CN115409273A

CN115409273A - Date prediction method, date prediction device, computer equipment and storage medium

Info

Publication number: CN115409273A
Application number: CN202211075096.9A
Authority: CN
Inventors: 杨尚航
Original assignee: Industrial and Commercial Bank of China Ltd ICBC
Current assignee: Industrial and Commercial Bank of China Ltd ICBC
Priority date: 2022-09-02
Filing date: 2022-09-02
Publication date: 2022-11-29

Abstract

The application relates to a date prediction method, a date prediction device, computer equipment, a storage medium and a computer program product, relates to the technical field of artificial intelligence, and can be used in the field of financial science and technology or other fields. The method comprises the following steps: acquiring current task progress index data corresponding to a target object; inputting the current task progress index data into a date prediction model to obtain the task prediction completion date of the target object; the date prediction model is obtained by processing historical task progress index data corresponding to a target object based on an initial date prediction model to obtain a historical task prediction completion date, determining a genetic probability parameter according to the actual completion date of the historical task, the historical task prediction completion date, a genetic algorithm and the current iteration number by taking parameters of the initial date prediction model as individuals, and updating the parameters of the initial date prediction model based on the genetic probability parameter. By adopting the method, the task completion date can be predicted more accurately.

Description

Date prediction method, date prediction device, computer equipment and storage medium

Technical Field

The present application relates to the field of artificial intelligence technologies, and in particular, to a date prediction method, apparatus, computer device, storage medium, and computer program product.

Background

For corporate financial management, the management of specific objects (such as properties, other assets, etc.) plays an important role. For example, financial institutions such as banks generally perform a house inventory and the like every year, and an inventory clerk needs to complete the inventory and the like of each house involved in the institution before a prescribed time. In order to supervise the smooth completion of the house inventory task, a manager needs to know the current task progress in real time.

In the related art, the manager generally estimates the task completion date through the current checked number of properties, the total number of properties to be checked and the current spent time, and by combining experience, so as to know the current task progress. For example, if the number of currently checked properties is 5, the total number of properties to be checked is 10, and the currently spent checking time is 1 month, the percentage of the checked properties is 50%, and it can be roughly estimated that 1 month is spent on checking the remaining 50% of the properties.

However, due to the difficulty or different situations in the house inventory process, the method for predicting the task completion date according to the manual experience has low prediction accuracy and cannot accurately reflect the progress of the house inventory task.

Disclosure of Invention

In view of the above, it is necessary to provide a date prediction method, apparatus, computer device, computer readable storage medium, and computer program product capable of improving the accuracy of prediction of a task completion date in view of the above technical problems.

In a first aspect, the present application provides a method of date prediction. The method comprises the following steps:

acquiring current task progress index data corresponding to a target object; the current task progress index data comprises the total number of the task objects, the current date, the number of the currently completed task objects, the task permission starting date and the task starting execution date;

inputting the current task progress index data into a date prediction model to obtain the task prediction completion date of the target object;

the date prediction model is obtained by processing historical task progress index data corresponding to the target object based on an initial date prediction model to obtain a historical task prediction completion date, determining a genetic probability parameter according to the actual completion date of the historical task corresponding to the target object, the historical task prediction completion date, a genetic algorithm and the current iteration number by taking parameters of the initial date prediction model as individuals, and updating the parameters of the initial date prediction model based on the genetic probability parameter.

In one embodiment, the training process of the date prediction model comprises:

acquiring the historical task progress index data and the actual completion date of the historical task corresponding to the target object; the historical task progress index data comprises the total number of historical task objects, historical dates, the number of completed task objects corresponding to the historical dates, historical task permission starting dates and historical task starting execution dates;

inputting the historical task progress index data into a plurality of initial date prediction models to obtain a plurality of historical task prediction completion dates; wherein the parameters of each initial date prediction model are different;

and determining the genetic probability parameters according to the actual completion date of the historical tasks, the predicted completion dates of the historical tasks, the genetic algorithm and the current iteration number by taking the parameters of the initial date prediction model as individuals, updating the parameters of each initial date prediction model based on the genetic probability parameters, and executing the steps of obtaining the historical task progress index data and the actual completion date of the historical tasks corresponding to the target object until a preset training stop condition is reached to obtain the trained date prediction model.

In one embodiment, the determining the genetic probability parameter according to the actual completion date of the historical tasks, the predicted completion dates of the historical tasks, the genetic algorithm, and the current iteration number includes:

aiming at each initial date prediction model, calculating a loss value corresponding to the initial date prediction model according to the actual completion date of the historical task, the predicted completion date of the historical task corresponding to the initial date prediction model and an objective function;

and determining the fitness value of the individual corresponding to each initial date prediction model according to the loss value corresponding to each initial date prediction model, and determining a genetic probability parameter based on the fitness value and the current iteration times.

In one embodiment, the genetic probability parameter comprises a cross probability parameter; determining the genetic probability parameter based on the fitness value and the current iteration number comprises:

determining a maximum fitness value and an average fitness value according to the fitness value of each individual;

selecting a plurality of target individual pairs from each individual according to the fitness value of each individual and a preset selection strategy;

and aiming at each target individual pair, determining the maximum value of the fitness values of the two individuals in the target individual pair as a target fitness value, and calculating a cross probability parameter corresponding to the target individual pair according to the target fitness value, the maximum fitness value, the average fitness value and the current iteration times.

In one embodiment, the calculating a cross probability parameter corresponding to the target individual pair according to the target fitness value, the maximum fitness value, the average fitness value, and the current iteration number includes:

under the condition that the target fitness value is greater than or equal to the average fitness value, calculating a cross probability parameter corresponding to the target individual pair according to the target fitness value, the maximum fitness value, the average fitness value and the current iteration times;

and under the condition that the target fitness value is smaller than the average fitness value, calculating a cross probability parameter corresponding to the target individual pair according to the current iteration times.

In one embodiment, the genetic probability parameter comprises a mutation probability parameter; determining the genetic probability parameter according to the fitness value corresponding to each individual and the current iteration number, wherein the determining comprises:

selecting a plurality of target individuals from the individuals according to the fitness value of each individual and a preset selection strategy; one of said target individuals corresponding to one of said initial date prediction models;

and calculating a variation probability parameter corresponding to each target individual according to the fitness value, the maximum fitness value, the average fitness value and the current iteration number of the target individual.

In one embodiment, the calculating, according to the fitness value of the target individual, the maximum fitness value, the average fitness value, and the current iteration number, a variation probability parameter corresponding to the target individual includes:

under the condition that the fitness value of the target individual is greater than or equal to the average fitness value, calculating a variation probability parameter corresponding to the target individual according to the fitness value of the target individual, the maximum fitness value, the average fitness value and the current iteration times;

and under the condition that the fitness value of the target individual is smaller than the average fitness value, calculating a variation probability parameter corresponding to the target individual according to the current iteration times.

