatively easy to predict. However, the prediction object is "overall" revenue. In actual use of the predicted data, it is often necessary to calculate revenue or ticket volume for specific markets or customer groups. If the trend of a specific market differs from the overall trend, significant deviations will occur. The trend of total revenue differs from the trend of total ticket volume because the average ticket price is not the same at each point in time. Therefore, when subsequent use scenarios require ticket volume, significant deviations will occur. Neither simple moving averages, SARIMA, nor LSTM can solve the above problems, making it difficult to further improve accuracy. The main difference between these models lies in their ability to extract trends from historical data, which mathematical models are powerless to address. Statistically, the accuracy of existing solutions is typically only around 75%-80%.
[0021] Furthermore, existing solutions only collect historical data on cost results in subsequent calculation stages. This historical cost time series is then input into a mathematical model to predict the total cost for the next period by analyzing historical trends. However, in reality, future changes are not limited to passenger market factors such as revenue and average ticket prices; they also include airline-specific rules. Existing rules may be modified, or new or abolished rules may be added. These adjustments are typically based on market changes and lack a fixed pattern. Because the historical cost time series does not include information related to revenue, average ticket prices, and rules, the mathematical model cannot accurately predict the total cost for the next period. Moreover, the prediction only covers the total cost for the next period and cannot be further refined, making it difficult to support the needs of sophisticated management.
[0022] To address the shortcomings of existing technologies, this invention provides a method and apparatus for predicting civil aviation passenger ticket volume, specifically a time-series prediction method for civil aviation passenger ticket volume that integrates classification factors. Figure 1 This is a flowchart illustrating the civil aviation passenger ticket volume prediction method in an embodiment of the present invention, as shown below. Figure 1 As shown, the method includes: Step 101: Collect sub-cabin information of passenger tickets for historical flight segments; the sub-cabin information includes ticket price and ticket quantity; Step 102: Perform cluster analysis using the sub-cabin information of passenger tickets carried on historical flight segments to obtain multiple clusters; the center point of each cluster includes the average ticket price of the sub-cabin. Step 103: Divide the sub-cabins of the historical flight segment passenger tickets into different price markets based on the average ticket price of each sub-cabin. Step 104: Collect historical ticket volume for each time period for different price market segments, predict ticket volume based on time series forecasting analysis, and output the predicted ticket volume value for each price market segment.
[0023] based on Figure 1 As shown in the method, this invention first performs key classification processing on segmented markets, and then predicts ticket volume. This invention predicts different price segments separately, replacing overall market trend predictions, to determine the criteria for distinguishing key markets. The market is segmented based on actual data feedback rather than airline-preset segmentation methods; a predictive indicator superior to revenue is found to ensure the accuracy of predicted revenue, ticket volume, and ticket information. The core of this invention lies in its ability to accurately predict ticket volume in segmented price segments, not based on common preset standards like first class, business class, and economy class, but on actual market conditions, achieving more effective market segmentation based on sub-cabin information.
[0024] The following describes in detail the method for predicting the number of civil aviation passenger tickets in the embodiments of the present invention.
[0025] First, collect information on the sub-cabin class of passenger tickets for historical flight segments; the sub-cabin class information includes ticket price and ticket quantity.
[0026] For example, to predict flight segments, information on the sub-cabins of tickets historically operated or sold can be collected.
[0027] Sub-cabins are further subdivided under the three main categories of First Class, Business Class, and Economy Class. They can be identified by different letter codes, and different sub-cabins differ in terms of fare discounts, benefits, etc.
[0028] In one embodiment, collecting sub-cabin information of historical flight segment tickets may include: The civil aviation passenger market is pre-divided into multiple regions. For a specific region, the flight segment with the highest sales volume in that region within a specified historical period is determined, and the sub-cabin information of the passenger tickets for that flight segment with the highest sales volume is collected.
[0029] For example, first determine whether the forecast target is ticket sales or ticket volume, and determine the length of the forecast period, such as month or week. Then, based on factors such as ticket pricing rules, customer characteristics, and whether they are similar or identical, divide the global passenger market into multiple regions. For example, it can be divided into domestic and international; it can also be further refined into domestic, Japan and South Korea, Europe, North America, etc. For each of these regions, conduct market segmentation and ticket volume forecasting separately.
[0030] For any given region, find the flight segment "AB" with the highest sales volume in that region over the past month, and then calculate the sub-cabin class, average fare for the sub-cabin class, and number of tickets for that sub-cabin class.
[0031] In step 102, cluster analysis is performed using the sub-cabin information of passenger tickets carried on historical flight segments to obtain multiple clusters; the center point of each cluster includes the average ticket price of the sub-cabin.
[0032] For example, by selecting a sub-cabin class as a key factor for market segmentation, and then clustering the sub-cabin class fares, the price range market to which the sub-cabin class belongs can be classified.
