Flight reservation prediction method and device, storage medium and electronic equipment

Abstract

The invention belongs to the technical field of freight rate search, and provides a flight reservation prediction method and device, a storage medium and electronic equipment, and the method comprises the steps: obtaining a classification method of a target flight, and carrying out the classification of a peak season P, a flat season D and a slack season O based on the corresponding classification method, and obtaining a classification result comprising a plurality of PDO marks; all PDO marks in the classification result are marked into a seat reservation data table of the target flight; and inputting the data in the seat reservation data table of the target flight into the prediction model for parameter estimation, and outputting a prediction result. According to the prediction method, each piece of data input into the prediction model is marked with the corresponding PDO mark, so that the reservation rate of the target flight on the specified date in the future can be accurately predicted finally.

Description

Technical Field

[0001] The present invention belongs to the technical field of fare search, and in particular, relates to a flight booking prediction method, device, storage medium and electronic device. Background Art

[0002] Flight demand forecasting, referred to as flight prediction, is the prediction of future passenger booking demand at different data processing points before the flight departs, which serves as the input to more complex flight seat control models.

[0003] Flight forecasting is one of the core issues of airline revenue management methods. Typical methods for flight forecasting research include: improvement of traditional forecasting models, time series models, grey topology models, clustering methods, BP neural networks, dual LSTM, deep learning networks, etc. Most of these studies are minor improvements on existing methods or the application of new forecasting methods to the problem of flight forecasting.

[0004] In engineering practice, simple and robust flight prediction models such as regression, increment, multiplication, addition, and zero booking are mainly used.

[0005] Flight prediction generally includes two basic steps, as described below: First, historical data is selected to learn model parameters, and then, combined with the real-time data of the flight to be predicted, the model parameters obtained by the above learning are used to predict the future booking demand of the flight. In these two steps, the selection of the historical data set for model parameter learning and the model parameters obtained by which historical data set should be substituted for the flight to be predicted play a crucial role in the accuracy of flight prediction.

[0006] The selection of existing model parameter estimation data sets generally takes into account seasonal characteristics, holiday characteristics, and especially DOW (Day of Week) characteristics.

[0007] From the perspective of civil aviation revenue management engineering practice, airlines generally use a combination of DOW and manual setting to estimate model parameters. The specific rules used are as follows:

[0008] Rule 1: For flights affected by holidays or special events, the departure dates of the future flights to be predicted are matched to the same year-on-year dates of the same holiday or special event in previous years, such as last year or the year before, to obtain a set of year-on-year dates. The data of historical flights in this set are used to estimate the model parameters for future flight predictions.

[0009] Rule 2: For flights that are not affected by holidays or special events, historical and future flights are matched according to DOW, and the historical date set should not include dates of holidays and special events.

[0010] Rule 3: Airline operators are not restricted by the above two rules and can manually specify the comparison relationship between future dates and historical dates of specific flights based on their own experience and judgment for flight prediction.

[0011] The above rules 1 and 2 can be automatically matched by computer programs, while rule 3 needs to be set manually.

[0012] The above three rules are widely used in engineering practice, but they have the following problems:

[0013] Problem 1: For the unconventional fluctuations that occur during the sales process and such fluctuations change repeatedly, since Rule 1 uses a relatively fixed date comparison method, it cannot adapt to such fluctuations, resulting in poor prediction results. For example: Assuming that the current date is July 9, 2024, for a flight to be predicted 14 days later, that is, on July 23, 2024, Tuesday, the above Rule 2 is used. The date set of historical flights that can be used for model parameter estimation is A = {July 2, 2024, June 25, 2024, June 18, 2024, June 11, 2024, ...}. When 14 days in advance, the model parameter estimation uses the actual number of bookings b of the flights in set A 14 days in advance. 14 and the number of seats booked before departure b0, if b of all flights in set A 14 The distribution of is not much different from the current (14 days in advance) booking number of the target flight on July 23, 2024, indicating that set A can be used for parameter estimation of the target flight prediction 14 days in advance; if the current date is updated to July 16 in another week, the lead time becomes 7 days, and in the past week, the booking of the target flight has fluctuated greatly from the previous Tuesday. At this time, the distribution of the number of bookings b7 of all flights in set A 7 days in advance is quite different from the current (7 days in advance) booking number of the target flight, then it is inappropriate to estimate the model parameters of the target flight on July 23 with a lead time of 7 days from set A. Further, 3 days later, on July 19, set A may be suitable for the prediction of the flight on July 23 again.

[0014] Question 2: For holidays and special events, Rule 2 is similar to Rule 1 when the booking data fluctuates repeatedly. In addition, the historical data set selected for Rule 2 only has one data per year, and a large amount of historical data needs to be stored, which is not conducive to the rapid operation of the automated program. Therefore, generally only the control data of the last two years is saved, and a simple weighted average is used for prediction, which is not suitable for parameter estimation of conventional models.

