Prediction model training method and device, storage medium and electronic device

Abstract

The present invention provides a training method and device for a prediction model, a storage medium and an electronic device, the method comprising: in response to a model training instruction, obtaining the original booking sample of each flight collected at each data collection time point; determining the preprocessing parameters of each data collection time point according to each original booking sample; calculating the booking demand data of each flight at each data collection time point based on the preprocessing parameters of each data collection time point and the original booking sample of each flight collected at each data collection time point; and using each booking demand data to train a pre-built prediction model. By using the method provided by the present invention, the original booking sample can be processed by the preprocessing parameter to remove the data bias of the original booking sample, so that the preprocessed data can accurately reflect the user's booking demand, thereby greatly improving the prediction accuracy of the prediction model.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a prediction model training method and device, a storage medium and an electronic device. Background Art

[0002] With the development of science and technology, air transportation has gradually become one of the main modes of transportation. As the number of flights increases, the difficulty of flight seat resource management is also increasing.

[0003] At present, in order to achieve refined management of flight seat resources, airlines usually build prediction models based on historical booking data and use the prediction models to make flight booking predictions.

[0004] However, historical booking data is usually affected by manual management of flight administrators and seasonal fluctuations, and has serious data bias. Therefore, directly building a prediction model based on historical booking data will result in low prediction accuracy of the prediction model. Summary of the invention

[0005] The technical problem to be solved by the present invention is to provide a prediction model training method that can improve the prediction accuracy of the prediction model.

[0006] The present invention also provides a prediction model training device to ensure the implementation and application of the above method in practice.

[0007] A prediction model training method, comprising:

[0008] In response to the model training instruction, obtaining the original booking sample of each flight collected in advance at each data collection time point;

[0009] Determining the preprocessing parameters of each data collection time point according to each original booking sample;

[0010] Based on the preprocessing parameters at each data collection time point and the original booking samples of each flight collected at each data collection time point, the booking demand data of each flight at each data collection time point is calculated;

[0011] The pre-built prediction model is trained using each of the reservation demand data.

[0012] A prediction model training device, comprising:

[0013] An acquisition unit, configured to acquire, in response to a model training instruction, an original booking sample of each flight collected in advance at each data collection time point;

[0014] A determination unit, configured to determine a preprocessing parameter for each of the data collection time points according to each of the original reservation samples;

[0015] A calculation unit, configured to calculate the seat booking demand data of each flight at each data collection time point based on the preprocessing parameters at each data collection time point and the original seat booking samples of each flight collected at each data collection time point;

[0016] A training unit is used to apply each of the reservation demand data to train a pre-built prediction model.

[0017] A storage medium includes storage instructions, wherein when the instructions are executed, the device where the storage medium is located is controlled to execute the training method of the prediction model as described above.

[0018] An electronic device includes a memory and one or more instructions, wherein the one or more instructions are stored in the memory and are configured to be executed by one or more processors to perform the above-mentioned prediction model training method.

[0019] Compared with the prior art, the present invention has the following advantages:

[0020] The present invention provides a training method and device for a prediction model, a storage medium and an electronic device, the method comprising: in response to a model training instruction, obtaining the original booking sample of each flight collected in advance at each data collection time point; determining the preprocessing parameters of each data collection time point according to each original booking sample; calculating the booking demand data of each flight at each data collection time point based on the preprocessing parameters of each data collection time point and the original booking sample of each flight collected at each data collection time point; and using each booking demand data to train a pre-constructed prediction model. By using the method provided by the present invention, the original booking sample can be processed by preprocessing parameters to remove the data bias of the original booking sample, so that the preprocessed data can accurately reflect the user's booking demand. Therefore, the booking data demand data training model obtained by processing the original booking sample with preprocessing parameters can greatly improve the prediction accuracy of the prediction model. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0022] Figure 1 A method flow chart of a prediction model training method provided by the present invention;

[0023] Figure 2 A flow chart of a process for determining preprocessing parameters at each data collection time point provided by the present invention;

[0024] Figure 3 A flowchart of a process for calculating and obtaining the seat reservation demand data of each flight at each data collection time point provided by the present invention;

[0025] Figure 4 A flowchart of another process of determining preprocessing parameters for each data collection time point provided by the present invention;

[0026] Figure 5 A flowchart of another process of calculating and obtaining the seat reservation demand data of each flight at each data collection time point provided by the present invention;

[0027] Figure 6 A schematic diagram of the structure of a prediction model training device provided by the present invention;

[0028] Figure 7 The present invention provides a schematic structural diagram of an electronic device. DETAILED DESCRIPTION

[0029] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0030] In this application, the terms "comprises", "comprising" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprising a ..." does not exclude the presence of other identical elements in the process, method, article or device comprising the element.

[0031] The present invention can be used in many general or special computing device environments or configurations, such as personal computers, server computers, handheld or portable devices, tablet devices, multi-processor devices, distributed computing environments including any of the above devices or devices, etc.

[0032] The embodiment of the present invention provides a prediction model training method, which can be applied to a variety of system platforms, and its execution subject can be a processor of an electronic device. The method flow chart of the method is as follows: Figure 1 As shown, specifically including:

[0033] S101: In response to a model training instruction, an original booking sample of each flight collected in advance at each data collection time point is obtained.

[0034] In an embodiment of the present invention, the data collection time point DCP is a time point based on the departure time of the flight. According to the departure time of each flight, multiple data collection time points can be set before the flight departs. When each data collection time point is reached, flight data in a specified data format can be obtained. The flight data may include reservation data, which can be used as an original reservation sample.

[0035] The data collection time point includes a fixed data collection time point or a floating data collection time point.

[0036] Specifically, in an embodiment of the present invention, it is also possible to determine whether the original booking samples of each flight collected at each data collection time point meet the application conditions; if the data application conditions are not met, the original booking samples of the flight can be eliminated; if the data application conditions are met, the subsequent steps are executed through the original booking samples that meet the data application conditions.

[0037] S102: Determine the preprocessing parameters of each data collection time point according to each original booking sample.

[0038] In an embodiment of the present invention, the preprocessing parameter at the data collection time point may be a de-restriction parameter, which may be a parameter calculated for the original booking sample based on the Baseline algorithm, or a parameter calculated for the original booking sample based on the EM algorithm.