In a second aspect, the present application also provides a date prediction apparatus. The device comprises:

the first acquisition module is used for acquiring current task progress index data corresponding to a target object; the current task progress index data comprises the total number of the task objects, the current date, the number of the currently completed task objects, the task permission starting date and the task starting execution date;

the prediction module is used for inputting the current task progress index data into a date prediction model to obtain the task prediction completion date of the target object; the date prediction model is obtained by processing historical task progress index data corresponding to the target object based on an initial date prediction model to obtain a historical task prediction completion date, determining a genetic probability parameter according to the actual completion date of the historical task corresponding to the target object, the historical task prediction completion date, a genetic algorithm and the current iteration frequency by taking parameters of the initial date prediction model as individuals, and updating the parameters of the initial date prediction model based on the genetic probability parameter.

In one embodiment, the apparatus further comprises:

the second acquisition module is used for acquiring the historical task progress index data and the historical task actual completion date corresponding to the target object; the historical task progress index data comprises the total number of historical task objects, historical dates, the number of completed task objects corresponding to the historical dates, historical task permission starting dates and historical task starting execution dates;

the input module is used for inputting the historical task progress index data into the plurality of initial date prediction models to obtain a plurality of historical task prediction completion dates; wherein the parameters of each initial date prediction model are different;

and the updating module is used for determining the genetic probability parameters according to the actual completion date of the historical tasks, the predicted completion dates of the historical tasks, the genetic algorithm and the current iteration number by taking the parameters of the initial date prediction model as individuals, updating the parameters of each initial date prediction model based on the genetic probability parameters, and executing the steps of obtaining the historical task progress index data corresponding to the target object and the actual completion date of the historical tasks until a preset training stopping condition is reached to obtain the trained date prediction model.

In one embodiment, the update module is specifically configured to:

aiming at each initial date prediction model, calculating a loss value corresponding to the initial date prediction model according to the actual completion date of the historical task, the historical task prediction completion date corresponding to the initial date prediction model and an objective function; and determining the fitness value of the individual corresponding to each initial date prediction model according to the loss value corresponding to each initial date prediction model, and determining the genetic probability parameter based on the fitness value and the current iteration times.

In one embodiment, the genetic probability parameter comprises a cross probability parameter; the update module is specifically configured to:

determining a maximum fitness value and an average fitness value according to the fitness value of each individual; selecting a plurality of target individual pairs from each individual according to the fitness value of each individual and a preset selection strategy; and aiming at each target individual pair, determining the maximum value of the adaptability values of the two individuals in the target individual pair as a target adaptability value, and calculating a cross probability parameter corresponding to the target individual pair according to the target adaptability value, the maximum adaptability value, the average adaptability value and the current iteration number.

In one embodiment, the update module is specifically configured to:

under the condition that the target fitness value is greater than or equal to the average fitness value, calculating a cross probability parameter corresponding to the target individual pair according to the target fitness value, the maximum fitness value, the average fitness value and the current iteration times; and under the condition that the target fitness value is smaller than the average fitness value, calculating a cross probability parameter corresponding to the target individual pair according to the current iteration times.

In one embodiment, the genetic probability parameter comprises a mutation probability parameter; the update module is specifically configured to:

determining a maximum fitness value and an average fitness value according to the fitness value of each individual; selecting a plurality of target individuals from the individuals according to the fitness value of each individual and a preset selection strategy; one of said target individuals corresponding to one of said initial date prediction models; and calculating a variation probability parameter corresponding to each target individual according to the fitness value, the maximum fitness value, the average fitness value and the current iteration number of the target individual.

In one embodiment, the update module is specifically configured to:

under the condition that the fitness value of the target individual is greater than or equal to the average fitness value, calculating a variation probability parameter corresponding to the target individual according to the fitness value of the target individual, the maximum fitness value, the average fitness value and the current iteration times; and under the condition that the fitness value of the target individual is smaller than the average fitness value, calculating a variation probability parameter corresponding to the target individual according to the current iteration times.

In a third aspect, the application also provides a computer device. The computer device comprises a memory storing a computer program and a processor implementing the steps of the method of the first aspect when executing the computer program.

In a fourth aspect, the present application further provides a computer-readable storage medium. The computer-readable storage medium having stored thereon a computer program which, when being executed by a processor, carries out the steps of the method of the first aspect.

In a fifth aspect, the present application further provides a computer program product. The computer program product comprising a computer program that, when executed by a processor, performs the steps of the method of the first aspect.

According to the date prediction method, the date prediction device, the computer equipment, the storage medium and the computer program product, the task prediction completion date of the target object is obtained by acquiring the current task progress index data of the target object and inputting the data into the date prediction model. The current task progress index data comprises the total number of task objects, the current date, the number of the currently completed task objects, the task permission starting date and the task starting execution date. The date prediction model is obtained by processing historical task progress index data corresponding to a target object based on an initial date prediction model to obtain a historical task prediction completion date, determining a genetic probability parameter according to the actual completion date of the historical task corresponding to the target object, the historical task prediction completion date, a genetic algorithm and the current iteration number by taking parameters of the initial date prediction model as individuals, and updating the parameters of the initial date prediction model based on the genetic probability parameter.

According to the method, the task prediction completion date of the target object is obtained by processing the current task progress index data corresponding to the target object through a date prediction model. In the training process of the date prediction model, parameters of the model are updated by adopting a genetic algorithm, wherein genetic probability parameters are determined according to the current iteration times, the global convergence of the genetic algorithm can be improved, namely parameters of the model with a better global state can be obtained, and the prediction accuracy of the date prediction model obtained by training on the task completion date is higher. Therefore, compared with the method for predicting the task progress according to manual experience, the method can more accurately predict the task completion date so as to more accurately reflect the task progress and provide a basis for the subsequent management of a manager.

Drawings

FIG. 1 is a flow diagram illustrating a method for date prediction in one embodiment;

FIG. 2 is a schematic flow chart diagram illustrating the training process for the date prediction model in one embodiment;

FIG. 3 is a schematic diagram of a structure of a date prediction model in one example;

FIG. 4 is a schematic flow chart of the determination of the genetic probability parameter in one embodiment;

FIG. 5 is a schematic flow chart of the determination of the genetic probability parameter in another embodiment;

FIG. 6 is a schematic flow chart of the determination of the genetic probability parameter in another embodiment;

FIG. 7 is a block diagram showing the construction of a date prediction apparatus according to an embodiment;

FIG. 8 is a diagram illustrating an internal structure of a computer device according to an embodiment.

Detailed Description

In order to make the objects, technical solutions and advantages of the present application more apparent, the present application is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present application and are not intended to limit the present application.