[0015] Question 3: If Rule 3 is adopted, the historical comparison date is completely set manually by the flight attendant, which will undoubtedly put tremendous work pressure on the flight attendant. It is impossible to guarantee the timeliness and accuracy of the update of the comparison date set, and it is also difficult to cope with the repeated fluctuations of the flight during the entire pre-sale period.

[0016] Based on the problem of repeated data fluctuations in the above-mentioned prior art, the prediction model cannot accurately predict the booking rate of the target flight on a specified future date based on the repeatedly fluctuating data.

[0017] How to solve the problem that the existing flight prediction model cannot accurately predict the booking rate of the target flight on a specified future date is a technical problem to be solved. Summary of the invention

[0018] Based on this, it is necessary to provide a flight booking prediction method, device, storage medium and electronic device to address the defect that the existing technology cannot accurately predict the booking rate of the target flight on a specified future date.

[0019] In a first aspect, an embodiment of the present invention provides a method for predicting flight booking, the method comprising:

[0020] Obtain a classification method for the target flight, and classify the peak season P, the flat season D, and the off-season O in turn based on the corresponding classification method to obtain a classification result including multiple PDO tags;

[0021] Marking all PDO marks in the classification results into the booking data table of the target flight;

[0022] The data in the booking data table of the target flight are input into the prediction model for parameter estimation, and the prediction result is output. The prediction model is used to predict the booking rate of the target flight on a future specified date, and each data is marked with a corresponding PDO mark.

[0023] Optionally, the method of obtaining a classification method for the target flight, and classifying the peak season P, the flat season D, and the off-season O in sequence based on the corresponding classification method, includes:

[0024] If the target flight is the first target flight, and the first target flight is a historical flight that has departed, a first classification method corresponding to the first target flight is obtained, and the first classification method is used to classify the peak season, the average season, and the off-season in turn.

[0025] Optionally, the adopting the first classification method to classify the peak season, the average season and the off-season in sequence includes:

[0026] Obtain the first peak season passenger load factor lower limit and the first off-season passenger load factor upper limit for the first target flight;

[0027] Determine the time period corresponding to the passenger load factor that is less than the lower limit of the passenger load factor in the first peak season and greater than the upper limit of the passenger load factor in the first off-season as the off-season;

[0028] Acquire a first historical data set within a first preset time period and corresponding to the first target flight;

[0029] Based on the first peak season occupancy rate lower limit and the first off-season occupancy rate upper limit, each first historical data in the first historical data set is classified, and each first historical data is marked with a corresponding first mark in turn, and the first marks include: P mark, D mark, O mark.

[0030] Optionally, after the peak season, the average season and the off-season are classified in sequence by the first classification method, the method further includes:

[0031] According to a preset ratio, the first historical data set is sorted from large to small, and the first historical data set is divided into three corresponding historical data subsets;

[0032] For the data in each of the three historical data subsets, the maximum number, the minimum number and the median corresponding to the peak season, the average season and the off-season are determined in turn to dynamically adjust the PDO parameters.

[0033] Optionally, the method of obtaining a classification method for the target flight, and classifying the peak season P, the flat season D, and the off-season O in sequence based on the corresponding classification method, includes:

[0034] If the target flight is a second target flight, and the second target flight is a future flight, a second classification method corresponding to the second target flight is obtained, and the second classification method is used to classify the peak season, the average season, and the off-season in turn.

[0035] Optionally, the adopting the second classification method to classify the peak season, the average season and the off-season in sequence includes:

[0036] Obtain the lower limit of the passenger load factor in the second peak season and the upper limit of the passenger load factor in the second off-season for the second target flight;

[0037] The time period corresponding to the passenger load factor that is less than the lower limit of the passenger load factor in the second peak season and greater than the upper limit of the passenger load factor in the second off-season is determined as the shoulder season;

[0038] Acquire a second historical data set within a second preset time period and corresponding to the future flight;

[0039] Based on the lower limit of the occupancy rate in the second peak season and the upper limit of the occupancy rate in the second off-season, each second historical data in the second historical data set is classified, and each second historical data is marked with a corresponding second mark in turn, wherein the second marks include: P mark, D mark, and O mark.

[0040] Optionally, after the peak season, the average season and the off-season are classified in sequence by the second classification method, the method further includes:

[0041] If the target flight is the third target flight, the third target flight is the future flight, and the prediction interval corresponding to the future flight is D i-1 -D0, the current PDO parameters are dynamically adjusted according to the preset adjustment method, where i is a positive integer equal to or greater than the preset value N.