[0039] S103: Based on the preprocessing parameters at each data collection time point and the original booking samples of each flight collected at each data collection time point, the booking demand data of each flight at each data collection time point is calculated.

[0040] In an embodiment of the present invention, a pre-set de-restriction algorithm can be applied to calculate the booking demand data of each flight at each data collection time point based on the preprocessing parameters at each data collection time point and the original booking samples of each flight collected at each data collection time point.

[0041] S104: Using each of the booking demand data to train a pre-built prediction model.

[0042] By applying the method provided by the present invention, the original booking samples can be processed through preprocessing parameters to remove the data bias of the original booking samples, so that the preprocessed data can accurately reflect the user's booking needs. Therefore, the booking data demand data training model obtained by processing the original booking samples with preprocessing parameters can greatly improve the prediction accuracy of the prediction model.

[0043] In the method provided in the embodiment of the present invention, based on the above implementation process, specifically, it also includes:

[0044] Obtaining current flight information of the flight to be predicted; the flight information includes booking data;

[0045] The prediction model is applied to process the flight information to obtain a booking prediction result of the flight to be predicted in a future time period.

[0046] In this embodiment, after obtaining the booking prediction result, the seat resources of the flight can be managed according to the booking prediction result. For example, a sales strategy for the flight seats can be set.

[0047] In the method provided in the embodiment of the present invention, based on the above implementation process, specifically, a feasible way of determining the preprocessing parameters of each data collection time point according to each original booking sample is as follows: Figure 2 As shown, specifically including:

[0048] S201: Determine the sample quantity of the original booking samples collected at each data collection time point; and determine the total booking quantity corresponding to the data collection time point based on the original booking samples collected at each data collection time point.

[0049] In the embodiment of the present invention, the total booking amount corresponding to each data collection time point can be obtained from the original booking samples of each flight at the data collection time point and the total booking amount at the previous data collection time point, as follows:

[0050]

[0051] Among them, Sum j is the total number of reservations at the jth data collection time point, Sum j-1 is the total number of bookings at the j-1th data collection time point, k can be the number of flights at each data collection time point, booking i,j It can be the number of bookings of the original booking sample of the i-th flight at the j-th data collection time point, j≥2.

[0052] Specifically, the total number of reservations at the first data collection time point can be:

[0053]

[0054] The number of data collection time points may be multiple, for example, 20.

[0055] S202: For each of the data collection time points, if the data collection time point is the first data collection time point, the number of samples at the data collection time point and the total number of bookings corresponding to the data collection time point are used as preprocessing parameters for the data collection time point; if the data collection time point is not the first data collection time point, the number of samples at the data collection time point, the total number of bookings corresponding to the data collection time point, and the total number of bookings at the previous data collection time point are used as preprocessing parameters for the data collection time point.

[0056] In the embodiment of the present invention, it is also possible to determine whether the preprocessing parameters at each data collection time point meet the application conditions. If the number of original booking samples at the data collection time point is less than a preset sample threshold, it can be determined that the preprocessing parameters at the data collection time point do not meet the application conditions. Otherwise, it can be determined that the preprocessing parameters at the data collection time point meet the application conditions.

[0057] Among them, when the preprocessing parameters at the data collection time point do not meet the application conditions, the preprocessing parameters at the data collection time point can be discarded. When the preprocessing parameters at the data collection time point meet the application conditions, the preprocessing parameters at the data collection time point can be applied to calculate the reservation demand data at the data collection time point.

[0058] In the embodiment of the present invention, based on the preprocessing parameters at each data collection time point in S202 above, correspondingly, based on the preprocessing parameters at each data collection time point and the original booking samples of each flight collected at each data collection time point, a feasible method for calculating the booking demand data of each flight at each data collection time point is as follows: Figure 3 As shown, it may include:

[0059] S301: Calculate a collection value at each data collection time point according to the preprocessing parameters at each data collection time point; the collection value represents the average increase in reservations at the data collection time point relative to the previous data collection time point.

[0060] Specifically, the calculation method of the collected value at the data collection time point is as follows:

[0061]

[0062]

[0063] Among them, hisPj is the data collection value at the data collection time point j, C j is the number of original booking samples for each flight at the data collection time point.

[0064] S302: Determine the increment of the number of bookings for each flight at each data collection time point based on the original booking samples of each flight collected at each data collection time point.

[0065] The incremental number of bookings is the increase in the number of bookings for the flight at the data collection time point relative to the previous data collection time point.

[0066] In the embodiment of the present invention, the calculation method of the seat booking increment of each flight at each data collection time point is as follows:

[0067] P i,j =booking i,j ,j=1;

[0068] P i,j =booking i,j -booking i,j-1 ,j≥2.

[0069] Among them, P i,j is the increase in the number of bookings for flight i at the data collection time point j, booking i,j is the original booking sample of flight i at data collection time point j.

[0070] S303: Calculate the seat demand data of each flight at each data collection time point according to the collection value at each data collection time point and the seat increment of each flight at each data collection time point.

[0071] In the embodiment of the present invention, the first reference value can be calculated by the collected value. If the collected value hisP j is greater than or equal to 0, then the first reference value hisC is determined ij =hisP j , otherwise, the first reference value hisC ij The calculation method is as follows:

[0072]

[0073] Among them, UnBooking i,j-1 is the booking demand data of flight i at the previous data collection time point.

[0074] Specifically, the second reference value can be determined by the increment of the number of seats booked. i,jis greater than or equal to 0, then determine the second reference value actC ij =P i,j Otherwise, the second reference value actC ij The calculation method is as follows:

[0075]

[0076] Among them, for each data collection time point, if the number of bookings at the data collection time point and the previous data collection time point are both less than the preset booking number threshold; or the number of bookings at the data collection time point is less than the booking number threshold, and the booking number increment is less than 0, then the booking demand data UnBooking of each flight at the data collection time point i,j Can be:

[0077] UnBooking i,j =UnBooking i,j-1 +actC ij ;

[0078] Otherwise, it is:

[0079] UnBooking i,j =UnBooking i,j-1 +max(hisC ij +actC ij ).

[0080] In one embodiment provided by the present invention, a feasible method for preprocessing the original booking sample at the floating data collection time point is also provided when there is a floating data collection time point between two fixed data collection time points, which is specifically as follows:

[0081] The previous fixed data collection time point of the floating data collection time point floatDcp is DCP(j-1), and the next fixed data collection time point of the floating data collection time point is DCP(j).