First, before specifically describing the technical solution of the embodiment of the present application, a technical background or a technical evolution context on which the embodiment of the present application is based is described. For enterprise financial management, the management of specific objects (such as real estate, assets, etc.) plays an important role. For example, financial institutions such as banks generally perform a house inventory and the like every year, and an inventory clerk needs to complete the inventory and the like of each house involved in the institution before a prescribed time. In order to supervise the smooth completion of the house inventory task, a manager needs to know the current task progress in real time. In the related art, a manager generally estimates a task completion date through the current checked number of properties, the total number of properties to be checked, and the current spent time, and by combining experience, so as to know the current task progress. For example, if the current checked number of properties is 5, the total number of properties to be checked is 10, and the current checking time is 1 month, the percentage of the checked properties is 50%, and it can be roughly estimated that it takes 1 month to check the remaining 50% of the properties.

However, due to the difficulty or different situations in the house inventory process, the method for predicting the task completion date according to manual experience has low prediction accuracy and cannot accurately reflect the progress of the house inventory task. Based on the background, the applicant provides the date prediction method through long-term research and development and experimental verification, and the accuracy of task completion date prediction can be improved. In addition, it should be noted that the applicant has paid a lot of creative efforts to find technical problems of the present application and technical solutions described in the following embodiments.

In an embodiment, as shown in fig. 1, a date prediction method is provided, and this embodiment is illustrated by applying the method to a terminal, and it is to be understood that the method may also be applied to a server, and may also be applied to a system including the terminal and the server, and is implemented through interaction between the terminal and the server. The terminal can be, but is not limited to, various personal computers, notebook computers, smart phones, tablet computers, internet of things devices, portable wearable devices and the like. In this embodiment, the method includes the steps of:

step 101, obtaining current task progress index data corresponding to a target object.

The target object refers to an object that needs to perform tasks such as specific object inventory, for example, the target object may be a financial institution such as a bank, or another target institution that needs to perform tasks such as specific object inventory, the specific object may be a real estate or another asset, and there are generally a plurality of specific objects. The current task progress index data refers to data related to tasks such as specific object inventory of a target object, and includes a task object total number (such as a total number of inventory of property to be reserved), a current date (generally, after a date of opening task permission and before the task is completed), a number of currently completed task objects (such as a number of currently completed property), a task permission opening date (such as a date of opening task permission for inventory of property for an inventory clerk), and a task start execution date (such as a date of starting inventory of first property for the inventory clerk).

In implementation, a user may input current task progress index data corresponding to a target object at a terminal, so that the terminal may obtain the current task progress index data corresponding to the target object based on the data input by the user. The current date may be the day on which the date prediction is made.

And 102, inputting the current task progress index data into a date prediction model to obtain the task prediction completion date of the target object.

The date prediction model is obtained by processing historical task progress index data corresponding to a target object based on an initial date prediction model to obtain historical task prediction completion dates, determining genetic probability parameters according to the actual completion dates of the historical tasks corresponding to the target object, the historical task prediction completion dates, a genetic algorithm and the current iteration times by taking parameters of the initial date prediction model as individuals, and updating the parameters of the initial date prediction model based on the genetic probability parameters.

In implementation, the terminal may input the current task progress index data to a pre-trained date prediction model, and process the current task progress index data through the date prediction model to obtain a task prediction completion date of the target object. The date prediction model may be a Back Propagation (BP) Neural Network or a Recurrent Neural Network (RNN), and may be obtained by training the initial date prediction model in advance through historical task progress index data and historical task prediction completion date of the target object. During the model training process, the parameters of the initial date prediction model can be updated by adopting a genetic algorithm. The genetic probability parameter of the genetic algorithm can be obtained by calculation according to the actual completion date of the historical tasks, the predicted completion dates of the historical tasks, the genetic algorithm and the current iteration number, so that the genetic probability parameter changes with the increase of the iteration number, for example, the genetic probability parameter can be gradually reduced with the increase of the iteration number. Therefore, the genetic probability parameters are larger at the initial stage of iteration so as to keep the diversity of the population, and the genetic probability parameters are smaller at the middle and later stages of iteration so as to prevent the optimal solution from being damaged, thereby improving the global convergence of the genetic algorithm and further improving the accuracy of the trained date prediction model in predicting the task completion date. The iteration number refers to the number of times of updating the model parameters by using a genetic algorithm, and when the genetic probability parameters are determined for the first time to update the parameters, the current iteration number is 1. The genetic probability parameter may specifically comprise one or more of a selection probability parameter, a cross probability parameter and a mutation probability parameter.

In some examples, for a prediction scene of the house inventory task progress, the task progress index data hardly have a mutual dependency relationship, so that the date prediction model can adopt a BP neural network, is more suitable for the progress prediction scene of tasks such as house inventory and the like, has a simpler network structure, and can give consideration to both prediction accuracy and training and prediction efficiency.

In the above-mentioned date prediction method, the task completion date of the target object is predicted by the current task progress index data corresponding to the target object and the date prediction model. In the training process of the date prediction model, parameters of the model are updated by adopting a genetic algorithm, wherein genetic probability parameters are determined according to the current iteration times, the global convergence of the genetic algorithm can be improved, namely parameters of the model with a better global state can be obtained, and the prediction accuracy of the date prediction model obtained by training on the task completion date is higher. Therefore, the method can accurately predict the task completion date so as to accurately reflect the task progress and provide a basis for the subsequent management of a manager.

In one embodiment, as shown in fig. 2, the training process of the date prediction model specifically includes the following steps:

step 201, obtaining historical task progress index data and historical task actual completion date corresponding to the target object.

The historical task progress index data is historical data generated by tasks such as specific object inventory and the like aiming at a target object, and comprises the total number of historical task objects, historical dates, the number of completed task objects corresponding to the historical dates, historical task permission starting dates and historical task starting execution dates. In some examples, the historical task progress indicator data may be data generated by performing a property inventory task on the target object last time (if the property inventory task is performed once a year, the last time is the last year), and accordingly, the historical task object total number (which may be N) may be a total number of properties to be inventoried of the target object at the last property inventory task, the historical task permission starting date (which may be dateS) may be a date when the property inventory task permission is started for the inventory member at the last property inventory task (i.e., from the day, the inventory member may start to inventory the properties to be inventoried), the historical task starting execution date (which may be date 0) may be a date when the inventory member starts to inventory the first property at the last property inventory task, and the historical task actual completion date (which may be date 1) may be an actual completion date of the last property inventory task. The historical date (which can be recorded as date) refers to the historical task authority opening date dateS and the actual completion date of the historical task date1, the number of completed task objects (which may be denoted as M) corresponding to the history date may be the number of properties of completed inventory corresponding to the history date.