[0042] Optionally, dynamically adjusting the current PDO parameters according to a preset adjustment method includes:

[0043] When the PDO parameter corresponding to the current flight is P peak season, if the current booking increment is less than the median of the flat season MiddleD x , the PDO parameter corresponding to the current flight is adjusted to D for the off-season; if the current booking increment is less than the off-season median MiddleO x When the current flight is in the off-season, the PDO parameter corresponding to the current flight is adjusted to O; or,

[0044] When the PDO parameter corresponding to the current flight is D in the off-season, if the current booking increment is greater than the median MiddleP in the peak season x , the PDO parameter corresponding to the current flight is adjusted to P peak season; if the current booking increment is less than the median MiddleO in the off-season x When the current flight is in the off-season, the PDO parameter corresponding to the current flight is adjusted to O; or,

[0045] When the PDO parameter corresponding to the current flight is O in the off-season, if the current booking increment is greater than the median MiddleD in the normal season x , the PDO parameter corresponding to the current flight is adjusted to D for the off-season; if the current booking increment is greater than the median MiddleP for the peak season x When the PDO parameter corresponding to the current flight is adjusted to P peak season.

[0046] In a second aspect, an embodiment of the present invention provides a flight booking prediction device, the device comprising:

[0047] An acquisition module is used to obtain the classification method of the target flight;

[0048] A classification module, used to classify the peak season P, the flat season D and the off-season O in turn based on corresponding classification methods, and obtain classification results including multiple PDO markers;

[0049] A marking module, used for marking all PDO marks in the classification result into the booking data table of the target flight;

[0050] The prediction module is used to input the data in the booking data table of the target flight into the prediction model for parameter estimation and output the prediction result. The prediction model is used to predict the booking rate of the target flight on a specified future date, and each data is marked with a corresponding PDO mark.

[0051] According to a third aspect, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed in a computer, the computer is caused to execute the method of the first aspect.

[0052] According to a fourth aspect, an electronic device is provided, comprising a memory and a processor, wherein the memory stores executable code, and when the processor executes the executable code, the method according to the first aspect is implemented.

[0053] In an embodiment of the present invention, a classification method of a target flight is obtained, and peak season P, flat season D and off-season O are classified in turn based on the corresponding classification method to obtain a classification result including multiple PDO tags; all PDO tags in the classification result are marked in the booking data table of the target flight; and the data in the booking data table of the target flight are input into the prediction model for parameter estimation, and a prediction result is output. The prediction model is used to predict the booking rate of the target flight on a future specified date. The prediction method provided by the embodiment of the present invention, through the prediction method, because each data input into the prediction model is marked with a corresponding PDO tag, it is finally possible to accurately predict the booking rate of the target flight on a future specified date. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] By referring to the following drawings, a more complete understanding of the exemplary embodiments of the present invention can be obtained. The drawings are used to provide a further understanding of the embodiments of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the drawings, the same reference numerals generally represent the same components or steps.

[0055] Figure 1 A flowchart of a method for predicting flight booking according to an exemplary embodiment of the present invention;

[0056] Figure 2 It is a schematic diagram of the structure of a flight booking prediction device 200 provided according to an exemplary embodiment of the present invention. DETAILED DESCRIPTION

[0057] The exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments described herein. On the contrary, these embodiments are provided in order to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0058] It should be noted that, unless otherwise specified, the technical terms or scientific terms used in the present invention should have the common meanings understood by those skilled in the art to which the present invention belongs.

[0059] In addition, the terms "first" and "second" etc. are used to distinguish different objects, rather than to describe a specific order. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, device, product or equipment that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units that are not listed, or may optionally include other steps or units that are inherent to these processes, methods, products or equipment.

[0060] Embodiments of the present invention provide a flight booking prediction method and device, a computer-readable medium, and an electronic device, which are described below in conjunction with the accompanying drawings.

[0061] Please refer to Figure 1 , which shows a flow chart of a flight booking prediction method provided by some embodiments of the present invention, such as Figure 1 As shown, the flight booking prediction method may include the following steps:

[0062] Step S101: obtaining a classification method for a target flight, and classifying peak season P, flat season D and off-season O in sequence based on the corresponding classification method to obtain a classification result including a plurality of PDO tags.

[0063] In one example, a classification method of a target flight is obtained, and peak season P, flat season D, and off-season O are classified in sequence based on the corresponding classification method, including the following steps:

[0064] If the target flight is the first target flight, and the first target flight is a historical flight that has departed, a first classification method corresponding to the first target flight is obtained, and the first classification method is used to classify the peak season, the average season, and the off-season in turn.