[0082] Calculate the ratio of the number of days from DCP(j-1) to floatDcp and from DCP(j-1) to DCP(j) as follows:

[0083] ratio=(dPrior[j-1]-dPrior[floatDcp]) / (dPrior[j-1]-dPrior[j]).

[0084] Calculate the incremental number of seats P for each flight at the data collection time point DCP(j) i,j and collected values.

[0085] The first reference value is calculated by the collected value. If the collected value hisP j is greater than or equal to 0, then the first reference value hisC is determined i floatDcp =ratio*hisP j , otherwise, the first reference value hisC i floatDcp The calculation method is as follows:

[0086]

[0087] The second reference value can be determined by the increment of the number of seats booked. If the increment of the number of seats booked is P i,j is greater than or equal to 0, then determine the second reference value actC i floaDcp =P i,j Otherwise, the second reference value actC i floatDcp The calculation method is as follows:

[0088]

[0089] If the number of seats booked at the floating data collection time point and the previous data collection time point are both less than the preset booking number threshold; or the number of seats booked at the data collection time point is less than the booking number threshold, and the booking number increment is less than 0, then the booking demand data UnBooking of each flight at the data collection time point is i,floatDcp Can be:

[0090] UnBooking i,floatDcp =UnBooking i,j-1 +actC i floatDcp ;

[0091] Otherwise, it is:

[0092] UnBooking i,floatDcp =UnBooking i,j-1 +max(hisC ij +actC i floatDcp ).

[0093] In the embodiment of the present invention, based on the above implementation process, specifically, a feasible way to determine the preprocessing parameters of each data collection time point according to each original booking sample is as follows: Figure 4 As shown, it may include:

[0094] S401: Determine the increment of the number of bookings for each flight at the data collection time point according to the original booking samples of each flight collected at the data collection time point.

[0095] In the embodiment of the present invention, the method of determining the increment of the number of seats booked for each flight at the data collection time point j is as follows:

[0096] Pi,j =booking i,j ,j=1;

[0097] P i,j =booking i,j -booking i,j-1 ,j≥2.

[0098] Among them, P i,j is the increase in the number of bookings for flight i at the data collection time point j, booking i,j is the original booking sample of flight i at data collection time point j.

[0099] S402: According to the increment of the number of seat reservations of each flight at the data collection time point, the sum of the increment of the number of seat reservations of each flight at the data collection time point and the square of the sum of the increment of the number of seat reservations are calculated.

[0100] In the embodiment of the present invention, the method for calculating the sum of the incremental number of seats booked for each flight at the data collection time point is as follows:

[0101]

[0102] Among them, Op j It is the sum of the incremental number of bookings for each flight at the time of data collection.

[0103] In the embodiment of the present invention, the method of calculating the square of the sum of the incremental number of seat reservations for each flight at the data collection time point is as follows:

[0104]

[0105] Among them, SQp j It is the square of the sum of the incremental number of bookings for each flight at the time of data collection.

[0106] S403: Calculate the mean and root variance of the sum of the incremental booking numbers based on the number of original booking samples of each flight collected at the data collection time point, the sum of the incremental booking numbers, and the square of the sum of the incremental booking numbers.

[0107] In the embodiment of the present invention, the mean of the sum of the increments of the number of seat reservations is calculated as follows:

[0108]

[0109] Among them, Mean j is the mean of the sum of the increments of the number of seats booked, Oc j is the number of original booking samples of each flight at the data collection time point j.

[0110] In the embodiment of the present invention, the root variance of the sum of the incremental number of seat reservations is calculated as follows:

[0111]

[0112] Among them, error j is the root variance of the sum of the increments in the number of seat reservations.

[0113] S404: Determine candidate seat number increments from each seat number increment according to the mean and the root variance.

[0114] In the embodiment of the present invention, each seat booking increment P can be determined i,j With Mean j The absolute value of the difference; and according to the root variance error j Determine the threshold value; determine whether each absolute value is greater than the corresponding threshold value, and determine the seat number increment to which the absolute value that is not greater than the corresponding threshold value belongs as the candidate seat number increment.

[0115] Specifically, the increase in the number of alternative bookings must meet the following conditions:

[0116] |P i,j -Mean j |<error j *R.

[0117] Among them, error j *R is the error threshold at data acquisition time point j, and R is the error multiplier.

[0118] S405: Calculate a first mean and a first root variance of the sum of the booking number increments corresponding to each of the alternative booking number increments based on the number of original booking samples of the flights to which each of the alternative booking number increments belongs, the sum of the booking number increments corresponding to each of the alternative booking number increments, and the square of the sum of the booking number increments corresponding to each of the alternative booking number increments.

[0119] In the embodiment of the present invention, the calculation method of the first mean and the first root variance may refer to the calculation method of the mean and the root variance in S403, which will not be described in detail here.

[0120] S406: Determine the first mean as a target mean, and determine the first root variance as a target root variance.

[0121] S407: Iterate the first operation until the number of iterations is greater than a preset number threshold, or the target mean value calculated by the current execution of the first operation satisfies a preset stop iteration condition; the first operation includes: determining a target seat increment in each seat increment according to the target mean value and the target root variance currently calculated; calculating the mean expectation and the variance expectation according to each target seat increment, the first mean value and the first root variance; determining the flight to which each target seat increment belongs as a target flight; determining the target flight according to the sum of the seat increments of each target flight and the mean value. According to the expectation, the sum of the new increments of the number of bookings for each of the target flights is calculated; according to the square of the sum of the increments of the number of bookings for each of the target flights and the variance expectation, the square of the sum of the new increments of the number of bookings for each of the target flights is calculated; according to the number of original booking samples of each of the target flights, the sum of the new increments of the number of bookings for each of the target flights, and the square of the sum of the new increments of the number of bookings for each of the target flights, the second mean and the second root variance of the sum of the increments of the number of bookings corresponding to each of the target increments of the number of bookings are calculated; the second mean is determined as the new target mean, and the second root variance is determined as the new target root variance.