Since the target organization generally relates to a plurality of branch organizations, each branch organization may respectively perform tasks such as inventory of respective specific objects, and historical task progress index data and actual historical task completion dates corresponding to each branch organization may be generated, each branch organization may be respectively used as a target object to respectively train a date prediction model corresponding to each branch organization, or may be used as a target object as a whole, and an identifier of each branch organization may be used as one of historical task progress index data, that is, the historical task progress index data may further include an identifier (which may be denoted as fancode) of a sub-target object, and the identifier fancode of the sub-target object has a corresponding relationship with other historical task progress index data and actual historical task completion dates. Correspondingly, if the historical task progress index data includes the identifier fancode of the sub-target object, the current task progress index data acquired in step 101 also includes the identifier fancode of the sub-target object, that is, the task prediction completion date of each sub-target object can be predicted according to the historical task progress index data of each sub-target object (such as a branch office).

In implementation, the historical task progress index data and the actual completion date of the historical task corresponding to the target object may be stored in the database in advance, and the terminal may directly obtain the historical task progress index data and the actual completion date from the database to be used as a sample of the training date prediction model. I.e. each training sample will contain historical task progress indicator data and the actual completion date of the historical task. Specifically, historical task progress index data and historical task actual completion dates generated by one or more historical tasks performed on the target object may be used as training samples. For the historical data generated by one historical task, one historical date corresponds to one training sample, and a plurality of historical dates can be set so as to obtain a plurality of training samples. For example, if the total number of history task objects of a task such as the last specific object inventory of the target object is 6 (i.e., N = 6), the history task authority open date is 4 months and 1 days (i.e., dateS =4 months and 1 days), the history task start execution date is 4 months and 15 days (i.e., date0=4 months and 15 days), the history task actual completion date is 10 months and 1 days (i.e., date1=10 months and 1 day), the history date may be set to 5 months and 1 days (i.e., date =5 months and 1 days), and the number of completed task objects corresponding to 5 months and 1 days is 1 (i.e., M = 1), then N =6, dateS =4 months and 1 day, date0=4 months and 15 days, date =5 months and 1 day, and date1=10 months and 1 day may be used as a training sample of the history task actual completion date. By setting a plurality of different historical dates, a plurality of training samples can be obtained.

Step 202, inputting the historical task progress index data into a plurality of initial date prediction models to obtain a plurality of historical task prediction completion dates.

In implementation, the terminal may input the acquired historical task progress index data corresponding to the target object to the plurality of initial date prediction models to obtain a plurality of historical task prediction completion dates. The number of the initial date prediction models can be set according to needs, and the parameters of the initial date prediction models are different. It can be understood that, for a date prediction model, multiple times of parameter initialization may be performed, where the parameters of each initialization are different, and the historical task progress indicators are input to the date prediction model after each initialization, so as to obtain multiple historical task prediction completion dates.

And 203, taking the parameters of the initial date prediction model as individuals, determining genetic probability parameters according to actual completion dates of the historical tasks, the predicted completion dates of the historical tasks, the genetic algorithm and the current iteration number, updating the parameters of each initial date prediction model based on the genetic probability parameters, and executing the steps of obtaining historical task progress index data and the actual completion dates of the historical tasks corresponding to the target object until preset training stopping conditions are reached to obtain the trained date prediction model.

In implementation, the terminal may use the parameters of the initial date prediction model as individuals, determine a genetic probability parameter according to actual completion dates of historical tasks, predicted completion dates of multiple historical tasks, a genetic algorithm, and the current iteration number, and update the parameters of each initial date prediction model based on the genetic probability parameter. That is, the terminal may update the parameters of the inception date prediction model using a genetic algorithm. Since the individual is generally represented by a binary character string, the terminal can encode the parameter of the initial date prediction model to obtain the binary character string as the individual. Each individual in the initial population (population of individuals) corresponds to a parameter of an initial date prediction model. For example, if a three-layer BP neural network structure (input layer, hidden layer, and output layer) is used as the initial date prediction model, the number N1 of neurons in the input layer matches the number of pieces of historical task progress index data, and may be 5 (including the identifier fancode of the sub-target object) or 6 (including the identifier fancode of the sub-target object). The number N2 of the hidden layer neural network and the number N1 of the input layer neurons have an approximate relation: n1=2N1+1, if N1=6, the number of hidden layer neurons is: n2=2 + 6+1=13. The neurons of the output layer may be set to 1. The BP neural network structure thus set (as an initial date prediction model) can be as shown in fig. 3: the input layer has 6 nodes, the hidden layer has 13 nodes, and the output layer has 1 node. In fig. 3, 6 nodes of the input layer correspond to: the number M of completed task objects corresponding to the history date, the total number N of history task objects, the identification fancode of the sub-object objects, the history date, the date0 of starting execution of the history task and the date of opening the permission of the history task are obtained. The initial date prediction model involves 6+ 13+ 1=91weights and 13+ 1=14thresholds altogether. I.e. the number of parameters of the inception date prediction model is 105. The 105 parameters may be encoded into a string as an individual.

Specifically, the terminal may input the historical task progress indicator data into an input layer of the initial date prediction model shown in fig. 3, then process data transmitted from the input layer in the hidden layer, and transmit the result to the output layer. The output layer calculates the predicted output of the initial date prediction model according to the data transmitted from the hidden layer, that is, outputs the predicted completion date of the historical task (corresponding to C shown in fig. 3). Then, the terminal can calculate a loss value according to the prediction output C and the expected output (namely, the actual completion date of the historical task date 1), and if the loss value is larger than the expected value, the terminal can update the individual by adopting a genetic algorithm based on the genetic probability parameters, namely, parameters such as the weight values and the threshold values of the hidden layer and the output layer of each initial date prediction model are correspondingly updated. The genetic probability parameter may specifically include one or more of a selection probability parameter, a cross probability parameter, and a mutation probability parameter. The genetic probability parameter can be obtained by calculation according to the actual completion date of the historical tasks, the predicted completion dates of the multiple historical tasks, the genetic algorithm and the current iteration number, so that the genetic probability parameter changes with the increase of the iteration number, for example, the genetic probability parameter can be gradually reduced with the increase of the iteration number. The iteration number refers to the number of times of updating the model parameter by using a genetic algorithm, and when the genetic probability parameter is determined for the first time to update the parameter, the current iteration number is 1.

Then, the terminal may return to the execution of acquiring historical task progress index data and actual historical task completion dates corresponding to the target objects based on the initial date prediction model after the parameters are updated (i.e., the model after the parameters are updated is used as a new initial date prediction model) (step 201), and then input the historical task progress index data into a plurality of initial date prediction models (i.e., new initial date prediction models) (step 202) until a preset training stop condition is reached, so as to obtain a trained date prediction model. The preset training stop condition may be that the number of iterations reaches a preset value, or that an error (loss value) between a predicted completion date of the historical task and an actual completion date of the historical task, which is predicted, is smaller than a preset value. For example, if the training stop condition is that the number of iterations reaches a preset value, one of the initial date prediction models after updating the parameters in the last iteration may be randomly selected as the trained date prediction model, or a model with the smallest error value between the predicted completion date of the historical task and the actual completion date of the historical task in the initial date prediction models after updating the parameters in the last iteration may be used as the trained date prediction model.