[0065] In one example, a first classification method is used to classify peak season, flat season and off-season in sequence, including the following steps:

[0066] Obtain the first peak season passenger load factor lower limit and the first off-season passenger load factor upper limit of the first target flight;

[0067] The time period corresponding to the passenger load factor that is less than the lower limit of the passenger load factor in the first peak season and greater than the upper limit of the passenger load factor in the first off-season is determined as the off-season;

[0068] Acquire a first historical data set within a first preset time period and corresponding to a first target flight;

[0069] Based on the lower limit of the occupancy rate in the first peak season and the upper limit of the occupancy rate in the first off-season, each first historical data in the first historical data set is classified, and each first historical data is marked with a corresponding first mark in turn. The first marks include: P mark, D mark, and O mark. The P mark represents the peak season, the D mark represents the flat season, and the O mark represents the off-season.

[0070] It should be noted that the first preset time period can be set to one year. Here, there are no specific restrictions on the first preset time period, the lower limit of the occupancy rate in the first peak season, and the upper limit of the occupancy rate in the first off-season, and they can be adjusted according to the needs of different application scenarios.

[0071] In the actual application scenario, if the target flight is a historical flight that has departed, the first classification method is used to classify the peak season, the average season and the off-season in turn as follows: sort the occupancy rates of the 365-day flights that have departed before the current input date from high to low, and take the 28th occupancy rate as the lower limit of the peak season occupancy rate. The lower limit of the peak season occupancy rate is LF p Indicates that the 71st passenger load factor is the upper limit of the off-season passenger load factor, and the upper limit of the off-season passenger load factor is LF o ; less than LF p and greater than LF o A load factor of represents the shoulder season.

[0072] According to LF p and LF o All historical data of 365 days are marked with P (representing peak season), D (representing shoulder season), and O (representing off-season), and each corresponding PDO mark is marked on each day of the historical flight.

[0073] It should be noted that the above two division points of 28 and 71 are division values ​​in a specific application scenario. In different application scenarios, the above division points can be adjusted and are not specifically limited here.

[0074] In the classification method for departed flights, event days are not taken into account, and event days and weekdays are calculated together; this classification method groups according to the prediction target, that is: if the flight level is predicted, PDO division is performed according to the flight grouping; if the flight cabin level is predicted, PDO division is performed according to the flight + cabin grouping; if the flight segment level is predicted, PDO division is performed according to the flight segment grouping.

[0075] In one example, after the peak season, the average season, and the off-season are classified in sequence using the first classification method, the flight booking prediction method provided by the embodiment of the present invention may further include the following steps:

[0076] According to the preset ratio, the first historical data set is sorted from large to small, and the first historical data set is divided into three corresponding historical data subsets;

[0077] For the data in each of the three historical data subsets, the maximum number, minimum number and median corresponding to the peak season, the average season and the off-season are determined in turn to dynamically adjust the PDO parameters.

[0078] In a specific application scenario, the preset ratio may be set to 3: 4: 3. In different application scenarios, the preset ratio may be adjusted, which is not specifically limited here.

[0079] In a specific application scenario, if the target flight is a departing flight, the steps of offline calculating the dynamic PDO parameters specifically include:

[0080] Step a1: Preliminary classification by DOW;

[0081] Step a2: Select historical data from the current date 2*365 days ago, select D x -D x+1 The number of cancellations is considered.

[0082] Step a3: Calculate the historical data by adjacent D i The increase in the number of seats booked (D x -D x+1 , consider the number of cancellations, set the current value to D x ) Sort the data from large to small, divide the data into three sections according to the first 30%, 40%, and 30%, and take MaxP for each section of data x 、MinP x 、MiddleP x 、MaxD x 、MinD x 、MiddleD x 、MaxO x 、MinO x 、MiddleO x, stored in the parameter table, where the subscript x indicates that the current data processing point is D x .

[0083] The basic idea of ​​the above method is to divide the historical data into three segments: P, D, and O, and take the maximum number Max, the minimum number Min, and the median Middle for each segment.

[0084] It should be noted that: D x It can express the cumulative number of flight bookings at data processing point x, or it can express data processing point x itself.

[0085] In another example, a classification method of a target flight is obtained, and peak season P, flat season D, and off-season O are classified in sequence based on the corresponding classification method, including:

[0086] If the target flight is the second target flight and the second target flight is a future flight, a second classification method corresponding to the second target flight is obtained, and the second classification method is used to classify the peak season, the average season and the off-season in turn.

[0087] In one example, the second classification method is used to classify the peak season, the shoulder season, and the off-season in sequence, including the following steps:

[0088] Obtain the lower limit of the passenger load factor in the second peak season and the upper limit of the passenger load factor in the second off-season for the second target flight;

[0089] The time period corresponding to the passenger load factor that is less than the lower limit of the passenger load factor in the second peak season and greater than the upper limit of the passenger load factor in the second off-season is determined as the shoulder season;

[0090] Acquire a second historical data set within a second preset time period and corresponding to the future flight;

[0091] Based on the lower limit of the occupancy rate in the second peak season and the upper limit of the occupancy rate in the second off-season, each second historical data in the second historical data set is classified, and each second historical data is marked with a corresponding second mark in turn, where the second marks include: P mark, D mark, and O mark, the P mark represents the peak season, the D mark represents the flat season, and the O mark represents the off-season.