[0033] In one embodiment, cluster analysis is performed using the sub-cabin information of passenger tickets carried on historical flight segments to obtain multiple clusters. This may include: using the average ticket price of the sub-cabin as a distance metric and the number of tickets in the sub-cabin as a weighting factor, and using the K-means clustering method to perform cluster analysis on the sub-cabin information of passenger tickets carried on historical flight segments to obtain multiple clusters.
[0034] For example, for each sub-cabin class of flight segment "AB", K-means clustering algorithm is used for cluster analysis, with K=3. The clustering process uses the "average ticket price of sub-cabin" as the distance metric and the "number of tickets in sub-cabin" as the sample weight, ultimately dividing the sub-cabin classes of the flight segment into 3 clusters. At the same time, 3 corresponding centroids are obtained, which are the average ticket prices of the clusters.
[0035] In step 103, the sub-cabins of the historical flight segment passenger tickets are divided into different price markets based on the average ticket price of each sub-cabin.
[0036] For example, after dividing into multiple clusters, the sub-cabins are further divided into price segment 1 market, price segment 2 market, price segment 3 market, price segment 4 market, etc., according to the average ticket price of the sub-cabins from high to low.
[0037] In one embodiment, cluster analysis of each sub-cabin of a flight segment can divide the market into three levels: high, medium, and low, namely, high-end market, mid-range market, and low-end market.
[0038] Sub-cabin classes and discount levels are equivalent, thus yielding the fare discount ranges for high, mid, and low-end markets. Domestically, the published fare for class Y on a flight segment is used as a strict benchmark, while internationally, the average fare for class Y (a sub-cabin class) on a flight segment is used as an approximate benchmark.
[0039] In one embodiment, after dividing the sub-cabin classes of historical flight segment tickets into different price markets based on the average fare of each sub-cabin class in each cluster, the method may further include: Compare the sub-cabin class of the historical flight segment passenger tickets collected with the full quantum cabin class value of the designated area. If there are missing sub-cabin classes in the collected information, the missing sub-cabin classes are classified into different price market to obtain the updated different price market. For different price segments, historical ticket volumes for each time period are collected. Ticket volume forecasts are then performed using time series forecasting analysis, outputting the predicted ticket volume for each price segment, including: For each updated price segment market, historical ticket volume for each time period is collected, and ticket volume is predicted based on time series forecast analysis, outputting the predicted ticket volume value for each price segment market.
[0040] For example, the historical sub-cabin information collected previously may differ from the total number of sub-cabins for that flight segment. To further improve the accuracy of ticket volume prediction, it is necessary to detect whether any sub-cabins are missing from the collected information and add the missing sub-cabins to the clusters. In practice, sub-cabins in the same cluster have similar characteristics, such as similar discounts. The missing sub-cabins can be classified into clusters with the corresponding discount ranges according to the historical discount range of the missing sub-cabins.
[0041] For example, the average fare (discount) for each sub-class within a cluster will fall within a continuous interval. The missing sub-class will be assigned to the cluster corresponding to the interval in which its historical average fare (discount) falls. If this is still uncertain, the cluster closest to the center will be used to assign it to the missing sub-class. Subsequent forecasts will then utilize the updated market classification for that flight segment.
[0042] In the final step 104, historical ticket volumes for different price market segments are collected for each time period. Ticket volume is predicted based on time series forecasting analysis, and the predicted ticket volume value for each price market is output.
[0043] In this embodiment of the invention, time series forecasts are performed for different price markets. The model can be exponential smoothing, SARIMA model or LSTM model, etc., and the final ticket volume forecast results can meet the needs of multiple scenarios.
[0044] In one embodiment, step 104, for different price market segments, collects historical ticket volumes for each time period, performs ticket volume forecasting based on time series forecasting analysis, and outputs the predicted ticket volume value for each price market segment, which may include: For any price range market, historical ticket volumes for each time period are collected, and multiple time series forecasting models are constructed. Each time series forecasting model uses the historical ticket volumes for each time period of the market at any price range as the input variable for forecasting, and outputs the predicted ticket volume value for the market at any price range. Each time series forecasting model has a different combination of model parameters. The optimal model is obtained by screening multiple time series prediction models using preset indicators; these preset indicators include error, robustness, and interpretability. The optimal model is used to predict ticket volume for different price markets, and the predicted ticket volume value for each price market is output.
[0045] In one embodiment, the time series forecasting analysis method includes triple exponential smoothing, specifically the Winters exponential smoothing method. Further, for different price market segments, historical ticket volumes for each time period are collected, and ticket volume forecasts are performed based on the time series forecasting analysis method, outputting the predicted ticket volume value for each price market segment, which may include: For any price market, historical ticket volumes for each time period are collected. Based on triple exponential smoothing, multiple time series forecasting models are constructed. Each time series forecasting model uses the historical ticket volumes for each time period in the market at any price level as the input variable for forecasting, and outputs the predicted ticket volume for that market at any price level. Each time series forecasting model has a different combination of model parameters. The combination of model parameters includes model type and dependent variable transformation method. The model type includes Winters additive and Winters multiplicative, and the dependent variable transformation method includes natural logarithm transformation, square root transformation, and no transformation. Each time series forecasting model is used to predict ticket volume and perform significance tests. Among the time series forecasting models that pass the significance test, the time series forecasting model with the largest adjusted coefficient of determination is selected as the optimal model. The optimal model is used to predict ticket volume for different price markets, and the predicted ticket volume value for each price market is output.