The embodiment provides a training method of a date prediction model. In the method, a date prediction model is trained by adopting historical data of a historical task (such as a real estate inventory task) aiming at a target object, and parameters of the date prediction model are updated by adopting a genetic algorithm in the training process of the date prediction model, wherein genetic probability parameters are determined according to the current iteration times, the global convergence of the genetic algorithm can be improved, namely parameters of the model with a better global condition can be obtained, so that the prediction accuracy of the obtained date prediction model is higher, namely the task prediction completion date predicted by the date prediction model is closer to the actual task completion date, so that the task progress can be reflected more accurately, and a basis is provided for the subsequent management of a manager.

In one embodiment, as shown in fig. 4, the process of determining the genetic probability parameter in step 203 specifically includes the following steps:

step 401, for each inception date prediction model, calculating a loss value corresponding to the inception date prediction model according to an actual completion date of a historical task, a historical task prediction completion date corresponding to the inception date prediction model, and an objective function.

In an implementation, for each inception date prediction model, the terminal may calculate a loss value corresponding to the inception date prediction model according to an actual completion date of the historical task, the predicted completion date of the historical task corresponding to the inception date prediction model obtained in step 202, and an objective function. The objective function can adopt a mean square error loss function, and a calculation formula of a loss value is as follows:

where n (an integer greater than or equal to 1) represents the total number of training samples (each training sample includes historical task progress indicator data of a target object and corresponding historical task actual completion date), y _i Showing the predicted completion date of the historical task obtained by inputting the historical task progress index data in the ith training sample into the initial date prediction model,

indicating the actual completion date of the historical task in the ith training sample. The two dates are subtracted to obtain the number of days between the two dates.

Step 402, determining fitness values of individuals corresponding to the initial date prediction models according to the loss values corresponding to the initial date prediction models, and determining a genetic probability parameter based on the fitness values and the current iteration times.

In an implementation, the terminal may determine the fitness value of the individual corresponding to each inception date prediction model according to the loss value corresponding to each inception date prediction model. Specifically, each individual corresponds to one inception date prediction model, and the fitness value of each individual can be determined according to the loss value corresponding to each inception date prediction model. The terminal can directly determine the reciprocal of each loss value as the corresponding fitness value of each individual, or perform reciprocal operation after normalizing each loss value to obtain the fitness value, wherein the smaller the loss value is, the larger the fitness value is. Then, the terminal may determine the genetic probability parameter based on the fitness value and the current iteration number, so that the genetic probability parameter, such as the mutation probability or the cross probability of the individual, changes with the increase of the iteration number, for example, the genetic probability parameter may gradually decrease with the increase of the iteration number.

In this embodiment, the loss value corresponding to each inceptive date prediction model is calculated according to the actual completion date of the historical task, the historical task prediction completion date corresponding to the inceptive date prediction model, and the objective function, and then the fitness value of each individual is determined according to the loss value, so that the genetic probability parameter is determined based on the fitness value and the current iteration number. The genetic probability parameters can change from large to small along with the increase of the iteration times, so that the genetic probability parameters such as crossover probability and mutation probability are large at the initial stage of iteration to keep the diversity of the population, and the genetic probability parameters such as crossover probability and mutation probability become small at the middle and later stages of iteration to prevent the optimal solution from being damaged, so that the global convergence of the genetic algorithm is improved, and the accuracy of the trained date prediction model for the task completion date is improved.

In one embodiment, the genetic probability parameter includes a cross probability parameter, as shown in fig. 5, the process of determining the genetic probability parameter in step 402 specifically includes the following steps:

step 501, according to the fitness value of each individual, determining a maximum fitness value and an average fitness value.

In implementation, the terminal may determine, according to the fitness value (which may be denoted as f) of the individual corresponding to each inception date prediction model, a maximum fitness value (which may be denoted as f) with the largest value among the fitness values _max ) That is, the maximum fitness value of each individual in the current population is calculated, and the average fitness value (which can be written as f) of the current population (the population composed of the individuals corresponding to the current iteration number) is calculated _avg )。

Step 502, selecting a plurality of target individual pairs from each individual according to the fitness value of each individual and a preset selection strategy.

In implementation, the terminal may select a plurality of target individual pairs from each individual according to the fitness value of each individual and a preset selection policy. The preset selection strategy can be a roulette method, the probability of each body to be selected is calculated according to the fitness value, and the larger the fitness value of each body is, the larger the probability to be selected is. Then, based on the probability of each individual body being selected, a target individual body is selected from each individual body, and the number of the target individual bodies is the same as that of the individual bodies in the initial population. The selected individuals with higher probability may be selected multiple times, that is, the target individual may include the same individual. Then, the terminal may pair the target individuals two by two, for example, may pair randomly to obtain a plurality of target individual pairs.

Step 503, for each target individual pair, determining the maximum value of the fitness values of the two individuals in the target individual pair as a target fitness value, and calculating the cross probability parameter corresponding to the target individual pair according to the target fitness value, the maximum fitness value, the average fitness value and the current iteration number.

In practice, for each target individual pair selected in step 502, the terminal may adapt the two individuals in the target individual pairThe strain values are compared, and the maximum fitness value of the two is taken as a target fitness value (which can be recorded as f) ^* ). Then, the terminal may calculate a cross probability parameter (which may be denoted as Pc) corresponding to the target individual pair according to the target fitness value, the maximum fitness value, the average fitness value, and the current iteration number. And the terminal carries out cross operation on each target individual pair based on the cross probability parameter corresponding to each target individual. In one example, the cross probability parameter Pc is calculated as follows:

wherein n is the current iteration number, f _max Is the maximum fitness value, f _avg Is an average fitness value, f ^* Is the target fitness value. As can be seen from the curve of the exponential function, when the iteration generation number n is small,

the speed of the reduction of the value is higher, so that the cross probability is faster to approach the expected probability; and when the iteration algebra n is large,

the value will also be greater than 0, making the crossover probability vary within a reasonable range. Therefore, the diversity of the population can be kept at the initial stage of iteration, the local optimum can be quickly skipped out, so that the global optimum is better converged, the global convergence of the genetic algorithm is improved, and the prediction accuracy of the trained date prediction model is improved.

In this embodiment, the genetic probability parameters include cross probability parameters, and an implementation manner for calculating the cross probability parameters based on the current iteration number is provided, so that the cross probability parameters change from large to small with the increase of the iteration algebra, so as to improve the global convergence of the genetic algorithm, and thus improve the accuracy of the trained date prediction model in predicting the task completion date.