[0092] It should be noted that the second preset time period can be set to two years. Here, there are no specific restrictions on the second preset time period, the lower limit of the second peak season occupancy rate, and the upper limit of the second off-season occupancy rate, and they can be adjusted according to the needs of different application scenarios.

[0093] In the actual scenario, if the target flight is a future flight and the predicted time point corresponding to the future flight is D i, the following classification method is used to classify the peak season, the normal season and the off-season in turn, where i is a positive integer equal to or greater than the preset value N. In a specific application scenario, N can be 8. The details are as follows:

[0094] First, define the date correspondence between future flights and flights of last year and the year before last. See Table 1 (Historical date correspondence method):

[0095]

[0096] Table 1

[0097] For future flights, select the PDO settings and flight load factors of the corresponding flights in the previous two years according to the rules in Table 1. Perform the initial settings of the PDO for future flights according to the method shown in Table 2 (Comparison table of initial settings of PDO for future flights).

[0098]

[0099]

[0100] Table 2

[0101] In one example, after the peak season, the average season, and the off-season are classified in sequence using the second classification method, the flight booking prediction method provided by the embodiment of the present invention may further include the following steps:

[0102] If the target flight is the third target flight, the third target flight is a future flight, and the prediction interval corresponding to the future flight is D i-1 -D0, the current PDO parameters are dynamically adjusted according to the preset adjustment method, where i is a positive integer equal to or greater than the preset value N.

[0103] In a specific application scenario, N may be set to 8, which is just an example. In different application scenarios, N is adjusted, and no specific limitation is made here.

[0104] In one example, dynamically adjusting the current PDO parameters according to a preset adjustment method includes the following steps:

[0105] When the PDO parameter corresponding to the current flight is P peak season, if the current booking increment is less than the median of the flat season MiddleD x , the PDO parameter corresponding to the current flight is adjusted to D for the off-season; if the current booking increment is less than the off-season median MiddleO x When the PDO parameter corresponding to the current flight is adjusted to the off-season O.

[0106] In another example, dynamically adjusting the current PDO parameters according to a preset adjustment method includes the following steps:

[0107] When the PDO parameter corresponding to the current flight is D in the off-season, if the current booking increment is greater than the median MiddleP in the peak season x , the PDO parameter corresponding to the current flight is adjusted to P peak season; if the current booking increment is less than the median MiddleO in the off-season x When the PDO parameter corresponding to the current flight is adjusted to the off-season O;

[0108] In another example, dynamically adjusting the current PDO parameters according to a preset adjustment method includes the following steps:

[0109] When the PDO parameter corresponding to the current flight is O in the off-season, if the current booking increment is greater than the median MiddleD in the normal season x , the PDO parameter corresponding to the current flight is adjusted to D for the off-season; if the current booking increment is greater than the median MiddleP for the peak season x When the PDO parameter corresponding to the current flight is adjusted to P peak season.

[0110] In the actual application scenario, if the target flight is the third target flight, the third target flight is a future flight, and the prediction interval corresponding to the future flight is D i-1 -D0 (for example, N is 8, then D7-D0), the current PDO parameters are dynamically adjusted according to the preset adjustment method.

[0111] If the current flight gets D x After the data is obtained (x is between 7-0), the dynamic adjustment process is as follows:

[0112] The current flight has been set up with PDO early according to the initial setting of PDO for future flights shown in Table 2 before D7.

[0113] If the current PDO of the current flight is P peak season, when D x The booking increment is less than the median value of the shoulder season in the parameter table MiddleD x When the current PDO of the current flight is adjusted from P peak season to D flat season; when D x The increase in bookings is less than the median in the off-season x When the current PDO of the current flight is adjusted from P peak season to O off season.

[0114] If the current PDO of the current flight is D shoulder season, when D x The booking increment is greater than the peak season median MiddleP in the parameter table x When D x The booking increment is less than MiddleOx When the current PDO of the current flight is adjusted from D shoulder season to O off season.

[0115] If the current PDO of the current flight is O off-season, when D x The booking increment is greater than the median value MiddleD in the parameter table during the off-season x When the current PDO of the current flight is adjusted from O off-season to D flat season; x The booking increment is greater than the peak season median MiddleP in the parameter table x When the current PDO of the current flight is adjusted from O low season to P peak season.

[0116] Step S102: All PDO marks in the classification results are marked in the booking data table of the target flight.

[0117] Step S103: Input all data in the booking data table of the target flight into the prediction model for parameter estimation, and output the prediction result. The prediction model is used to predict the booking rate of the target flight on a future specified date, and each data is marked with a corresponding PDO mark.

