CN112613997A

CN112613997A - Method and apparatus for forecasting combined investment of money fund

Info

Publication number: CN112613997A
Application number: CN202011492671.6A
Authority: CN
Inventors: 吴波
Original assignee: Ping An Consumer Finance Co Ltd
Current assignee: Ping An Consumer Finance Co Ltd
Priority date: 2020-12-17
Filing date: 2020-12-17
Publication date: 2021-04-06

Abstract

The invention relates to the technical field of intelligent decision making, and aims to improve the prediction accuracy of fund combination investment. The invention provides a combined investment prediction method and a prediction device of a money fund, wherein the method comprises the following steps: acquiring first historical performance data of a plurality of funds, and respectively determining the predicted income of each fund by using an income prediction model; acquiring first historical operating characteristics of a target merchant, and determining predicted floating funds of the merchant by using a floating fund prediction model; obtaining current operation characteristics of a target merchant in a plurality of different time periods, and respectively taking each current operation characteristic, the predicted income and the predicted liquidity as decision factors of a random forest model to obtain candidate fund combinations of the target merchant in the different time periods; wherein every two adjacent time intervals contain a crossing time interval; and determining a recommended fund combination according to the occurrence density of different funds contained in the plurality of candidate fund combinations.

Description

Method and apparatus for forecasting combined investment of money fund

Technical Field

The invention relates to the technical field of intelligent decision, in particular to a combined investment prediction method and a combined investment prediction device for a money fund.

Background

Random Forest models (Random Forest) are widely applied in various fields such as medical treatment, finance, meteorology and the like at present. The random forest model can process a large number of input variables and evaluate the importance of the variables, so that classification decision is realized.

In the financial field, random forest models are commonly used to make financial product investment decisions. Because the investment products relate to a large number of characteristic parameters and the characteristic parameters have strong timeliness, the combined investment products predicted by the random forest models in a short time interval may come in and go out greatly, the combined investment products are not ideal in stability or accuracy, and actually the reference value provided by the existing random forest models for investors is not large.

Therefore, how to provide more accurate and reliable combined investment prediction for investors becomes a technical problem to be urgently solved by the technical personnel in the field.

Disclosure of Invention

The invention aims to provide a combined investment prediction scheme capable of improving the prediction accuracy and stability so as to solve the problems in the prior art.

To achieve the above object, the present invention provides a method for forecasting a portfolio investment of a money fund, comprising:

acquiring first historical performance data of a plurality of funds, and respectively determining the predicted income of each fund by using an income prediction model;

acquiring first historical operating characteristics of a target merchant, and determining predicted floating funds of the merchant by using a floating fund prediction model;

obtaining the current operation characteristics of a target merchant, and taking the current operation characteristics, the predicted income and the predicted liquidity as decision factors of a random forest model to obtain a candidate fund combination corresponding to the target merchant;

and acquiring a plurality of candidate fund combinations, and determining a recommended fund combination based on the occurrence density of different funds contained in the plurality of candidate fund combinations.

According to the combined investment prediction method of the money fund, the step of acquiring historical performance data of a plurality of funds and respectively determining the predicted income of each fund by using an income prediction model comprises the following steps:

acquiring any one or more of newly-increased purchase quantity, newly-increased redemption quantity, Shanghai 300 index, bank daytime borrowing rate, fund company financial statement, fund company establishment duration and similar fund ranking corresponding to each fund as the historical performance data;

inputting the historical performance data into a profit prediction model to output the predicted profit of each fund in a preset time period; wherein the revenue prediction model is trained by a linear regression model.

According to the forecasting method for the combined investment of the money fund, the step of inputting the historical performance data into a profit forecasting model to output the forecasting profit of each fund in the preset time period comprises the following steps:

dividing the historical performance data into working day historical performance data and resting calendar historical performance data;

inputting the performance data of the working calendar history into the income prediction model so as to output the predicted working day income of the fund in a preset time period;

inputting the weekday historical performance data into the revenue prediction model to output predicted weekday revenue of the fund over a preset time period;

and weighting and summing the predicted weekday benefits and the predicted weekday benefits to obtain the predicted benefits.