In one embodiment, the process of calculating the cross probability parameter corresponding to the target individual pair in step 503 specifically includes the following steps: under the condition that the target fitness value is greater than or equal to the average fitness value, calculating a cross probability parameter corresponding to the target individual pair according to the target fitness value, the maximum fitness value, the average fitness value and the current iteration times; and under the condition that the target adaptability value is smaller than the average adaptability value, calculating a cross probability parameter corresponding to the target individual pair according to the current iteration times.

In implementation, after determining the target fitness value of the target individual pair, the terminal may compare the target fitness value with the average fitness value, and then calculate the cross probability parameter. Specifically, if the target fitness value is greater than or equal to the average fitness value, the terminal may calculate a cross probability parameter corresponding to the target individual pair according to the target fitness value, the maximum fitness value, the average fitness value, and the current iteration number; and if the target fitness value is smaller than the average fitness value, the terminal calculates the cross probability parameter corresponding to the target individual pair according to the current iteration times. In one example, the cross probability parameter Pc is calculated as follows:

wherein n is the current iteration number, f _max Is the maximum fitness value, f _avg Is an average fitness value, f ^* Is the target fitness value.

The embodiment provides another implementation mode for calculating the cross probability parameter based on the current iteration number, so that the cross probability parameter changes from large to small along with the increase of the iteration algebra, the setting rationality of the cross probability parameter is considered, the global convergence of the genetic algorithm is improved, and the prediction accuracy of the trained date prediction model on the task completion date is improved.

In one embodiment, the genetic probability parameter includes a mutation probability parameter, as shown in fig. 6, the process of determining the genetic probability parameter in step 402 specifically includes the following steps:

step 601, determining a maximum fitness value and an average fitness value according to the fitness values of the individuals.

In implementation, the terminal may determine the maximum fitness value f with the largest value among the fitness values according to the fitness values f of the individuals corresponding to the initial date prediction models _max I.e. the maximum fitness value of each individual in the current population, and calculates the average fitness value f of the current population (the population consisting of the individuals corresponding to the current iteration number) _avg 。

Step 602, selecting a plurality of target individuals from each individual according to the fitness value of each individual and a preset selection strategy.

In implementation, the terminal may select a plurality of target individuals from each individual according to the fitness value of each individual and a preset selection policy. The preset selection strategy can be a roulette method, the probability of each body to be selected is calculated according to the fitness value, and the larger the fitness value of each body is, the larger the probability to be selected is. And then selecting a plurality of target individuals from the individuals on the basis of the selected probability of each individual, wherein the number of the target individuals is the same as that of the individuals in the initial population, and one target individual corresponds to one initial date prediction model.

Step 603, for each target individual, calculating a variation probability parameter corresponding to the target individual according to the fitness value, the maximum fitness value, the average fitness value and the current iteration number of the target individual.

In implementation, for the target individuals selected in step 602, the terminal calculates a variation probability parameter (which may be denoted as Pm) corresponding to each target individual with respect to the fitness value, the maximum fitness value, the average fitness value, and the current iteration count of each target individual. And the terminal performs mutation operation on each target individual based on the mutation probability parameter of each target individual. In one example, the calculation formula of the variation probability parameter Pm is as follows:

wherein n is the current iteration number, f _max Is the maximum fitness value, f _avg Is the average fitness value, and f is the fitness value of the target individual.

In this embodiment, the genetic probability parameter includes a mutation probability parameter, and an implementation manner for calculating the mutation probability parameter based on the current iteration number is provided, so that the mutation probability parameter changes from large to small with the increase of the iteration algebra, so as to improve the global convergence of the genetic algorithm, and thus improve the accuracy of the trained date prediction model in predicting the task completion date.

In one embodiment, the process of calculating the variation probability parameter corresponding to the target individual in step 603 includes the following steps: under the condition that the fitness value of the target individual is greater than or equal to the average fitness value, calculating a variation probability parameter corresponding to the target individual according to the fitness value, the maximum fitness value, the average fitness value and the current iteration times of the target individual; and under the condition that the fitness value of the target individual is smaller than the average fitness value, calculating a variation probability parameter corresponding to the target individual according to the current iteration times.

In implementation, the terminal may compare the fitness value of each target individual with the average fitness value, and further calculate a variation probability parameter of each target individual. Specifically, if the fitness value of the target individual is greater than or equal to the average fitness value, the terminal may calculate a variation probability parameter of the target individual according to the fitness value, the maximum fitness value, the average fitness value, and the current iteration number of the target individual; if the fitness value of the target individual is smaller than the average fitness value, the terminal can calculate the variation probability parameter of the target individual according to the current iteration times. In one example, the calculation formula of the mutation probability parameter Pm is as follows:

The embodiment provides another implementation mode for calculating the variation probability parameter based on the current iteration number, so that the variation probability parameter changes from large to small along with the increase of the iteration algebra, the setting rationality of the variation probability parameter is considered, the global convergence of the genetic algorithm is improved, and the prediction accuracy of the trained date prediction model on the task completion date is improved.

The application also provides an example of a training process of the date prediction model, in the example, the genetic probability parameters include a cross probability parameter and a variation probability parameter, and the selection operation of the genetic algorithm can be performed according to a preset selection strategy, and the method specifically comprises the following steps:

step 701, obtaining historical task progress index data and historical task actual completion date corresponding to the target object.

The historical task progress index data comprise the total number of the historical task objects, the historical date, the number of the completed task objects corresponding to the historical date, the historical task permission starting date and the historical task starting execution date.

Step 702, inputting the historical task progress index data into a plurality of initial date prediction models to obtain a plurality of historical task prediction completion dates.

The parameters of each inception date prediction model are different.

And 703, calculating a loss value corresponding to the initial date prediction model according to the actual completion date of the historical task, the historical task prediction completion date corresponding to the initial date prediction model and the objective function for each initial date prediction model.

And step 704, determining the fitness value of the individual corresponding to each initial date prediction model according to the loss value corresponding to each initial date prediction model.

Step 705, determining a maximum fitness value and an average fitness value according to the fitness value of each individual.

And 706, selecting a plurality of target individuals from each individual according to the fitness value of each individual and a preset selection strategy.

Wherein a target individual corresponds to an initial date prediction model.

And 707, pairing the multiple target individuals to obtain multiple target individual pairs, determining a maximum value of the fitness values of the two individuals in each target individual pair as a target fitness value, and calculating a cross probability parameter corresponding to each target individual pair according to the target fitness value, the maximum fitness value, the average fitness value and the current iteration number.

Wherein the current iteration number is counted from 1.

Step 708, for each target individual, calculating a variation probability parameter corresponding to the target individual according to the fitness value, the maximum fitness value, the average fitness value and the current iteration number of the target individual.