[0046] In this example, different combinations of model parameters are used to screen the model, and the results are evaluated through significance testing and adjusted R-squared values. 2 To select the optimal model.
[0047] Figure 2 This is a specific example diagram of the civil aviation passenger ticket volume prediction method in an embodiment of the present invention, as shown in the figure. Figure 2 As shown, after outputting the predicted ticket volume for each price segment in step 104, the method may further include: Step 201: Predict ticket revenue based on the predicted ticket volume for each price segment and the average ticket price for each sub-cabin class in each price segment.
[0048] In this example, the ticket volume of different price markets is predicted separately. The predicted revenue of a market is obtained by multiplying the predicted ticket volume of each price market by the average ticket price of the corresponding cluster. The predicted revenue of all price markets is then aggregated, and finally, the sales revenue or transportation revenue is predicted based on the ticket volume.
[0049] The direct objective of this invention is to significantly improve the accuracy of predicting total revenue and ticket volume for the next time period, while simultaneously accurately predicting the ticket price structure for the next period. This ticket price structure reflects different growth trends in ticket volume for different price segments within the same market. For example, this can be achieved by accurately predicting ticket volume for the high, medium, and low-priced ticket markets.
[0050] The indirect objectives of this invention are as follows: First, to use the predicted ticket volume to generate more accurate detailed ticket forecast data records that include revenue, ticket volume, and fare structure distribution; second, to combine the detailed ticket forecast data records with cost calculation rules to calculate the predicted cost in a traditional system. This is the subsequent practical forecasting application of this solution.
[0051] The ultimate goal of this invention is to improve the accuracy of direct forecast results and subsequent applications, covering ticket volume, revenue, expenses, and various forecast analyses. This reflects airlines' accuracy requirements for subsequent practical forecast applications.
[0052] Figure 3 This is another specific example diagram of the civil aviation passenger ticket volume prediction method in the embodiments of the present invention, as shown in the figure. Figure 3 As shown, the method includes: Step 301: Select the prediction area and prediction period length; Step 302: Select a data source; Step 303: Select flight segment cluster analysis; Step 304: Based on the cluster analysis results, divide the sub-cabins of the flight segment into three levels: high, medium, and low. Step 305: Supplement the missing sub-cabins for each level of the flight segment based on historical data; Step 306: Predict ticket volume for each class of cabin and generate multiple predicted trends; Step 307: For each class of sub-classes, retain the most suitable predicted trend; Step 308: Compare the fitted and predicted data with historical data to observe and judge the rationality of the predicted trend of each class of sub-classes.
[0053] Combination Figure 3 A specific example is described below.
[0054] 1. Determine the length of the time period for predicting passenger ticket volume, with options of month or week; at the same time, determine whether the prediction object is the sales volume or the transportation volume.
[0055] 2. Based on factors such as ticket pricing rules, customer characteristics, and whether they are uniform or similar, the global passenger transport market can be divided into multiple regional markets. For example, it can be divided into domestic and international markets.
[0056] 3. For each of the above-mentioned major passenger transport regions, conduct market segmentation and ticket volume forecasting.
[0057] The following example uses a major region within China: 4. Market segmentation in major domestic regions.
[0058] 4.1 Find the domestic region with the highest sales volume of the flight segment "AB" in the past month, and count the sub-cabins, average ticket price of the sub-cabins, and number of tickets for the sub-cabins carried by the flight segment.
[0059] 4.2 For each sub-cabin class of flight segment "AB", K-means clustering (K=3) algorithm is used for cluster analysis. The clustering process uses the "average ticket price of sub-cabin class" as the distance metric and the "number of tickets in sub-cabin class" as the sample weight, ultimately dividing the sub-cabin classes of the flight segment into 3 clusters. Simultaneously, 3 corresponding centroids are obtained, representing the average ticket price of each cluster.
[0060] 4.3. Based on the order of cluster centers from high to low, the sub-cabins are divided into three groups of sub-cabins in the "AB" segment: high-end, mid-end, and low-end markets.
[0061] 4.4. Sub-cabin classes and discount levels are equivalent, thus yielding the fare discount ranges for high, medium, and low-end markets. The domestic market uses the published fare for class Y of a flight segment as a strict benchmark, while the international market uses the average fare for class Y of a flight segment as an approximate benchmark.