According to the combined investment prediction method of the money fund provided by the invention, the step of obtaining the historical operating characteristics of the target merchant and determining the predicted liquidity of the merchant by using a liquidity prediction model comprises the following steps:

acquiring any one or more of historical sale amount, historical price, holiday factor, historical refund proportion, monthly and annual rate withdrawal amount, monthly and annual rate expenditure amount, promotion factor and purchase factor of the main operation commodity corresponding to the target merchant as the historical operation characteristic;

inputting the historical operating characteristics into a liquidity forecasting model to output the forecasted liquidity of the merchant; wherein the liquidity forecasting model is trained based on an autocorrelation model.

According to the forecasting method for the combined investment of the money fund, the step of obtaining a plurality of candidate fund combinations and determining the recommended fund combination based on the occurrence density of different funds contained in the plurality of fund combinations comprises the following steps:

acquiring fund numbers contained in each current candidate fund combination;

calculating the occurrence density of each fund number in all current candidate fund combinations, wherein the occurrence density is the proportion of the fund number in all fund numbers;

and determining a recommended fund combination according to the occurrence density.

According to the combined investment prediction method of the money fund, the training process of the profit prediction model comprises the following steps:

obtaining second historical performance data for the plurality of funds, the second historical performance data comprising performance data for the plurality of funds over a first time period prior to a training time point;

dividing the second historical performance data into a plurality of different training samples and testing samples, each testing sample including performance data for one of the funds over a second time period; the second time period is less than the first time period;

performing mathematical operation on all numerical values contained in each training sample to obtain operation sample data, wherein the mathematical operation comprises any of summation, averaging, variance, maximum and minimum;

and adding the operation sample data into the training sample to obtain a new training sample, training the profit prediction model by using the new sample data, and testing by using the test sample.

According to the combined investment prediction method of the money fund provided by the invention, the training process of the predicted fund flow model comprises the following steps:

acquiring second historical operation data of the target merchant, wherein the second historical operation data comprises the historical operation data of the target merchant in a first time period before a training time point;

dividing the second historical operating data into a plurality of different training samples, each training sample comprising performance data of one of the funds within a second time period; the second time period is less than the first time period;

removing the maximum value data and the minimum value data contained in the training sample to obtain a new training sample;

training the liquidity forecasting model with the new sample data.

To achieve the above object, the present invention also provides a combination investment prediction apparatus for a money fund, comprising:

the profit prediction module is suitable for acquiring first historical performance data of a plurality of funds and determining the predicted profit of each fund by using the profit prediction model;

the mobile fund prediction module is suitable for acquiring first historical operating characteristics of a target merchant and determining the predicted mobile fund of the merchant by using a mobile fund prediction model;

the random forest module is suitable for obtaining the current operation characteristics of a target merchant, and the current operation characteristics, the predicted income and the predicted liquidity are used as decision factors of a random forest model to obtain a candidate fund combination corresponding to the target merchant;

and the recommending module is used for acquiring a plurality of candidate fund combinations and determining the recommended fund combination based on the occurrence density of different funds contained in the plurality of candidate fund combinations.

To achieve the above object, the present invention further provides a computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the steps of the above method when executing the computer program.

To achieve the above object, the present invention also provides a computer-readable storage medium having stored thereon a computer program which, when being executed by a processor, carries out the steps of the above method.

According to the combined investment prediction method and the prediction device for the money fund, the prediction benefits of the multiple funds are determined through the benefit prediction model, the prediction liquidity of the target merchant is predicted through the liquidity prediction model, and the prediction benefits of the multiple funds and the prediction liquidity of the target merchant are used as decision factors of the random forest model so as to obtain the candidate fund combinations corresponding to the target merchant in different time periods. In order to improve the accuracy of prediction, the recommended fund combination with the highest probability is determined according to the occurrence density of each fund on the basis of acquiring a plurality of candidate fund combinations. The scheme can ensure that the output recommended fund combination is closer to the real fund operation condition, thereby improving the accuracy of the prediction recommendation.