[0118] In actual application scenarios, the prediction model can be a time series model, and the algorithm used by the time series model can be a Bayesian prediction algorithm or an exponential smoothing algorithm.

[0119] In actual application scenarios, when predicting future flights, the historical data set with the same flight number and PDO+DOW can be selected from HisBook based on the FutBook.PDO and FutBook.DOW fields for parameter estimation of various prediction models. Based on the parameter estimation results, the future flight demand is predicted, and the prediction results are finally stored in the FcstResult table.

[0120] Flight prediction needs to be processed at each data processing point Di. For example, if the current date sysdate is June 1, 2024, the flights on June 2, June 3, June 4, and June 15, 2024 are at D1, D2, D3, and D4 respectively. 14 The PDO processing of historical flights and the estimation of prediction parameters for future flights must be compared with the flight data processing point based on the comparison of DOW+PDO. In actual application scenarios, the specific data processing point can be obtained based on the difference between the flight departure date in the database and the current date sysdate. For the same current date sysdate, multiple different data processing points need to be processed for flights on different dates in the future.

[0121] It should be noted that the database tables of the above database include PDOPra, HisBook, FutBook, and FcstResult (prediction result table). The structures of the above three tables are explained as follows. In specific implementation, they can be appropriately cut based on the different designs of data processing points.

[0122] The following are examples for PDOPara (dynamic PDO setting parameter table), HisBook (historical flight booking data table), and FutBook (future flight booking data table):

[0123] In one example, the exemplary table structure of the dynamic PDO setting parameter table PDOPara for storing offline calculated dynamic PDO parameters is shown in Table 3 below:

[0124]

[0125]

[0126] Table 3

[0127] In one example, an exemplary table structure of the historical flight booking data table HisBook is shown in Table 4 below:

[0128]

[0129]

[0130] Table 4

[0131] In one example, the exemplary table structure of the future flight booking data table FutBook is as follows:

[0132] The structure of this table is the same as HisBook, but it only stores future flight data. For example, if a flight has already left the airport, the flight data can be processed at 0:00 every day and migrated from this table to the HisBook table.

[0133] The flight booking prediction method provided by the embodiment of the present invention can accurately predict the booking rate of the target flight on a specified future date, because each data in the input prediction model is marked with a corresponding PDO mark.

[0134] In the above embodiment, a flight booking prediction method is provided, and correspondingly, the present invention also provides a flight booking prediction device. The flight booking prediction device provided in the embodiment of the present invention can implement the above flight booking prediction method, and the flight booking prediction device can be implemented by software, hardware, or a combination of software and hardware. For example, the flight booking prediction device can include integrated or separate functional modules or units to perform the corresponding steps in the above methods.

[0135] Please refer to Figure 2 , which shows a schematic diagram of a flight booking prediction device provided by some embodiments of the present invention. Since the device embodiment is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment. The device embodiment described below is only illustrative.

[0136] like Figure 2 As shown, the flight booking prediction device 200 may include:

[0137] An acquisition module 201 is used to acquire a classification method of a target flight;

[0138] A classification module 202, for classifying the peak season P, the flat season D and the off-season O in sequence based on corresponding classification methods, and obtaining classification results including a plurality of PDO markers;

[0139] A marking module 203, used to mark all PDO marks in the classification result into the booking data table of the target flight;

[0140] The prediction module 204 is used to input each data in the booking data table of the target flight into the prediction model for parameter estimation and output the prediction result. The prediction model is used to predict the booking rate of the target flight on a specified future date. Each data is marked with a corresponding PDO mark.

[0141] In some implementations of the embodiments of the present invention, the classification module 202 is used to:

[0142] If the target flight is the first target flight, and the first target flight is a historical flight that has departed, a first classification method corresponding to the first target flight is obtained, and the first classification method is used to classify the peak season, the average season, and the off-season in turn.

[0143] In some implementations of the embodiments of the present invention, the classification module 202 is specifically used to:

[0144] Obtain the first peak season passenger load factor lower limit and the first off-season passenger load factor upper limit of the first target flight;

[0145] The time period corresponding to the passenger load factor that is less than the lower limit of the passenger load factor in the first peak season and greater than the upper limit of the passenger load factor in the first off-season is determined as the off-season;

[0146] Acquire a first historical data set within a first preset time period and corresponding to a first target flight;

[0147] Based on the lower limit of the occupancy rate in the first peak season and the upper limit of the occupancy rate in the first off-season, each first historical data in the first historical data set is classified, and each first historical data is marked with a corresponding first mark in turn, and the first marks include: P mark, D mark, and O mark.