[0122] In an embodiment of the present invention, the number threshold can be set according to actual needs, for example, it can be 10 times, and the condition for stopping iteration can be that the difference between the current target mean and the target mean calculated last time is less than or equal to the difference threshold, and the difference threshold can be 0.01.

[0123] Among them, a feasible way to calculate the expected mean and expected variance according to each of the target seat booking increments, the first mean and the first root variance is: substitute the target seat booking increment, the first mean and the first root variance into the expectation formula of the EM algorithm to obtain the expected mean NewP j And the expected variance NewSQp j .

[0124] Optionally, according to the sum of the incremental seat number of each target flight and the expected mean, the method for calculating the sum of the incremental seat number of each target flight is as follows:

[0125] NewP j =Op+Pc*expP

[0126] Among them, NewP j is the sum of the new booking increments, Pc is the number of original booking samples of each current target flight, and expP is the expected mean.

[0127] Specifically, according to the square of the sum of the incremental number of seat reservations of each target flight and the expected variance, the square of the sum of the incremental number of seat reservations of each target flight is calculated as follows:

[0128] QqI j =SQp j +Pc*expSQP

[0129] Among them, NewSQp j is the square of the sum of the new increments in the number of seat reservations, and expSQP is the expected variance.

[0130] In the embodiment of the present invention, the method for calculating the second mean and the second root variance of the sum of the booking number increments corresponding to each target booking number increment is based on the number of original booking samples of each target flight, the sum of the new booking number increments of each target flight, and the square of the sum of the new booking number increments of each target flight. The method for calculating the mean and the root variance in S403 can be referred to, and will not be repeated here.

[0131] S408: After stopping the iteration of the first operation, the current target mean and target root variance are used as preprocessing parameters at the data collection time point.

[0132] In the embodiment of the present invention, based on the preprocessing parameters of each data collection time point obtained in S408, correspondingly, based on the preprocessing parameters of each data collection time point and the original booking samples of each flight collected at each data collection time point, the booking demand data of each flight at each data collection time point is calculated, such as Figure 5 As shown, including:

[0133] S501: Calculate the expected increase in the number of bookings for each flight at each data collection time point according to the increase in the number of bookings for each flight at each data collection time point and the preprocessing parameters at each data collection time point.

[0134] The P i,j and target mean TMean j and target variance Terror j Substitute the expected formula of the EM algorithm to obtain the expected increase in the number of bookings EXPT ij .

[0135] S502: Based on the original booking samples of each flight collected at each data collection time point and the expected incremental number of bookings of each flight at each data collection time point, the booking number demand data of each flight at each data collection time point is calculated.

[0136] In the embodiment of the present invention, for each flight i at the data collection time point j, if the number of original booking samples Oc at the data collection time point j is j The sample quantity threshold is greater than the preset value, the target mean is greater than 0, and the booking increment P i,j Less than or equal to the error threshold Terror j *R, then determine the third reference value HC ij =EXPT ij Otherwise, calculate the first reference value hisC ij , the first reference value hisC ij as the third reference value.

[0137] Optionally, a second reference value actC may also be calculated ij , wherein the calculation method of the first reference value and the second reference value can refer to the above-mentioned S303 part, which will not be repeated here.

[0138] In the embodiment of the present invention, for each data collection time point, if the number of bookings at the data collection time point and the previous data collection time point are both less than the preset booking number threshold; or the number of bookings at the data collection time point is less than the booking number threshold, and the booking number increment is less than 0, then the booking demand data UnBooking of each flight at the data collection time point is i,j Can be:

[0139] UnBooking i,j =UnBooking i,j-1 +actC ij ;

[0140] Otherwise, it is:

[0141] UnBooking i,j =UnBooking i,j-1 +max(HC ij +actC ij ).

[0142] In one embodiment provided by the present invention, another feasible way of preprocessing the original booking sample at the floating data collection time point is provided when there is a floating data collection time point between two fixed data collection time points, which is specifically as follows:

[0143] The previous fixed data collection time point of the floating data collection time point floatDcp is DCP(j-1), and the next fixed data collection time point of the floating data collection time point is DCP(j).

[0144] If the number of original booking samples at the data collection time point j is Oc jThe sample quantity threshold is greater than the preset value, the target mean is greater than 0, and the booking increment P i,j Less than or equal to the error threshold Terror j *R, then determine the third reference value HC ifloatDcp =ratio*EXPT ij Otherwise, calculate the first reference value hisC ij , the first reference value hisC ij As the third reference value, UnBooking is calculated i,floatDcp The calculation principle is the same as that of UnBooking in S502. i,j The principle is the same as that of , which can be referred to and will not be repeated here.

[0145] In an embodiment provided by the present invention, a feasible method for preprocessing the original reservation sample at the last data collection time point is also provided, which is as follows:

[0146] When the data collection time point j is the last data collection time point, if booking i,j >0, then the second ratio Otherwise, showRate = 1. The last data collection time point may be a boarding collection time point, that is, a data collection point on the flight departure date.

[0147] If showRate>0.70, or booking i,j >5, then

[0148] UnBooking i,j =UnBooking i,j-1 +actC ij ;otherwise

[0149] UnBooking i,j =UnBooking i,j-1 +max(HC ij +actC ij ).

[0150] and Figure 1 Corresponding to the method described above, the embodiment of the present invention also provides a prediction model training device for Figure 1 In the specific implementation of the method, the training device of the prediction model provided in the embodiment of the present invention can be applied to an electronic device, and its structural diagram is as shown in FIG. Figure 6 As shown, specifically including:

[0151] An acquisition unit 601 is used to acquire, in response to a model training instruction, an original booking sample of each flight collected in advance at each data collection time point;

[0152] A determination unit 602, configured to determine a preprocessing parameter for each of the data collection time points according to each of the original reservation samples;

[0153] The calculation unit 603 is used to calculate the booking demand data of each flight at each data collection time point based on the preprocessing parameters at each data collection time point and the original booking samples of each flight collected at each data collection time point;

[0154] The training unit 604 is used to apply each of the reservation demand data to train a pre-built prediction model.

[0155] In an embodiment provided by the present invention, based on the above implementation process, specifically, it further includes: an execution unit,

[0156] The execution unit is used to obtain the current flight information of the flight to be predicted; the flight information includes booking data; and the prediction model is used to process the flight information to obtain the booking prediction result of the flight to be predicted in the future time period.