And step 709, updating the parameters of each initial date prediction model based on the genetic probability parameters.

Specifically, the crossover operation may be performed on each target individual pair based on a crossover probability parameter corresponding to each target individual pair, and then the mutation operation may be performed on each target individual after the crossover operation based on a mutation probability parameter corresponding to each target individual. Each target individual after the mutation operation corresponds to the parameters of the updated model. One target individual corresponds to one model. The target individuals after each mutation operation can be decoded to obtain parameters such as the weight and the threshold of each model, and the parameters of each model are updated.

Step 710, determining whether a preset training stop condition is reached.

Specifically, the preset training stop condition may be whether the iteration number reaches a preset value. If the stopping condition is not met, the model with each updated parameter is used as a plurality of new initial date prediction models (it can be understood that the parameters of the plurality of new initial date prediction models may be the same), the process returns to step 701, and the current iteration number is +1. If the stop condition is reached, step 711 is executed.

And 711, inputting the test sample set to each model with updated parameters to obtain a historical task prediction completion date corresponding to each model with updated parameters, wherein each sample in the test sample set comprises historical task progress index data corresponding to the target object and an actual historical task completion date, calculating a loss value corresponding to each model with updated parameters according to the historical task prediction completion date and the actual historical task completion date, and taking the model with the updated parameters with the minimum loss value as a trained date prediction model.

It should be understood that, although the steps in the flowcharts related to the embodiments described above are shown in sequence as indicated by the arrows, the steps are not necessarily performed in sequence as indicated by the arrows. The steps are not performed in the exact order shown and described, and may be performed in other orders, unless explicitly stated otherwise. Moreover, at least a part of the steps in the flowcharts related to the embodiments described above may include multiple steps or multiple stages, which are not necessarily performed at the same time, but may be performed at different times, and the execution order of the steps or stages is not necessarily sequential, but may be rotated or alternated with other steps or at least a part of the steps or stages in other steps.

Based on the same inventive concept, the embodiment of the application also provides a date prediction device for realizing the date prediction method. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the method, so specific limitations in one or more embodiments of the date prediction device provided below can be referred to the limitations of the date prediction method in the foregoing, and details are not described herein again.

In one embodiment, as shown in fig. 7, there is provided a date prediction apparatus 700 comprising: a first obtaining module 701 and a predicting module 702, wherein:

a first obtaining module 701, configured to obtain current task progress indicator data corresponding to a target object; the current task progress index data comprises the total number of task objects, the current date, the number of the currently completed task objects, the task permission starting date and the task starting execution date.

The prediction module 702 is configured to input the current task progress indicator data into the date prediction model to obtain a task prediction completion date of the target object; the date prediction model is obtained by processing historical task progress index data corresponding to a target object based on an initial date prediction model to obtain historical task prediction completion dates, determining genetic probability parameters according to the actual completion dates of the historical tasks corresponding to the target object, the historical task prediction completion dates, a genetic algorithm and the current iteration times by taking parameters of the initial date prediction model as individuals, and updating the parameters of the initial date prediction model based on the genetic probability parameters.

In one embodiment, the apparatus further comprises: second acquisition module, input module and update module, wherein:

the second acquisition module is used for acquiring historical task progress index data and historical task actual completion dates corresponding to the target object; the historical task progress index data comprises the total number of historical task objects, historical date, the number of completed task objects corresponding to the historical date, historical task permission starting date and historical task starting execution date.

The input module is used for inputting the historical task progress index data into the plurality of initial date prediction models to obtain a plurality of historical task prediction completion dates; the parameters of each inception date prediction model are different.

And the updating module is used for determining a genetic probability parameter according to the actual completion date of the historical tasks, the predicted completion dates of the historical tasks, the genetic algorithm and the current iteration number by taking the parameters of the initial date prediction model as individuals, updating the parameters of each initial date prediction model based on the genetic probability parameter, and executing the steps of obtaining the historical task progress index data and the actual completion date of the historical tasks corresponding to the target object until a preset training stopping condition is reached to obtain the trained date prediction model.

In one embodiment, the update module is specifically configured to: aiming at each initial date prediction model, calculating a loss value corresponding to the initial date prediction model according to the actual completion date of the historical task, the predicted completion date of the historical task corresponding to the initial date prediction model and an objective function; and determining the fitness value of the individual corresponding to each initial date prediction model according to the loss value corresponding to each initial date prediction model, and determining the genetic probability parameter based on the fitness value and the current iteration times.

In one embodiment, the genetic probability parameter comprises a crossover probability parameter. The update module is specifically configured to: determining a maximum fitness value and an average fitness value according to the fitness value of each individual; selecting a plurality of target individual pairs from each individual according to the fitness value of each individual and a preset selection strategy; and determining the maximum value of the fitness values of the two individuals in each target individual pair as a target fitness value, and calculating the cross probability parameter corresponding to the target individual pair according to the target fitness value, the maximum fitness value, the average fitness value and the current iteration times.

In one embodiment, the update module is specifically configured to: under the condition that the target fitness value is greater than or equal to the average fitness value, calculating a cross probability parameter corresponding to the target individual pair according to the target fitness value, the maximum fitness value, the average fitness value and the current iteration times; and under the condition that the target fitness value is smaller than the average fitness value, calculating a cross probability parameter corresponding to the target individual pair according to the current iteration times.

In one embodiment, the genetic probability parameter comprises a mutation probability parameter. The update module is specifically configured to: determining a maximum fitness value and an average fitness value according to the fitness value of each individual; selecting a plurality of target individuals from each individual according to the fitness value of each individual and a preset selection strategy; a target individual corresponds to an initial date prediction model; and calculating a variation probability parameter corresponding to each target individual according to the fitness value, the maximum fitness value, the average fitness value and the current iteration number of the target individual.

In one embodiment, the update module is specifically configured to: under the condition that the fitness value of the target individual is greater than or equal to the average fitness value, calculating a variation probability parameter corresponding to the target individual according to the fitness value, the maximum fitness value, the average fitness value and the current iteration times of the target individual; and under the condition that the fitness value of the target individual is smaller than the average fitness value, calculating a variation probability parameter corresponding to the target individual according to the current iteration times.

The modules in the date prediction device can be wholly or partially implemented by software, hardware and a combination thereof. The modules can be embedded in a hardware form or independent from a processor in the computer device, and can also be stored in a memory in the computer device in a software form, so that the processor can call and execute operations corresponding to the modules.