[0148] In some implementations of the embodiments of the present invention, the flight booking prediction device 200 may further include:

[0149] The first adjustment module (in Figure 2 ), after classifying the peak season, the normal season and the off-season in sequence by the first classification method, sorting the first historical data set from large to small according to a preset ratio, and sequentially dividing the first historical data set into three corresponding historical data subsets; and determining the maximum number, the minimum number and the median corresponding to the peak season, the normal season and the off-season in sequence for the data in each of the three historical data subsets, so as to dynamically adjust the PDO parameters.

[0150] In some implementations of the embodiments of the present invention, the classification module 202 is used to:

[0151] If the target flight is the second target flight and the second target flight is a future flight, a second classification method corresponding to the second target flight is obtained, and the second classification method is used to classify the peak season, the average season and the off-season in turn.

[0152] In some implementations of the embodiments of the present invention, the classification module 202 is specifically used to:

[0153] Obtain the lower limit of the passenger load factor in the second peak season and the upper limit of the passenger load factor in the second off-season for the second target flight;

[0154] The time period corresponding to the passenger load factor that is less than the lower limit of the passenger load factor in the second peak season and greater than the upper limit of the passenger load factor in the second off-season is determined as the shoulder season;

[0155] Acquire a second historical data set within a second preset time period and corresponding to the future flight;

[0156] Based on the lower limit of the occupancy rate in the second peak season and the upper limit of the occupancy rate in the second off-season, each second historical data in the second historical data set is classified, and each second historical data is marked with a corresponding second mark in turn, where the second marks include: P mark, D mark, and O mark.

[0157] In some implementations of the embodiments of the present invention, the flight booking prediction device 200 may further include:

[0158] The second adjustment module (in Figure 2 ), after the peak season, the flat season and the off-season are classified in sequence by the second classification method, if the target flight is the third target flight, the third target flight is a future flight, and the prediction interval corresponding to the future flight is D i-1 -D0, the current PDO parameters are dynamically adjusted according to the preset adjustment method, where i is a positive integer equal to or greater than the preset value N.

[0159] In some implementations of the embodiments of the present invention, the second adjustment module is specifically used to:

[0160] When the PDO parameter corresponding to the current flight is P peak season, if the current booking increment is less than the median of the flat season MiddleD x , the PDO parameter corresponding to the current flight is adjusted to D for the off-season; if the current booking increment is less than the off-season median MiddleO x When the current flight is in the off-season, the PDO parameter corresponding to the current flight is adjusted to O; or,

[0161] When the PDO parameter corresponding to the current flight is D in the off-season, if the current booking increment is greater than the median MiddleP in the peak season x , the PDO parameter corresponding to the current flight is adjusted to P peak season; if the current booking increment is less than the median MiddleO in the off-season x When the current flight is in the off-season, the PDO parameter corresponding to the current flight is adjusted to O; or,

[0162] When the PDO parameter corresponding to the current flight is O in the off-season, if the current booking increment is greater than the median MiddleD in the normal season x , the PDO parameter corresponding to the current flight is adjusted to D for the off-season; if the current booking increment is greater than the median MiddleP for the peak season x When the PDO parameter corresponding to the current flight is adjusted to P peak season.

[0163] In some implementations of the embodiments of the present invention, the flight booking prediction device 200 provided by the embodiments of the present invention is based on the same inventive concept as the flight booking prediction method provided by the aforementioned embodiments of the present invention and has the same beneficial effects.

[0164] According to another embodiment, there is also provided a computer-readable storage medium having a computer program stored thereon, which, when executed in a computer, causes the computer to execute a combination of Figure 1 The method described.

[0165] According to another embodiment of the present invention, there is also provided an electronic device, comprising a memory and a processor, wherein the memory stores an executable code, and when the processor executes the executable code, the Figure 1 The method described.

[0166] Those skilled in the art should be aware that in one or more of the above examples, the functions described in the present invention can be implemented by hardware, software, firmware or any combination thereof. When implemented by software, these functions can be stored in a computer-readable medium or transmitted as one or more instructions or codes on a computer-readable medium.

[0167] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made on the basis of the technical solution of the present invention should be included in the scope of protection of the present invention.

Claims

1. A flight booking prediction method, characterized in that: The method comprises: Obtain a classification method for the target flight, and classify the peak season P, the flat season D, and the off-season O in turn based on the corresponding classification method to obtain a classification result including multiple PDO tags; Marking all PDO marks in the classification results into the booking data table of the target flight; Each data in the booking data table of the target flight is input into the prediction model for parameter estimation, and a prediction result is output. The prediction model is used to predict the booking rate of the target flight on a future specified date, and each data is marked with a corresponding PDO mark.

2. The prediction method according to claim 1, characterized in that: The method of obtaining the classification of the target flights, and classifying the peak season P, the flat season D and the off-season O in turn based on the corresponding classification method, includes: If the target flight is the first target flight, and the first target flight is a historical flight that has departed, a first classification method corresponding to the first target flight is obtained, and the first classification method is used to classify the peak season, the average season, and the off-season in turn.