[0157] In an embodiment provided by the present invention, based on the above implementation process, specifically, the determining unit 602 includes:

[0158] The first determination subunit is used to determine the sample quantity of the original booking samples collected at each data collection time point; and determine the total booking quantity corresponding to the data collection time point based on the original booking samples collected at each data collection time point;

[0159] The first execution sub-unit is used for, for each of the data collection time points, if the data collection time point is the first data collection time point, using the number of samples at the data collection time point and the total number of bookings corresponding to the data collection time point as preprocessing parameters of the data collection time point; if the data collection time point is not the first data collection time point, using the number of samples at the data collection time point, the total number of bookings corresponding to the data collection time point, and the total number of bookings at the data collection time point before the data collection time point as preprocessing parameters of the data collection time point.

[0160] In an embodiment provided by the present invention, based on the above implementation process, specifically, the calculation unit 603 includes:

[0161] A first calculation subunit is used to calculate a collection value at each data collection time point according to the preprocessing parameters of each data collection time point; the collection value represents an average increase in the number of seats booked at the data collection time point relative to the previous data collection time point;

[0162] A second determining subunit is used to determine the increment of the number of bookings for each flight at each data collection time point based on the original booking samples of each flight collected at each data collection time point;

[0163] The second calculation subunit is used to calculate the booking demand data of each flight at each data collection time point according to the collection value at each data collection time point and the increment of the number of bookings of each flight at each data collection time point.

[0164] In an embodiment provided by the present invention, based on the above implementation process, specifically, the determining unit 602 includes:

[0165] A third determining subunit is used to determine the increment of the number of bookings for each flight at the data collection time point according to the original booking samples of each flight collected at the data collection time point;

[0166] A third calculation subunit is used to calculate the sum of the increments of the number of bookings for each flight at the data collection time point and the square of the sum of the increments of the number of bookings;

[0167] a fourth calculation subunit, configured to calculate the mean and root variance of the sum of the incremental booking numbers according to the number of original booking samples of each flight collected at the data collection time point, the sum of the incremental booking numbers, and the square of the sum of the incremental booking numbers;

[0168] a fourth determining subunit, configured to determine a candidate seat number increment from each seat number increment according to the mean and the root variance;

[0169] a fifth calculation subunit, configured to calculate a first mean and a first root variance of the sum of the seat number increments corresponding to each of the alternative seat number increments according to the number of original seat booking samples of the flights to which each of the alternative seat number increments belongs, the sum of the seat number increments corresponding to each of the alternative seat number increments, and the square of the sum of the seat number increments corresponding to each of the alternative seat number increments;

[0170] a fifth determining subunit, configured to determine the first mean as a target mean, and determine the first root variance as a target root variance;

[0171] The iterative subunit is used to iterate the first operation until the number of iterations is greater than a preset number threshold, or the target mean value calculated by the current execution of the first operation meets the preset stop iteration condition; the first operation includes: determining a target seat number increment in each seat number increment according to the target mean value and the target root variance calculated currently; calculating the mean expectation and the variance expectation according to each target seat number increment, the first mean value and the first root variance; determining the flight to which each target seat number increment belongs as a target flight; according to the sum of the seat number increments of each target flight and the mean expectation, Calculate the sum of the new increments of the number of bookings for each of the target flights; calculate the square of the sum of the new increments of the number of bookings for each of the target flights according to the square of the sum of the increments of the number of bookings for each of the target flights and the expected variance; calculate the second mean and the second root variance of the sum of the increments of the number of bookings corresponding to each of the target increments of the number of bookings according to the number of original booking samples of each of the target flights, the sum of the new increments of the number of bookings for each of the target flights, and the square of the sum of the new increments of the number of bookings for each of the target flights; determine the second mean as the new target mean, and determine the second root variance as the new target root variance;

[0172] The second execution subunit is used to use the current target mean and the target root variance as preprocessing parameters of the data collection time point after stopping iterating the first operation.

[0173] In an embodiment provided by the present invention, based on the above implementation process, specifically, the calculation unit 603 includes:

[0174] A sixth calculation subunit, configured to calculate an expected increase in the number of seats booked for each flight at each data collection time point according to the increase in the number of seats booked for each flight at each data collection time point and a preprocessing parameter at each data collection time point;

[0175] The seventh calculation subunit is used to calculate the booking number demand data of each flight at each data collection time point based on the original booking samples of each flight collected at each data collection time point and the expected incremental booking number of each flight at each data collection time point.

[0176] The specific principles and execution processes of each unit and module in the training device of the prediction model disclosed in the above embodiment of the present invention are the same as the training method of the prediction model disclosed in the above embodiment of the present invention. Please refer to the corresponding parts of the training method of the prediction model provided in the above embodiment of the present invention, and will not be repeated here.

[0177] An embodiment of the present invention further provides a storage medium, which includes stored instructions, wherein when the instructions are executed, the device where the storage medium is located is controlled to execute the above-mentioned prediction model training method.

[0178] The embodiment of the present invention further provides an electronic device, the structural diagram of which is shown in FIG. Figure 7 As shown, it specifically includes a memory 701 and one or more instructions 702, wherein the one or more instructions 702 are stored in the memory 701 and are configured to be executed by one or more processors 703 to perform the following operations:

[0179] In response to the model training instruction, obtaining the original booking sample of each flight collected in advance at each data collection time point;

[0180] Determining the preprocessing parameters of each data collection time point according to each original booking sample;

[0181] Based on the preprocessing parameters at each data collection time point and the original booking samples of each flight collected at each data collection time point, the booking demand data of each flight at each data collection time point is calculated;

[0182] The pre-built prediction model is trained using each of the reservation demand data.

[0183] According to one or more embodiments of the present disclosure, Figure 1 An embodiment provides a prediction model training method, the method comprising: in response to a model training instruction, obtaining an original booking sample of each flight collected in advance at each data collection time point;

[0184] Determining the preprocessing parameters of each data collection time point according to each original booking sample;

[0185] Based on the preprocessing parameters at each data collection time point and the original booking samples of each flight collected at each data collection time point, the booking demand data of each flight at each data collection time point is calculated;

[0186] The pre-built prediction model is trained using each of the reservation demand data.