In one embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as shown in fig. 8. The computer device includes a processor, a memory, a communication interface, a display screen, and an input device connected by a system bus. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device comprises a nonvolatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operating system and the computer program to run on the non-volatile storage medium. The communication interface of the computer device is used for carrying out wired or wireless communication with an external terminal, and the wireless communication can be realized through WIFI, a mobile cellular network, NFC (near field communication) or other technologies. The computer program is executed by a processor to implement a method of date prediction. The display screen of the computer equipment can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer equipment can be a touch layer covered on the display screen, a key, a track ball or a touch pad arranged on the shell of the computer equipment, an external keyboard, a touch pad or a mouse and the like.

Those skilled in the art will appreciate that the architecture shown in fig. 8 is merely a block diagram of some of the structures associated with the disclosed aspects and is not intended to limit the computing devices to which the disclosed aspects apply, as particular computing devices may include more or less components than those shown, or may combine certain components, or have a different arrangement of components.

In one embodiment, a computer device is provided, comprising a memory and a processor, the memory having stored therein a computer program, the processor implementing the steps of the above-described method embodiments when executing the computer program.

In an embodiment, a computer-readable storage medium is provided, on which a computer program is stored, which computer program, when being executed by a processor, carries out the steps of the above-mentioned method embodiments.

In an embodiment, a computer program product is provided, comprising a computer program which, when being executed by a processor, carries out the steps of the above-mentioned method embodiments.

The date prediction method, the date prediction device, the computer equipment, the storage medium and the computer program product relate to the technical field of artificial intelligence, can be used in the field of financial science and technology or other fields, and are not limited in application fields.

It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) referred to in the present application are information and data authorized by the user or sufficiently authorized by each party.

It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above may be implemented by hardware instructions of a computer program, which may be stored in a non-volatile computer-readable storage medium, and when executed, may include the processes of the embodiments of the methods described above. Any reference to memory, databases, or other media used in the embodiments provided herein can include at least one of non-volatile and volatile memory. The nonvolatile Memory may include a Read-Only Memory (ROM), a magnetic tape, a floppy disk, a flash Memory, an optical Memory, a high-density embedded nonvolatile Memory, a resistive Random Access Memory (ReRAM), a Magnetic Random Access Memory (MRAM), a Ferroelectric Random Access Memory (FRAM), a Phase Change Memory (PCM), a graphene Memory, and the like. Volatile Memory can include Random Access Memory (RAM), external cache Memory, and the like. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM), among others. The databases referred to in various embodiments provided herein may include at least one of relational and non-relational databases. The non-relational database may include, but is not limited to, a block chain based distributed database, and the like. The processors referred to in the embodiments provided herein may be general purpose processors, central processing units, graphics processors, digital signal processors, programmable logic devices, quantum computing based data processing logic devices, etc., without limitation.

The technical features of the above embodiments can be arbitrarily combined, and for the sake of brevity, all possible combinations of the technical features in the above embodiments are not described, but should be considered as the scope of the present specification as long as there is no contradiction between the combinations of the technical features.

The above-mentioned embodiments only express several embodiments of the present application, and the description thereof is specific and detailed, but not construed as limiting the scope of the present application. It should be noted that, for a person skilled in the art, several variations and modifications can be made without departing from the concept of the present application, and these are all within the scope of protection of the present application. Therefore, the protection scope of the present application shall be subject to the appended claims.

Claims

1. A method of date prediction, the method comprising:

the date prediction model is obtained by processing historical task progress index data corresponding to the target object based on an initial date prediction model to obtain a historical task prediction completion date, determining a genetic probability parameter according to the actual completion date of the historical task corresponding to the target object, the historical task prediction completion date, a genetic algorithm and the current iteration frequency by taking parameters of the initial date prediction model as individuals, and updating the parameters of the initial date prediction model based on the genetic probability parameter.

2. The method of claim 1, wherein the training process of the date prediction model comprises:

acquiring the historical task progress index data and the actual completion date of the historical task corresponding to the target object; the historical task progress index data comprise the total number of historical task objects, historical dates, the number of completed task objects corresponding to the historical dates, historical task permission opening dates and historical task starting execution dates;

inputting the historical task progress index data into a plurality of initial date prediction models to obtain a plurality of historical task prediction completion dates; wherein the parameters of each of the initial date prediction models are different;

and taking the parameters of the initial date prediction model as individuals, determining the genetic probability parameters according to the actual completion date of the historical tasks, the predicted completion dates of the historical tasks, the genetic algorithm and the current iteration number, updating the parameters of each initial date prediction model based on the genetic probability parameters, and executing the steps of obtaining the historical task progress index data and the actual completion date of the historical tasks corresponding to the target object until a preset training stop condition is reached to obtain the trained date prediction model.

3. The method of claim 2, wherein determining the genetic probability parameter based on the historical task actual completion dates, the plurality of historical task predicted completion dates, the genetic algorithm, and the current number of iterations comprises:

aiming at each initial date prediction model, calculating a loss value corresponding to the initial date prediction model according to the actual completion date of the historical task, the historical task prediction completion date corresponding to the initial date prediction model and an objective function;

and determining the fitness value of the individual corresponding to each initial date prediction model according to the loss value corresponding to each initial date prediction model, and determining the genetic probability parameter based on the fitness value and the current iteration number.

4. The method of claim 3, wherein the genetic probability parameter comprises a crossover probability parameter; determining the genetic probability parameter based on the fitness value and the current iteration number comprises:

5. The method of claim 4, wherein the calculating the cross probability parameter corresponding to the target individual pair according to the target fitness value, the maximum fitness value, the average fitness value, and the current iteration number comprises:

6. The method of claim 3, wherein the genetic probability parameter comprises a mutation probability parameter; determining the genetic probability parameter according to the fitness value corresponding to each individual and the current iteration number, wherein the determining comprises:

7. The method of claim 6, wherein the calculating the variation probability parameter corresponding to the target individual according to the fitness value, the maximum fitness value, the average fitness value, and the current iteration number of the target individual comprises:

8. A date prediction apparatus, characterized in that the apparatus comprises:

the acquisition module is used for acquiring current task progress index data corresponding to the target object; the current task progress index data comprises the total number of the task objects, the current date, the number of the currently completed task objects, the task permission starting date and the task starting execution date;

the prediction module is used for inputting the current task progress index data into a date prediction model to obtain the task prediction completion date of the target object; the date prediction model is obtained by processing historical task progress index data corresponding to the target object based on an initial date prediction model to obtain a historical task prediction completion date, determining a genetic probability parameter according to the actual completion date of the historical task corresponding to the target object, the historical task prediction completion date, a genetic algorithm and the current iteration number by taking parameters of the initial date prediction model as individuals, and updating the parameters of the initial date prediction model based on the genetic probability parameter.

9. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that the processor, when executing the computer program, implements the steps of the method of any of claims 1 to 7.

10. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out the steps of the method of any one of claims 1 to 7.

11. A computer program product comprising a computer program, characterized in that the computer program realizes the steps of the method of any one of claims 1 to 7 when executed by a processor.