[0062] 4.5. Compare the current statistics of sub-cabin slots for the "AB" flight segment with the historical values for sub-cabin slots in this region. If any sub-cabin slots are missing, categorize them according to historical discount rates into the high, mid, or low-end market sub-cabin slot value columns corresponding to their respective discount ranges. If the location remains uncertain, assign them to the cluster closest to the center of the current cluster. This step ensures the completeness and mutual complementarity of sub-cabin slots across the high, mid, and low-end markets.
[0063] 4.6 In subsequent forecasts, the high, medium and low-end market sub-cabins of the "AB" segment will be approximately regarded as the high, medium and low-end market sub-cabins of the domestic region.
[0064] 5. Forecast ticket sales for the high-end, mid-end, and low-end markets in different regions of China.
[0065] Taking the high-end market and the Winters exponential smoothing model as an example: 5.1 Using the high-end market sub-cabin group in the domestic region as the screening criterion, the number of tickets sold in each historical period in China was statistically analyzed.
[0066] 5.2 A time series forecasting model was constructed using the Winters exponential smoothing model, with historical ticket volume in the high-end market of major domestic regions as the forecast variable; 6 different combinations of model parameters were set, with each set of parameters corresponding to one independent forecast, for a total of 6 forecasts.
[0067] 5.3. The six different combinations of model parameters (including model type and dependent variable transformation method) are as follows: Winters additive property, dependent variable transformation - none (i.e. no transformation); Winters multiplicative property, dependent variable transformation - none; Winters additive properties, dependent variable transformation - natural logarithm; Winters multiplication, dependent variable transformation - natural logarithm; Winters additive property, dependent variable transformation - square root; Winters multiplicative property, dependent variable transformation - square root.
[0068] 5.4 The model selection rules are as follows: Prioritize parameter combinations whose overall model passes the significance test; among the models that pass the significance test, further select the adjusted R-values. 2 The largest model is selected as the optimal model. Based on this optimal model, the predicted ticket volume for the high-end market in the domestic region for the next time period is output.
[0069] 6. After processing each item, the predicted ticket volume for the next period in the high-end, mid-end, and low-end markets of each major region is obtained, achieving complete coverage of the predicted ticket volume for all markets.
[0070] 7. Plot the historical and predicted ticket volumes for high, medium, and low-end markets in major regions into curves. Observe and judge the rationality of the predicted trend to avoid basic operational errors.
[0071] 8. Subsequent Use: The passenger ticket volume forecast can be used for revenue forecasting in different market segments and rule-based cost estimation; if aggregated, it can also be used as the total passenger ticket volume forecast for the next period.
[0072] In summary, this invention predicts the time series of three markets—high, medium, and low-priced—separately. Combining the characteristics of civil aviation operations with comparative traversal experiments, it confirms that the time series trends of the high, medium, and low-priced markets exhibit significant differences and can represent changes in ticket price structure. Sub-cabin class is determined as the key field for distinguishing between these markets. A clustering algorithm is used to classify the high, medium, and low-priced markets to which each sub-cabin class belongs based on sub-cabin class and ticket volume. Ticket volume is used instead of revenue as the prediction metric. In this embodiment, the high, medium, and low-priced markets obtained through data classification based on sub-cabin class show significant differences in long-term trends and good distinguishability. Separate predictions not only achieve accuracy in predicting each market and the accuracy of ticket price structure but also improve overall prediction accuracy. The design of the predicted ticket volume ensures correct relationships between the information in the subsequently generated predicted ticket details, further guaranteeing the structural accuracy of the cost calculation results. Statistical data shows that the accuracy of the subsequent prediction of total costs is on par with the accuracy of the prediction of total ticket volume (in existing solutions, revenue prediction is relatively inaccurate, and the deviation increases significantly after cost calculation).
[0073] Comparative verification shows that, in terms of total volume: the old scheme has an accuracy rate of approximately 80% in directly estimating total revenue and approximately 75% in indirectly estimating total tickets; the scheme of this invention achieves an accuracy rate of approximately 92% in directly estimating total tickets, with an error reduction of approximately 70%, and an accuracy rate of approximately 93% in indirectly estimating total revenue, with an error reduction of approximately 60%. Furthermore, the scheme of this invention performs better in various market segments, and even after aggregating the costs from each market segment, the accuracy rate still reaches 92%-93%.
[0074] Figure 4 This is another specific example diagram of the civil aviation passenger ticket volume prediction method in the embodiments of the present invention, as shown in the figure. Figure 4 As shown, the overall scheme for cost estimation includes: inputting historical basic data, predicting ticket volume trends, generating estimated basic data, maintaining calculation rules, and calculating estimated costs.
[0075] The following aspects need to be considered when estimating the basic data: the need for ticket-level information in the calculation, the significant differences in growth trends in different markets (such as domestic and international markets, and high, medium and low-priced ticket markets), the quantitative prediction of growth trends, and the possibility that experts may make additional adjustments to the growth ratio.