[0187] In the above method, optionally, determining the preprocessing parameters of each data collection time point according to each original reservation sample includes:

[0188] Determine the sample quantity of the original booking samples collected at each data collection time point; and determine the total booking quantity corresponding to the data collection time point based on the original booking samples collected at each data collection time point;

[0189] For each of the data collection time points, if the data collection time point is the first data collection time point, the number of samples at the data collection time point and the total number of bookings corresponding to the data collection time point are used as preprocessing parameters for the data collection time point; if the data collection time point is not the first data collection time point, the number of samples at the data collection time point, the total number of bookings corresponding to the data collection time point, and the total number of bookings at the data collection time point before the data collection time point are used as preprocessing parameters for the data collection time point.

[0190] The above method, optionally, the calculation of the booking demand data of each flight at each data collection time point based on the preprocessing parameters at each data collection time point and the original booking samples of each flight collected at each data collection time point includes:

[0191] A collection value at each data collection time point is calculated based on the preprocessing parameters at each data collection time point; the collection value represents the average increase in seat reservations at the data collection time point relative to the previous data collection time point;

[0192] Determine the increment of the number of bookings for each flight at each data collection time point based on the original booking samples of each flight collected at each data collection time point;

[0193] The seat booking demand data of each flight at each data collection time point is calculated based on the collection value at each data collection time point and the seat booking number increment of each flight at each data collection time point.

[0194] In the above method, optionally, determining the preprocessing parameters of each data collection time point according to each original reservation sample includes:

[0195] Determine the increment of the number of bookings for each flight at the data collection time point according to the original booking samples of each flight collected at the data collection time point;

[0196] According to the increment of the number of bookings of each flight at the data collection time point, calculate the sum of the increment of the number of bookings of each flight at the data collection time point, and the square of the sum of the increment of the number of bookings;

[0197] The mean and root variance of the sum of the incremental booking numbers are calculated based on the number of original booking samples of each flight collected at the data collection time point, the sum of the incremental booking numbers, and the square of the sum of the incremental booking numbers;

[0198] Determine the candidate seat number increments in each seat number increment according to the mean and the root variance;

[0199] Calculate a first mean and a first root variance of the sum of the seat number increments corresponding to each of the alternative seat number increments according to the number of original seat booking samples of the flights to which each of the alternative seat number increments belongs, the sum of the seat number increments corresponding to each of the alternative seat number increments, and the square of the sum of the seat number increments corresponding to each of the alternative seat number increments;

[0200] determining the first mean as a target mean, and determining the first root variance as a target root variance;

[0201] Iterate the first operation until the number of iterations is greater than a preset number threshold, or the target mean value calculated by the current execution of the first operation meets the preset stop iteration condition; the first operation includes: determining a target seat number increment in each seat number increment according to the target mean value and the target root variance calculated currently; calculating the mean expectation and the variance expectation according to each target seat number increment, the first mean value and the first root variance; determining the flight to which each target seat number increment belongs as a target flight; calculating the expected mean value according to the sum of the seat number increments of each target flight and the expected mean value the sum of the new increments of the number of bookings for each of the target flights; calculating the square of the sum of the new increments of the number of bookings for each of the target flights according to the square of the sum of the increments of the number of bookings for each of the target flights and the expected variance; calculating the second mean and the second root variance of the sum of the increments of the number of bookings corresponding to each of the target increments of the number of bookings according to the number of original booking samples of each of the target flights, the sum of the new increments of the number of bookings for each of the target flights, and the square of the sum of the new increments of the number of bookings for each of the target flights; determining the second mean as the new target mean, and determining the second root variance as the new target root variance;

[0202] After stopping iterating the first operation, the current target mean and the target root variance are used as preprocessing parameters of the data collection time point.

[0203] The above method, optionally, the calculation of the booking demand data of each flight at each data collection time point based on the preprocessing parameters at each data collection time point and the original booking samples of each flight collected at each data collection time point includes:

[0204] Calculate the expected increase in the number of seats for each flight at each data collection time point according to the increase in the number of seats for each flight at each data collection time point and the preprocessing parameters at each data collection time point;

[0205] Based on the original booking samples of each flight collected at each data collection time point and the expected incremental number of bookings for each flight at each data collection time point, the booking demand data for each flight at each data collection time point is calculated.

[0206] The above method may optionally further include:

[0207] Obtaining current flight information of the flight to be predicted; the flight information includes booking data;

[0208] The prediction model is applied to process the flight information to obtain a booking prediction result of the flight to be predicted in a future time period.

[0209] According to one or more embodiments of the present disclosure, Figure 6 An embodiment of the present invention provides a training device for a prediction model, the device comprising:

[0210] An acquisition unit, configured to acquire, in response to a model training instruction, an original booking sample of each flight collected in advance at each data collection time point;

[0211] A determination unit, configured to determine a preprocessing parameter for each of the data collection time points according to each of the original reservation samples;

[0212] A calculation unit, configured to calculate the seat booking demand data of each flight at each data collection time point based on the preprocessing parameters at each data collection time point and the original seat booking samples of each flight collected at each data collection time point;

[0213] A training unit is used to apply each of the reservation demand data to train a pre-built prediction model.

[0214] The above device may optionally further include: an execution unit,

[0215] The execution unit is used to obtain the current flight information of the flight to be predicted; the flight information includes booking data; and the prediction model is used to process the flight information to obtain the booking prediction result of the flight to be predicted in the future time period.

[0216] In the above device, optionally, the determining unit includes:

[0217] The first determination subunit is used to determine the sample quantity of the original booking samples collected at each data collection time point; and determine the total booking quantity corresponding to the data collection time point based on the original booking samples collected at each data collection time point;

[0218] The first execution sub-unit is used for, for each of the data collection time points, if the data collection time point is the first data collection time point, using the number of samples at the data collection time point and the total number of bookings corresponding to the data collection time point as preprocessing parameters of the data collection time point; if the data collection time point is not the first data collection time point, using the number of samples at the data collection time point, the total number of bookings corresponding to the data collection time point, and the total number of bookings at the data collection time point before the data collection time point as preprocessing parameters of the data collection time point.