Drawings

FIG. 1 is a flow chart of a first embodiment of a method for forecasting the combined investment of a monetary fund according to the present invention;

FIG. 2A shows a schematic flow diagram of training a revenue prediction model, according to an embodiment of the invention;

FIG. 2B shows a schematic flow chart for partitioning a training sample and a test sample, according to an embodiment of the invention;

FIG. 3A shows a schematic flow diagram for establishing an ARIMA model according to an embodiment of the invention;

FIG. 3B shows a schematic flow diagram of training a liquidity forecasting model in accordance with an embodiment of the invention;

FIG. 4 shows a schematic flow chart diagram of determining a recommended fund combination according to an embodiment of the present invention;

FIG. 5 shows a schematic flow diagram of training a random forest model according to an embodiment of the invention;

FIG. 6 is a schematic diagram of program modules of a first embodiment of the portfolio investment prediction device of the present invention;

fig. 7 is a schematic diagram showing a hardware configuration of a first embodiment of the portfolio investment predicting apparatus of the present invention.

Detailed Description

In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention 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 invention and are not intended to limit the invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

Example one

The combined investment prediction method and the combined investment prediction device of the money fund can be applied to terminals or servers. The terminal can comprise intelligent equipment such as a smart phone, a notebook computer and a tablet computer, and the server can comprise a PC (personal computer), a workgroup server, an enterprise-level server and the like. Referring to fig. 1, the present embodiment provides a method for forecasting the portfolio investment of a money fund, comprising the following steps:

and S100, acquiring first historical performance data of a plurality of funds, and respectively determining the predicted income of each fund by using an income prediction model.

The first historical performance data may include any of a newly-added purchase quantity, a newly-added redemption quantity, a Shanghai depth 300 index, a bank daytime borrowing rate, a fund company financial report, a fund company establishment duration and a similar fund ranking corresponding to each fund in the first time period. The first time period may be set according to different needs, such as the last two days, the last week, or the last month, etc.

The revenue prediction model in this embodiment may be obtained by training a linear regression model. For the first historical performance data for each fund entered, the revenue prediction model may output revenue data, such as ten thousand revenue rates, for the fund after the second time period. The specific value of the second time period is related to the training sample of the profit prediction model, and can be selected for one week.

S200, acquiring a first historical operating characteristic of a target merchant, and determining the predicted liquidity of the merchant by using a liquidity prediction model.

The first historical operation data may include any of historical sale amount, historical price, holiday factors, historical refund proportion, monthly-to-annular withdrawal amount, monthly-to-monthly same-proportion expenditure amount, sales promotion factors and purchase factors of the main operation commodities corresponding to the target merchant in the first time period.

The liquidity forecasting model in the embodiment may be obtained by training an autoregressive model. For the first historical operating data of the target merchant, the liquidity forecasting model may output liquidity funds required by the merchant after a second time period. The specific value of the second time period is related to a training sample of the liquidity forecasting model, and can be generally selected to be one week.

S300, obtaining current operation characteristics of a target merchant in a plurality of different time periods, and respectively taking each current operation characteristic, the predicted income and the predicted liquidity as decision factors of a random forest model to obtain candidate fund combinations of the target merchant in the different time periods; wherein every two adjacent time intervals contain a crossing time interval.

The random forest model in this embodiment finally outputs whether a decision-making target merchant should make an application for a certain monetary fund and purchase or redeem shares by constructing a plurality of two-classification decision trees. The predicted profit of each fund obtained in step S100 and the predicted liquidity of the target merchant obtained in step S200 are both used as decision factors of the classification decision tree. In addition, the decision factor of the random forest model also comprises the current operation characteristics of the target merchants in a plurality of different time periods. When the money fund contains a plurality of funds, the random forest model respectively outputs whether the fund should be purchased and the purchase applying share for each fund, and all the funds determined as purchase applying are collected to form the candidate fund combination of the embodiment.

It will be appreciated that the output of the random forest model is related to a plurality of decision factors. If the decision factor is stable, the output of the random forest model is relatively stable; if the decision factor is unstable, the output of the random forest model will also be relatively unstable. When the output of the random forest model is obviously different along with the change of the decision factor, the reference value of the output result is not high. In order to improve the classification accuracy of the random forest model, the current operation characteristics of the target merchant in a plurality of different time periods are used as decision factors.

The plurality of different periods may be time regions adjacent to the first period of time and having a duration less than the first period of time. For example, the first time period is measured in weeks, and the specific dates are assumed to be from 10 months 11 days to 10 months 17 days for a period of seven days. The plurality of different time periods may be time regions having a duration of less than seven days and adjacent to 10 months 11 to 10 months 17, for example, a plurality of time periods having a duration of 3 days may include a time period T1(10 months 10 to 10 months 12), a time period T2(10 months 13 to 10 months 15) and a time period T3(10 months 16 to 10 months 18). And taking the different time periods as decision factors of the random forest model respectively, and accordingly obtaining three candidate fund combinations.