[0076] The solution generates ticket-level forecast data by using historical data, differentiating between different markets, quantifying and predicting trends separately for each market, and allowing users to set adjustment rates for different markets based on forecasts and expert opinions. The key aspects are differentiating between different markets and quantifying and predicting trends separately for each market, which utilizes clustering and time series methods respectively.
[0077] The sales trends of high, medium, and low-priced tickets are completely different. For example, in 2019, the number of tickets sold increased, but the total sales revenue decreased, which was particularly evident. In recent years, marketing activities such as significant discounts on premium economy class and first and business class tickets by various airlines have made it impossible to directly use the main cabin class and sub-cabin class to distinguish the high, medium, and low-priced ticket markets. This solution adopts a K-order clustering model, which repeatedly calculates three clustering points based on distance in the fare space of all ticket chains. The model uses the last stable clustering point as the center position of the high, medium, and low-priced ticket markets. The advantage of this method is that it utilizes all the data and reclassifies the sub-cabin classes based on two dimensions: market sales volume and discount level, thus defining the high, medium, and low-priced tickets that are recognized by the market.
[0078] Figure 5 This is a schematic diagram illustrating the long-term sales trend prediction of the high, medium, and low-end markets in an embodiment of the present invention, such as... Figure 5 As shown, the long-term sales trend forecast results for the high, medium, and low-end markets are displayed from top to bottom. To the left of the black vertical line: red represents actual data, and blue represents fitted data; to the right of the black vertical line: blue represents predicted data. Figure 5The horizontal and vertical axes represent dates and months, while the vertical axis represents sales volume. It can be seen that the long-term sales forecasts for the high, medium, and low-end markets in this embodiment of the invention are consistent with the actual figures.
[0079] In this embodiment of the invention: 1. Introduce the concept and methodology of market segmentation at different price points. Sales trends in different market segments vary significantly. By forecasting separately and then aggregating the data, accuracy is improved at both the overall volume and structural levels.
[0080] 2. Market segmentation using clustering algorithms. This algorithm categorizes markets based on ticket prices and sales volume for each sub-class, fully reflecting the actual market acceptance. Compared to segmenting the market according to airline-preset product types such as first class, business class, and economy class, it is more accurate in distinguishing the trends of high, medium, and low-end markets and is more conducive to predicting market trends.
[0081] 3. Select sub-cabin classes as the key factor for market segmentation, i.e., clustering objects. Sub-cabin classes are highly matched with the discount gradient of civil aviation tickets; each airline has more than 20 sub-cabin classes, which can cover all fares from free tickets to first class, and the differentiation of discounts is moderate; the number and identification of sub-cabin classes are stable over the years, which is easier to debug and operate in the long term compared with product codes, fare bases, and route segment value columns, which need to be added or abolished continuously; at the same time, sub-cabin classes are basic ticket information and do not require additional processing, which can reduce operating costs.
[0082] 4. Predict ticket sales rather than revenue.
[0083] Ticket information includes details on the ticket price and product, passenger, reservation, sales, transportation, and settlement, as well as revenue and ticket volume information. The multi-dimensional information of a single ticket is closely linked and cannot be adjusted individually. In the subsequent steps of generating predicted detailed data records based on predicted trends, existing solutions, if they estimate an increase in revenue and adjust ticket prices accordingly, would cause a disconnect between the estimated revenue and the sub-cabin class, fare base, and product code, leading to subsequent usage errors. However, the solution in this invention adjusts the corresponding ticket volume by estimating ticket volume trends, ensuring consistency across all fields of the predicted ticket details while simultaneously adjusting revenue. Because revenue naturally increases proportionally when the number of tickets increases from 1 to 2, this guarantees the uniformity and accuracy of the predicted ticket detail records, laying the foundation for subsequent high-quality data processing.
[0084] From the perspective of the nature of the civil aviation ticketing business, revenue changes are caused by changes in ticket volume in different sub-classes (which are essentially equivalent to discounts), rather than airlines raising or lowering the price level of specific sub-classes. Therefore, forecasting based on ticket volume in sub-classes is more accurate.
[0085] Revenue can be used as an indirect forecast result, derived from ticket volume forecasts, and its total amount and structure are more accurate.
[0086] 5. Employ mathematical models to analyze historical trends and predict future sales. Commonly used prediction models include exponential smoothing, SARIMA, and LSTM. The choice of model depends on market volatility: for large markets where various noises cancel each other out and fluctuations are relatively regular, exponential smoothing can be used; for small airline markets where noise is more abundant and difficult to cancel out, the LSTM model is more effective. In this solution, the choice of prediction model depends on the characteristics of the data itself. For large regional markets, since the data trends are relatively stable, exponential smoothing is usually sufficient; if market fluctuations are more complex, a more suitable model can be used. The specific prediction model chosen is not the core focus of this solution; those skilled in the art can choose based on the actual situation.