3. The prediction method according to claim 2, characterized in that: The first classification method is used to classify the peak season, the average season and the off-season in sequence, including: Obtain the first peak season passenger load factor lower limit and the first off-season passenger load factor upper limit for the first target flight; Determine the time period corresponding to the passenger load factor that is less than the lower limit of the passenger load factor in the first peak season and greater than the upper limit of the passenger load factor in the first off-season as the off-season; Acquire a first historical data set within a first preset time period and corresponding to the first target flight; Based on the first peak season occupancy rate lower limit and the first off-season occupancy rate upper limit, each first historical data in the first historical data set is classified, and each first historical data is marked with a corresponding first mark in turn, and the first marks include: P mark, D mark, O mark.

4. The prediction method according to claim 3, characterized in that: After the peak season, the average season and the off-season are classified in sequence by the first classification method, the method further includes: According to a preset ratio, the first historical data set is sorted from large to small, and the first historical data set is divided into three corresponding historical data subsets; For the data in each of the three historical data subsets, the maximum number, minimum number and median corresponding to the peak season, the average season and the off-season are determined in turn to dynamically adjust the PDO parameters.

5. The prediction method according to claim 1, characterized in that: The method of obtaining the classification of the target flights, and classifying the peak season P, the flat season D and the off-season O in turn based on the corresponding classification method, includes: If the target flight is a second target flight, and the second target flight is a future flight, a second classification method corresponding to the second target flight is obtained, and the second classification method is used to classify the peak season, the average season, and the off-season in turn.

6. The prediction method according to claim 5, characterized in that: The second classification method is used to classify the peak season, the average season and the off-season in sequence, including: Obtain the lower limit of the passenger load factor in the second peak season and the upper limit of the passenger load factor in the second off-season for the second target flight; The time period corresponding to the passenger load factor that is less than the lower limit of the passenger load factor in the second peak season and greater than the upper limit of the passenger load factor in the second off-season is determined as the shoulder season; Acquire a second historical data set within a second preset time period and corresponding to the future flight; Based on the lower limit of the occupancy rate in the second peak season and the upper limit of the occupancy rate in the second off-season, each second historical data in the second historical data set is classified, and each second historical data is marked with a corresponding second mark in turn, wherein the second marks include: P mark, D mark, and O mark.

7. The prediction method according to claim 5, characterized in that: After the peak season, the average season and the off-season are classified in sequence by the second classification method, the method further includes: If the target flight is the third target flight, the third target flight is the future flight, and the prediction interval corresponding to the future flight is D i-1 -D0, the current PDO parameters are dynamically adjusted according to the preset adjustment method, where i is a positive integer equal to or greater than the preset value N.

8. The prediction method according to claim 7, characterized in that: The dynamically adjusting the current PDO parameters according to the preset adjustment method includes: When the PDO parameter corresponding to the current flight is P peak season, if the current booking increment is less than the median of the flat season MiddleD x , the PDO parameter corresponding to the current flight is adjusted to D for the off-season; if the current booking increment is less than the off-season median MiddleO x When the current flight is in the off-season, the PDO parameter corresponding to the current flight is adjusted to O; or, When the PDO parameter corresponding to the current flight is D in the off-season, if the current booking increment is greater than the median MiddleP in the peak season x , the PDO parameter corresponding to the current flight is adjusted to P peak season; if the current booking increment is less than the median MiddleO in the off-season x When the current flight is in the off-season, the PDO parameter corresponding to the current flight is adjusted to O; or, When the PDO parameter corresponding to the current flight is O in the off-season, if the current booking increment is greater than the median MiddleD in the normal season x , the PDO parameter corresponding to the current flight is adjusted to D for the off-season; if the current booking increment is greater than the median MiddleP for the peak season x When the PDO parameter corresponding to the current flight is adjusted to P peak season.

9. A flight booking prediction device, characterized in that: The device comprises: An acquisition module is used to obtain the classification method of the target flight; A classification module, used to classify the peak season P, the flat season D and the off-season O in turn based on corresponding classification methods, and obtain classification results including multiple PDO markers; A marking module, used for marking all PDO marks in the classification result into the booking data table of the target flight; The prediction module is used to input the data in the booking data table of the target flight into the prediction model for parameter estimation and output the prediction result. The prediction model is used to predict the booking rate of the target flight on a specified future date, and each data is marked with a corresponding PDO mark.

10. A computer-readable storage medium, characterized in that: A computer program is stored thereon, and when the computer program is executed in a computer, the computer is caused to execute the method according to any one of claims 1 to 8.

11. An electronic device, characterized in that: The method comprises a memory and a processor, wherein the memory stores executable codes, and when the processor executes the executable codes, the method according to any one of claims 1 to 8 is implemented.

Images