And S400, determining a recommended fund combination according to the occurrence density of different funds contained in the candidate fund combinations.

As previously mentioned, when selecting the target current business characteristics at different time periods as decision factors, the candidate fund combinations obtained by the random forest model may be different. In different sets of fund combinations, several funds with higher occurrence density are selected for recombination in the embodiment to serve as recommended fund combinations.

Because the recommended fund combination comprehensively considers decision factors generated among a plurality of adjacent time periods and is determined according to the high-to-low sequence of fund product occurrence density, the recommended fund combination has higher accuracy and reliability compared with the candidate fund combination.

As described above, the profit prediction model in the present embodiment may be obtained by training a linear regression model. Wherein the linear regression model is preferably a cable model in a penalized linear regression model. The first historical performance data for the fund is input into a linear regression model, and the predicted revenue of the fund over a preset time period can be output. It will be appreciated that the performance of the fund product on weekdays and on holidays will vary due to the varying degrees of concern. In order to predict fund returns more accurately, the embodiment first divides the first historical performance data into the working day historical performance data and the resting calendar historical performance data, and then inputs the working calendar historical performance data and the resting calendar historical performance data into the linear regression model respectively so as to obtain the predicted working day returns P1 and the predicted resting day returns P2 of the fund in a preset time period. Finally, the predicted weekday benefits P1 and the predicted weekday benefits P2 are weighted and summed to obtain a composite predicted benefit P. P may be represented by the formula:

P＝αP1+βP2

in the above formula, α and β are weighting coefficients, and may be specifically determined according to the ratio of the working day to the rest day in the first time period.

FIG. 2A shows a schematic flow diagram of training a revenue prediction model, according to an embodiment of the invention. As shown in FIG. 2, the training process of the revenue prediction model includes the following steps:

and S210, acquiring second historical performance data of the plurality of funds, wherein the second historical performance data comprises the performance data of the plurality of funds in a first time period before the training time point. In one example, the first time period may be three months.

S220, dividing the second historical performance data into a plurality of different training samples, wherein each testing sample comprises the performance data of one fund in a second time period; the second time period is less than the first time period.

In one example, the second time period is 2 weeks. At this time, the first period of 3 months was 6 times the second period of 2 weeks. Assuming a second historical performance data of a total of n funds, 6n training samples are available for each fund accordingly.

And S230, performing mathematical operation on all numerical values contained in each training sample to obtain operation sample data, wherein the mathematical operation comprises any of summation, averaging, variance, maximum and minimum.

Assuming that data originally contained in the training sample is [ x1, x2, … … xn ], the data originally contained is respectively subjected to operations such as summation, averaging, variance, maximum and minimum, and the like, so that y1, y2 and … … ym can be obtained.

And S240, adding the operation sample data into the training sample to obtain a new training sample, and training the profit prediction model by using the new sample data.

According to the example above, the new sample data is [ x1, x2, … … xn, y1, y2, … … ym ]. Through the steps, the diversity of the training samples can be increased, and the prediction accuracy of the profit prediction model is improved.

The test sample in this embodiment may be from historical performance data in the same period as the training sample, and 80% of the historical performance data is randomly selected as a model training sample and the remaining 20% is used as a test sample during each training. Specifically, as shown in fig. 2B, the process of dividing the training sample and the test sample is as follows:

and S210', calculating a time point of the barreled data according to the current time point. The sub-bucket data refers to data in the second time period, and the sub-bucket data time point corresponds to the specific duration of each second time period.

S220': and respectively acquiring the bucket dividing data according to the bucket dividing data time points.

S230': and judging whether the time point of the acquired sub-bucket data is within a preset first time period.

S240': if the time point of the data in the sub-bucket is within the first time period, a random number between 0 and 1 is sequentially generated for the data in the sub-bucket, and the random number may include any one of 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9, and 1.0.

S250': judging whether the random number is less than or equal to 0.8; if so, taking the corresponding sub-bucket data as a training set; and if not, taking the corresponding sub-bucket data as a test set.