[0087] In airline monthly forecast scenarios, airlines generally require forecast accuracy of over 95%. Through decomposition and reverse engineering, the trend forecasting modeling stage needs to achieve an accuracy rate of over 85% in both total volume and structure. Verification analysis shows that this solution not only meets the requirements but also significantly exceeds airline expectations in terms of accuracy. While other earlier solutions introduced various forecasting models, parameter tuning alone could not meet the accuracy requirements. This invention takes into account the differences in customer groups across different ticket price markets, whose sales trends vary significantly in terms of seasonality, annual trends, and volatility. Therefore, it proposes forecasting separately for each scenario to improve overall and structural accuracy.
[0088] On the other hand, in the scenario of monthly estimated billing for airlines, the rules for calculating costs are related to dozens of information items such as revenue, ticket volume, products, cabin class, flight segment, domestic and international routes, and flight dates. How to select information to differentiate markets and customer groups becomes a key issue. This embodiment of the invention selects sub-cabin classes as the criteria for market segmentation. Sub-cabin classes have advantages such as a limited number (more than 20, neither too many nor too few), a strong correlation with discounts in specific regions, and no new additions in the long term. In contrast, other dimensions: only 3-4 cabin classes, which is too coarse; discount calculation requires the introduction of domestically published fares, and international rules are more complex; the number of basic fares reaches tens of thousands, and the continuity is poor; business travel routes lack differentiation between routes and lack differentiation within routes. From the results of the solution verification, sub-cabin classes do indeed have good differentiation within a large region. The trends of high-end, mid-end, and low-end markets segmented according to sub-cabin classes are significantly different, which also reflects the structural differences in revenue and ticket volume.
[0089] Furthermore, this embodiment of the invention employs a clustering method for sub-cabin classes, combining the average ticket price and ticket volume of each sub-cabin class actually sold in the market to derive high-end, mid-end, and low-end market classifications, which have gained market acceptance. The trends of this classification better reflect changes in market structure and can meet the scenario requirements of detailed dimension-based calculations in cost calculation rules. This approach ensures that market segmentation is more realistic and more accurately identifies the trends of different market segments. In later applications, the accuracy of the cost estimation calculation stage is indeed basically on par with the accuracy of ticket volume prediction in this solution.
[0090] During the verification of the solution, the early sub-cabin data needs to be cleaned first to improve the accuracy of sub-cabin prediction; if the historical data accumulated after the airline changes is long enough, there is no need to use the early sub-cabin data.
[0091] This invention also provides a civil aviation passenger ticket volume prediction device, as described in the following embodiments. Since the principle behind this device's problem-solving is similar to that of the civil aviation passenger ticket volume prediction method, its implementation can refer to the implementation of the civil aviation passenger ticket volume prediction method; repeated details will not be elaborated further.
[0092] Figure 6 This is a schematic diagram of a civil aviation passenger ticket volume prediction device in an embodiment of the present invention, such as... Figure 6 As shown, the device 600 includes: Data acquisition module 601 is used to collect sub-cabin information of passenger tickets carried on historical flight segments; the sub-cabin information includes ticket price and ticket quantity; The market segmentation processing module 602 is used to perform cluster analysis using the sub-cabin information of passenger tickets carried on historical flight segments to obtain multiple clusters; the center point of each cluster includes the average fare of the sub-cabin; and the sub-cabin of passenger tickets carried on historical flight segments is divided into different price markets based on the average fare of the sub-cabin in each cluster. The ticket volume prediction calculation module 603 is used to collect historical ticket volumes for different price market periods, perform ticket volume prediction based on time series forecast analysis, and output the ticket volume prediction value for each price market.
[0093] In one embodiment, the data acquisition module 601 is specifically used for: The civil aviation passenger market is pre-divided into multiple regions. For a specific region, the flight segment with the highest sales volume in that region within a specified historical period is determined, and the sub-cabin information of the passenger tickets for that flight segment with the highest sales volume is collected.
[0094] In one embodiment, the market segmentation processing module 602 is specifically used for: Using the average ticket price of each sub-cabin as the distance metric and the number of tickets allocated to each sub-cabin as the weighting factor, the K-means clustering method was used to perform cluster analysis on the sub-cabin information of passenger tickets carried on historical flight segments, resulting in multiple clusters.
[0095] In one embodiment, the device 600 further includes: The missing sub-cabin class supplementation module is used to compare the collected sub-cabin class data of historical flight segment operating tickets with the full quantum cabin class value column of the designated area after the market classification processing module 602 divides the sub-cabin class of historical flight segment operating tickets into different price market based on the average ticket price of the sub-cabin class. If there are missing sub-cabin classes in the collected information, the missing sub-cabin classes are assigned to different price market, and the updated different price market is obtained. The ticket volume prediction calculation module 603 is specifically used for: For each updated price segment market, historical ticket volume for each time period is collected, and ticket volume is predicted based on time series forecast analysis, outputting the predicted ticket volume value for each price segment market.