[0219] In the above device, optionally, the computing unit comprises:

[0220] A first calculation subunit is used to calculate a collection value at each data collection time point according to the preprocessing parameters of each data collection time point; the collection value represents an average increase in the number of seats booked at the data collection time point relative to the previous data collection time point;

[0221] A second determining subunit is used to determine the increment of the number of bookings for each flight at each data collection time point based on the original booking samples of each flight collected at each data collection time point;

[0222] The second calculation subunit is used to calculate the booking demand data of each flight at each data collection time point according to the collection value at each data collection time point and the increment of the number of bookings of each flight at each data collection time point.

[0223] In the above device, optionally, the determining unit includes:

[0224] A third determining subunit is used to determine the increment of the number of bookings for each flight at the data collection time point according to the original booking samples of each flight collected at the data collection time point;

[0225] A third calculation subunit is used to calculate the sum of the increments of the number of bookings for each flight at the data collection time point and the square of the sum of the increments of the number of bookings;

[0226] a fourth calculation subunit, configured to calculate the mean and root variance of the sum of the incremental booking numbers according to the number of original booking samples of each flight collected at the data collection time point, the sum of the incremental booking numbers, and the square of the sum of the incremental booking numbers;

[0227] a fourth determining subunit, configured to determine a candidate seat number increment from each seat number increment according to the mean and the root variance;

[0228] a fifth calculation subunit, configured to calculate a first mean and a first root variance of the sum of the seat number increments corresponding to each of the alternative seat number increments according to the number of original seat booking samples of the flights to which each of the alternative seat number increments belongs, the sum of the seat number increments corresponding to each of the alternative seat number increments, and the square of the sum of the seat number increments corresponding to each of the alternative seat number increments;

[0229] a fifth determining subunit, configured to determine the first mean as a target mean, and determine the first root variance as a target root variance;

[0230] The iterative subunit is used to iterate the first operation until the number of iterations is greater than a preset number threshold, or the target mean value calculated by the current execution of the first operation meets the preset stop iteration condition; the first operation includes: determining a target seat number increment in each seat number increment according to the target mean value and the target root variance calculated currently; calculating the mean expectation and the variance expectation according to each target seat number increment, the first mean value and the first root variance; determining the flight to which each target seat number increment belongs as a target flight; according to the sum of the seat number increments of each target flight and the mean expectation, Calculate the sum of the new increments of the number of bookings for each of the target flights; calculate the square of the sum of the new increments of the number of bookings for each of the target flights according to the square of the sum of the increments of the number of bookings for each of the target flights and the expected variance; calculate the second mean and the second root variance of the sum of the increments of the number of bookings corresponding to each of the target increments of the number of bookings according to the number of original booking samples of each of the target flights, the sum of the new increments of the number of bookings for each of the target flights, and the square of the sum of the new increments of the number of bookings for each of the target flights; determine the second mean as the new target mean, and determine the second root variance as the new target root variance;

[0231] The second execution subunit is used to use the current target mean and the target root variance as preprocessing parameters of the data collection time point after stopping iterating the first operation.

[0232] In the above device, optionally, the computing unit comprises:

[0233] A sixth calculation subunit, configured to calculate an expected increase in the number of seats booked for each flight at each data collection time point according to the increase in the number of seats booked for each flight at each data collection time point and a preprocessing parameter at each data collection time point;

[0234] The seventh calculation subunit is used to calculate the booking number demand data of each flight at each data collection time point based on the original booking samples of each flight collected at each data collection time point and the expected incremental booking number of each flight at each data collection time point.

[0235] It should be noted that each embodiment in this specification is described in a progressive manner, and each embodiment focuses on the differences from other embodiments, and the same or similar parts between the embodiments can be referred to each other. For the device embodiment, since it 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.

[0236] Finally, it should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.

[0237] For the convenience of description, the above device is described as being divided into various units according to their functions. Of course, when implementing the present invention, the functions of each unit can be implemented in the same or multiple software and / or hardware.

[0238] It can be known from the description of the above implementation methods that those skilled in the art can clearly understand that the present invention can be implemented by means of software plus a necessary general hardware platform. Based on such an understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art can be embodied in the form of a software product, which can be stored in a storage medium such as ROM / RAM, a magnetic disk, an optical disk, etc., and includes a number of instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in the various embodiments of the present invention or certain parts of the embodiments.

[0239] The above is a detailed introduction to the training method of a prediction model provided by the present invention. Specific examples are used in this article to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea; at the same time, for those skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation method and application scope. In summary, the content of this specification should not be understood as a limitation on the present invention.

Claims

1. A prediction model training method, characterized in that: include: In response to the model training instruction, obtaining the original booking sample of each flight collected in advance at each data collection time point; Determining the preprocessing parameters of each data collection time point according to each original booking sample; Based on the preprocessing parameters at each data collection time point and the original booking samples of each flight collected at each data collection time point, the booking demand data of each flight at each data collection time point is calculated; Using each of the booking demand data to train a pre-built prediction model; Wherein, determining the preprocessing parameters of each data collection time point according to each original booking sample includes: Determine the increment of the number of bookings for each flight at the data collection time point according to the original booking samples of each flight collected at the data collection time point; According to the increment of the number of seat bookings of each flight at the data collection time point, calculate the sum of the increment of the number of seat bookings of each flight at the data collection time point, and the square of the sum of the increment of the number of seat bookings; The mean and root variance of the sum of the incremental booking numbers are calculated based on the number of original booking samples of each flight collected at the data collection time point, the sum of the incremental booking numbers, and the square of the sum of the incremental booking numbers; Determine the candidate seat number increments in each seat number increment according to the mean and the root variance; Calculate a first mean and a first root variance of the sum of the seat number increments corresponding to each of the alternative seat number increments according to the number of original seat booking samples of the flights to which each of the alternative seat number increments belongs, the sum of the seat number increments corresponding to each of the alternative seat number increments, and the square of the sum of the seat number increments corresponding to each of the alternative seat number increments; determining the first mean as a target mean, and determining the first root variance as a target root variance; Iterate the first operation until the number of iterations is greater than a preset number threshold, or the target mean value calculated by the current execution of the first operation meets the preset stop iteration condition; the first operation includes: determining a target seat number increment in each seat number increment according to the target mean value and the target root variance calculated currently; calculating the mean expectation and the variance expectation according to each target seat number increment, the first mean value and the first root variance; determining the flight to which each target seat number increment belongs as a target flight; calculating the expected mean value according to the sum of the seat number increments of each target flight and the expected mean value the sum of the new increments of the number of bookings for each of the target flights; calculating the square of the sum of the new increments of the number of bookings for each of the target flights according to the square of the sum of the increments of the number of bookings for each of the target flights and the expected variance; calculating the second mean and the second root variance of the sum of the increments of the number of bookings corresponding to each of the target increments of the number of bookings according to the number of original booking samples of each of the target flights, the sum of the new increments of the number of bookings for each of the target flights, and the square of the sum of the new increments of the number of bookings for each of the target flights; determining the second mean as the new target mean, and determining the second root variance as the new target root variance; After stopping iterating the first operation, the current target mean and the target root variance are used as preprocessing parameters of the data collection time point.