S260': and if the time point of the sub-bucket data is out of the first time period, taking the corresponding sub-bucket data as a test set.

S270': and circularly acquiring data until all the sub-bucket data are acquired.

As described above, the liquidity prediction model in the present embodiment may be obtained by training an autoregressive model. Wherein the autoregressive model is preferably an ARIMA model. And inputting the first historical operation data of the target merchant into the autoregressive model, and outputting the predicted liquidity of the target merchant in a preset time period.

The modeling process of the differential autoregressive moving average model ARIMA mainly comprises determining parameters (p, d, q). The order d of the difference can be generally selected to be 1 or 2 according to the waveform diagram, p and q are determined by the attenuation mode of the autocorrelation function ACF and the partial autocorrelation function PACF, p represents that the p number of the partial autocorrelation function PACF presents truncation attenuation, and q represents that the q number of the autocorrelation function ACF presents truncation attenuation. As shown in fig. 3A, the specific steps of establishing the ARIMA model in this embodiment are as follows:

s310', acquiring training sample data of the target merchant;

s320', calculating an autocorrelation function ACF and a partial autocorrelation function PACF according to training sample data;

s330 ', determining whether differential processing is needed according to the balance test result, if yes, turning to the step S340'; if not, go to step S350';

s340': carrying out difference processing;

s350': modeling according to an autocorrelation function ACF and a partial autocorrelation function PACF;

s360': estimating model parameters;

s370': calculating a residual error statistic value;

s380': checking whether the residual statistic is appropriate; if so, finishing the model training; if not, the model is optimized, and the process returns to step S350'.

FIG. 3B shows a schematic flow diagram of training a liquidity fund prediction model, according to an embodiment of the invention. As shown in fig. 3B, the step of training the liquidity forecasting model comprises:

s310, obtaining second historical operation data of the target merchant, wherein the second historical operation data comprises performance data of the multiple funds in a first time period before the training time point.

S320, dividing the second historical operation data into a plurality of different training samples, wherein each training sample comprises the performance data of one fund in a second time period; the second time period is less than the first time period.

The steps S310 and S320 are similar to the steps S210 and S220, and are not described herein again.

And S330, removing the maximum value data and the minimum value data contained in the training sample to obtain a new training sample.

And S340, training the liquidity forecasting model by using the new sample data.

Through the steps, the smoothness of the training samples of the liquidity forecasting model can be enhanced, and therefore the forecasting accuracy of the liquidity forecasting model is improved.

FIG. 4 shows a schematic flow chart of determining a recommended fund combination according to an embodiment of the invention. As shown in fig. 5, step S400 includes:

and S410, acquiring fund numbers contained in each current candidate fund combination.

Three candidate fund combinations are assumed to be in total, the first candidate fund combination is assumed to comprise J1J2J3J4, the second candidate fund combination comprises J1J3J5J7, and the third candidate fund combination comprises J1J2J5J 6. J1, J2, J3, J4, J5 and J6 are the fund numbers.

And S420, calculating the occurrence density of each fund number in all current candidate fund combinations, wherein the occurrence density is the proportion of the fund number in all fund numbers.

As can be seen, fund J1 appears 3 times in the three candidate fund combinations with an appearance density of 3 ÷ 100%; fund J2 appeared 2 times in the three candidate fund combinations with an appearance density of 2 ÷ 3 ═ 67%; (ii) a Fund J3 appeared 2 times in the three candidate fund combinations with an appearance density of 2 ÷ 3 ═ 67%; fund J4 appeared 1 time in the three candidate fund combinations with an appearance density of 1 ÷ 3 ═ 33%; fund J5 appeared 2 times in the three candidate fund combinations with an appearance density of 2 ÷ 3 ═ 67%; fund J6 appeared 1 time in the three candidate fund combinations with an appearance density of 1 ÷ 3 ═ 33%; fund J7 appeared 1 time in the three candidate fund combinations with an appearance density of 1 ÷ 3 ═ 33%.

And S430, determining a recommended fund combination according to the occurrence density.

The ranking is performed according to the order of the appearance density from high to low, and it can be seen that J1> J2 ═ J3 ═ J5> J4 ═ J6 ═ J7, and if the top 4 funds are selected as the recommended fund combination, the recommended fund combination can be determined to be J1J2J3J 5.