[0096] In one embodiment, the ticket volume prediction calculation module 603 is specifically used for: For any price range market, historical ticket volumes for each time period are collected, and multiple time series forecasting models are constructed. Each time series forecasting model uses the historical ticket volumes for each time period of the market at any price range as the input variable for forecasting, and outputs the predicted ticket volume value for the market at any price range. Each time series forecasting model has a different combination of model parameters. The optimal model is obtained by screening multiple time series prediction models using preset indicators; the preset indicators include error and robustness. The optimal model is used to predict ticket volume for different price markets, and the predicted ticket volume value for each price market is output.
[0097] In one embodiment, the time series forecasting analysis method includes triple exponential smoothing; The ticket volume prediction calculation module 603 is specifically used for: For any price range market, historical ticket volumes for each time period are collected, and multiple time series forecasting models are constructed using triple exponential smoothing. Each time series forecasting model uses the historical ticket volumes for each time period in that arbitrary price range market as the input variable for forecasting, and outputs the predicted ticket volume value for that arbitrary price range market. Each time series forecasting model has a different combination of model parameters. These model parameter combinations include model type and dependent variable transformation method. Model types include Winters additive and Winters multiplicative transformations, and dependent variable transformation methods include natural logarithm transformation, square root transformation, and no transformation. Each time series forecasting model is used to predict ticket volume and perform significance tests. Among the time series forecasting models that pass the significance test, the time series forecasting model with the largest adjusted coefficient of determination is selected as the optimal model. The optimal model is used to predict ticket volume for different price markets, and the predicted ticket volume value for each price market is output.
[0098] In one embodiment, the device 600 further includes: The cost estimation module is used to predict ticket revenue based on the predicted ticket volume for each price market and the average ticket price for each sub-cabin class after the ticket volume prediction calculation module 603 outputs the predicted ticket volume for each price market.
[0099] This invention also provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the above-described civil aviation passenger ticket volume prediction method.
[0100] Figure 7 This is a schematic diagram of a server in an embodiment of the present invention, as shown below. Figure 7 As shown, this embodiment of the invention also provides a server, including a processor, a storage medium, a database, an input / output interface, an operating system based on the processor, an application program, etc., and the processor implements the above-mentioned civil aviation passenger ticket volume prediction method when executing the computer program.
[0101] This invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described civil aviation passenger ticket volume prediction method.
[0102] This invention also provides a computer program product, which includes a computer program that, when executed by a processor, implements the above-described civil aviation passenger ticket volume prediction method.
[0103] For scenarios involving estimated accounting and structured calculations or analysis, the predictive cost complexity of this invention is moderate, while its accuracy and interpretability are sufficiently high. In terms of performance, this solution is superior.
[0104] In terms of total volume: the old solution had an accuracy rate of about 80% in directly estimating total revenue and about 75% in indirectly estimating ticket volume, which did not meet the client's requirement of 85% accuracy in revenue estimation.
[0105] The accuracy rate of directly estimating total ticket volume in this embodiment of the invention can reach approximately 92%, with an error reduction of about 70%; the accuracy rate of indirectly estimating total revenue can also reach approximately 92%, with an error reduction of about 60%. Furthermore, this embodiment of the invention performs even better in various market segments, and after aggregating the costs from each market segment, the accuracy rate still reaches 92%-93%, significantly exceeding the client's requirement of 85% accuracy in revenue estimation.
[0106] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0107] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.
[0108] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.
[0109] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 Figure 1 The steps of the function specified in one or more boxes.
[0110] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.
Claims
1. A method for predicting the volume of civil aviation passenger tickets, characterized in that, include: Collect sub-cabin information of passenger tickets carried on historical flight segments; the sub-cabin information includes ticket price and ticket quantity; Cluster analysis was performed using sub-cabin information of passenger tickets carried on historical flight segments to obtain multiple clusters; the centroid of each cluster includes the average ticket price of the sub-cabin. Based on the average fare of each sub-cabin class in each cluster, the passenger tickets for historical flight segments are divided into different price markets. For different price markets, historical ticket volumes for each time period are collected, and ticket volume is predicted based on time series forecasting analysis, outputting the predicted ticket volume value for each price market.
2. The method as described in claim 1, characterized in that, Collect sub-cabin information for tickets carried on historical flight segments, including: The civil aviation passenger market is pre-divided into multiple regions. For a specific region, the flight segment with the highest sales volume in that region within a specified historical period is determined, and the sub-cabin information of the passenger tickets for that flight segment with the highest sales volume is collected.
3. The method as described in claim 1, characterized in that, Cluster analysis was performed using sub-cabin information of passenger tickets carried on historical flight segments, resulting in multiple clusters, including: Using the average ticket price of each sub-cabin as the distance metric and the number of tickets allocated to each sub-cabin as the weighting factor, the K-means clustering method was used to perform cluster analysis on the sub-cabin information of passenger tickets carried on historical flight segments, resulting in multiple clusters.