2. The method according to claim 1, characterized in that The determining of the preprocessing parameters of each data collection time point according to each original reservation sample includes: Determine the sample quantity of the original booking samples collected at each data collection time point; and determine the total booking quantity corresponding to the data collection time point based on the original booking samples collected at each data collection time point; For each of the data collection time points, if the data collection time point is the first data collection time point, the number of samples at the data collection time point and the total number of bookings corresponding to the data collection time point are used as preprocessing parameters for the data collection time point; if the data collection time point is not the first data collection time point, the number of samples at the data collection time point, the total number of bookings corresponding to the data collection time point, and the total number of bookings at the data collection time point before the data collection time point are used as preprocessing parameters for the data collection time point.

3. The method according to claim 2, characterized in that The calculation based on the preprocessing parameters at each data collection time point and the original booking samples of each flight collected at each data collection time point to obtain the booking demand data of each flight at each data collection time point includes: A collection value at each data collection time point is calculated based on the preprocessing parameters at each data collection time point; the collection value represents the average increase in seat reservations at the data collection time point relative to the previous data collection time point; Determine the increment of the number of bookings for each flight at each data collection time point based on the original booking samples of each flight collected at each data collection time point; The seat booking demand data of each flight at each data collection time point is calculated based on the collection value at each data collection time point and the seat booking number increment of each flight at each data collection time point.

4. The method according to claim 1, characterized in that The calculation based on the preprocessing parameters at each data collection time point and the original booking samples of each flight collected at each data collection time point to obtain the booking demand data of each flight at each data collection time point includes: Calculate the expected increase in the number of seats for each flight at each data collection time point according to the increase in the number of seats for each flight at each data collection time point and the preprocessing parameters at each data collection time point; The seat demand data of each flight at each data collection time point is calculated based on the original booking samples of each flight collected at each data collection time point and the expected incremental number of bookings of each flight at each data collection time point.

5. The method according to claim 1, characterized in that Also includes: Get the current flight information of the flight to be predicted; The flight information includes booking data; The prediction model is applied to process the flight information to obtain a booking prediction result of the flight to be predicted in a future time period.

6. A training device for a prediction model, characterized in that: include: An acquisition unit, configured to acquire, in response to a model training instruction, an original booking sample of each flight collected in advance at each data collection time point; A determination unit, configured to determine a preprocessing parameter for each of the data collection time points according to each of the original reservation samples; A calculation unit, configured to calculate the booking demand data of each flight at each data collection time point based on the preprocessing parameters at each data collection time point and the original booking samples of each flight collected at each data collection time point; A training unit, used for applying each of the booking demand data to train a pre-built prediction model; The determining unit comprises: A third determining subunit is used to determine the increment of the number of bookings for each flight at the data collection time point according to the original booking samples of each flight collected at the data collection time point; A third calculation subunit is used to calculate the sum of the increments of the number of bookings for each flight at the data collection time point and the square of the sum of the increments of the number of bookings; a fourth calculation subunit, configured to calculate the mean and root variance of the sum of the incremental booking numbers according to the number of original booking samples of each flight collected at the data collection time point, the sum of the incremental booking numbers, and the square of the sum of the incremental booking numbers; a fourth determining subunit, configured to determine a candidate seat number increment from each seat number increment according to the mean and the root variance; a fifth calculation subunit, configured to calculate a first mean and a first root variance of the sum of the seat number increments corresponding to each of the alternative seat number increments according to the number of original seat booking samples of the flights to which each of the alternative seat number increments belongs, the sum of the seat number increments corresponding to each of the alternative seat number increments, and the square of the sum of the seat number increments corresponding to each of the alternative seat number increments; a fifth determining subunit, configured to determine the first mean as a target mean, and determine the first root variance as a target root variance; The iterative subunit is used to iterate the first operation until the number of iterations is greater than a preset number threshold, or the target mean value calculated by the current execution of the first operation meets the preset stop iteration condition; the first operation includes: determining a target seat number increment in each seat number increment according to the target mean value and the target root variance calculated currently; calculating the mean expectation and the variance expectation according to each target seat number increment, the first mean value and the first root variance; determining the flight to which each target seat number increment belongs as a target flight; according to the sum of the seat number increments of each target flight and the mean expectation, Calculate the sum of the new increments of the number of bookings for each of the target flights; calculate the square of the sum of the new increments of the number of bookings for each of the target flights according to the square of the sum of the increments of the number of bookings for each of the target flights and the expected variance; calculate the second mean and the second root variance of the sum of the increments of the number of bookings corresponding to each of the target increments of the number of bookings according to the number of original booking samples of each of the target flights, the sum of the new increments of the number of bookings for each of the target flights, and the square of the sum of the new increments of the number of bookings for each of the target flights; determine the second mean as the new target mean, and determine the second root variance as the new target root variance; The second execution subunit is used to use the current target mean and the target root variance as preprocessing parameters of the data collection time point after stopping iterating the first operation.

7. The device according to claim 6, characterized in that Also includes: an execution unit, The execution unit is used to obtain the current flight information of the flight to be predicted; the flight information includes booking data; and the prediction model is used to process the flight information to obtain the booking prediction result of the flight to be predicted in the future time period.

8. A storage medium, characterized in that: The storage medium includes storage instructions, wherein when the instructions are executed, the device where the storage medium is located is controlled to execute the training method of the prediction model according to any one of claims 1 to 5.

9. An electronic device, characterized in that: It comprises a memory and one or more instructions, wherein the one or more instructions are stored in the memory and are configured to be executed by one or more processors to perform the training method of the prediction model as described in any one of claims 1 to 5.

Images