4. The method as described in claim 2, characterized in that, After dividing the historical flight segment's passenger tickets into different price markets based on the average fare of each sub-cabin class, it also includes: Compare the sub-cabin class of the historical flight segment passenger tickets collected with the full quantum cabin class value of the designated area. If there are missing sub-cabin classes in the collected information, the missing sub-cabin classes are classified into different price market to obtain the updated different price market. For different price segments, historical ticket volumes for each time period are collected. Ticket volume forecasts are then performed using time series forecasting analysis, outputting the predicted ticket volume for each price segment, including: For each updated price segment market, historical ticket volume for each time period is collected, and ticket volume is predicted based on time series forecast analysis, outputting the predicted ticket volume value for each price segment market.
5. The method as described in claim 1, characterized in that, For different price segments, historical ticket volumes for each time period are collected. Ticket volume forecasts are then performed using time series forecasting analysis, outputting the predicted ticket volume for each price segment, including: For any price range market, historical ticket volumes for each time period are collected, and multiple time series forecasting models are constructed. Each time series forecasting model uses the historical ticket volumes for each time period of the market at any price range as the input variable for forecasting, and outputs the predicted ticket volume value for the market at any price range. Each time series forecasting model has a different combination of model parameters. The optimal model is obtained by screening multiple time series prediction models using preset indicators; the preset indicators include error and robustness. The optimal model is used to predict ticket volume for different price markets, and the predicted ticket volume value for each price market is output.
6. The method as described in claim 5, characterized in that, After outputting the ticket volume forecast for each price segment, the following is also included: Ticket revenue is predicted based on the predicted ticket volume for each price segment and the average ticket price for each sub-cabin class in each price segment.
7. A civil aviation passenger ticket volume prediction device, characterized in that, include: The data acquisition module is used to collect sub-cabin information of passenger tickets carried on historical flight segments; the sub-cabin information includes ticket price and ticket quantity. The market segmentation module is used to perform cluster analysis on the sub-cabin information of passenger tickets carried on historical flight segments to obtain multiple clusters; the centroid of each cluster includes the average fare of the sub-cabin; and the sub-cabins of passenger tickets carried on historical flight segments are divided into different price markets based on the average fare of the sub-cabins in each cluster. The ticket volume prediction calculation module is used to collect historical ticket volumes for different price market segments at different time periods, perform ticket volume prediction based on time series forecast analysis, and output the predicted ticket volume value for each price market segment.
8. The apparatus as claimed in claim 7, characterized in that, The data acquisition module is specifically used for: The civil aviation passenger market is pre-divided into multiple regions. For a specific region, the flight segment with the highest sales volume in that region within a specified historical period is determined, and the sub-cabin information of the passenger tickets for that flight segment with the highest sales volume is collected.
9. The apparatus as claimed in claim 7, characterized in that, The market segmentation processing module is specifically used for: Using the average ticket price of each sub-cabin as the distance metric and the number of tickets allocated to each sub-cabin as the weighting factor, the K-means clustering method was used to perform cluster analysis on the sub-cabin information of passenger tickets carried on historical flight segments, resulting in multiple clusters.
10. The apparatus as claimed in claim 8, characterized in that, Also includes: The missing sub-cabin class supplementation module is used to compare the collected sub-cabin class data of historical flight segment operating tickets with the full quantum cabin class value column of the designated area after the market classification processing module divides the sub-cabin class of historical flight segment operating tickets into different price market based on the average ticket price of the sub-cabin class. If there are missing sub-cabin classes in the collected information, the missing sub-cabin classes are assigned to different price market, and the updated different price market is obtained. The ticket volume prediction and calculation module is specifically used for: For each updated price segment market, historical ticket volume for each time period is collected, and ticket volume is predicted based on time series forecast analysis, outputting the predicted ticket volume value for each price segment market.
11. The apparatus as claimed in claim 7, characterized in that, The ticket volume prediction and calculation module is specifically used for: For any price range market, historical ticket volumes for each time period are collected, and multiple time series forecasting models are constructed. Each time series forecasting model uses the historical ticket volumes for each time period of the market at any price range as the input variable for forecasting, and outputs the predicted ticket volume value for the market at any price range. Each time series forecasting model has a different combination of model parameters. The optimal model is obtained by screening multiple time series prediction models using preset indicators; the preset indicators include error and robustness. The optimal model is used to predict ticket volume for different price markets, and the predicted ticket volume value for each price market is output.
12. The apparatus as claimed in claim 11, characterized in that, Also includes: The cost estimation module is used to predict ticket revenue based on the predicted ticket volume for each price market and the average ticket price for each sub-cabin class after the ticket volume prediction calculation module outputs the predicted ticket volume for each price market.
13. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method of any one of claims 1 to 6.
14. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the method of any one of claims 1 to 6.
15. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the method of any one of claims 1 to 6.