It can be seen that the recommended fund combination is different from any one of the candidate fund combinations. Because the recommended fund combination comprehensively considers decision factors generated among a plurality of adjacent time periods and is determined according to the high-to-low sequence of fund product occurrence density, the recommended fund combination has higher accuracy and reliability compared with the candidate fund combination.

As indicated previously, the candidate fund combinations of the target merchant at the plurality of different time periods are obtained through a random forest model. As shown in fig. 5, the random forest model in this embodiment can be obtained by training through the following steps:

and S510, randomly extracting 80% of data from the sample space, constructing a training subset, and segmenting the training set data according to the characteristic conditions and the preset values.

And S520, calculating the importance of the evaluation feature of the Gini coefficient, and reserving the preset feature quantity.

And randomly selecting N characteristics, finding out the optimal characteristic sequence, characteristic value and segmented data subset for segmenting the current data set, and creating a sub-segmenter for the data subset until the classification is finished. The maximum depth of the decision tree can be set to be 3, the minimum sample number of the child nodes is 10, and the splitting is stopped when the sample size is 100, so that the decision tree is created.

S530: and (4) creating a random forest model by using a bagging mode and training the model.

S540: and (4) performing prediction by using the test data, calculating an AUC score for the test result, and adjusting the characteristic parameters of the model to retrain if the AUC score is lower than 85%.

With continued reference to fig. 6, a corporate investment prediction apparatus for monetary funds is shown, in this embodiment, the corporate investment prediction apparatus 60 may include or be divided into one or more program modules, which are stored in a storage medium and executed by one or more processors to implement the present invention and the method of forecasting corporate investment described above. The program modules referred to herein are a series of computer program instruction segments that perform certain functions and are more suitable than the program itself for describing the execution of the portfolio investment prediction device 60 in a storage medium. The following description will specifically describe the functions of the program modules of the present embodiment:

the profit prediction module 61 is suitable for acquiring first historical performance data of a plurality of funds and determining the predicted profit of each fund by using a profit prediction model;

the mobile fund prediction module 62 is adapted to obtain a first historical operating characteristic of a target merchant, and determine a predicted mobile fund of the merchant by using a mobile fund prediction model;

a random forest module 63, adapted to obtain a current operation characteristic of a target merchant, and use the current operation characteristic, the predicted profit, and the predicted liquidity as decision factors of a random forest model to obtain a candidate fund combination corresponding to the target merchant;

the recommending module 64 acquires a plurality of candidate fund combinations, and determines a recommended fund combination based on the occurrence densities of different funds included in the plurality of candidate fund combinations.

Further, the profit prediction module 61 includes:

a historical performance obtaining unit 611, adapted to obtain any one or more of a newly added purchase quantity, a newly added redemption quantity, a Shanghai depth 300 index, a bank daytime borrowing rate, a fund company financial report, a fund company establishment duration and a similar fund ranking corresponding to each fund as the historical performance data;

a predicted benefit unit 612 adapted to input the historical performance data into a benefit prediction model to output a predicted benefit for each fund over a preset time period; wherein the revenue prediction model is trained by a linear regression model.

Further, the predicted revenue unit 612 includes:

a data dividing subunit 6121 adapted to divide the historical performance data into working day historical performance data and resting calendar historical performance data;

a working day prediction subunit 6122, adapted to input the working calendar history performance data into the income prediction model, so as to output the predicted working day income of the fund in a preset time period;

a holiday prediction subunit 6123 adapted to input the holiday historical performance data into the revenue prediction model to output a predicted holiday revenue of the fund over a preset time period;

a comprehensive prediction subunit 6124, adapted to perform weighted summation on the predicted weekday gains and the predicted weekday gains to obtain the predicted gains.

Further, the liquidity fund prediction module 62 comprises:

a historical operation characteristic obtaining unit 621, adapted to obtain any one or more of historical sales volume, historical price, holiday factor, historical refund proportion, monthly-to-monthly cycle ratio withdrawal amount, monthly-to-monthly parity payout amount, sales promotion factor and purchase factor of the main operation commodity corresponding to the target merchant, as the historical operation characteristic;

a fund prediction unit 622 adapted to input the historical operating characteristics into a liquidity prediction model to output predicted liquidity for the merchant; wherein the liquidity forecasting model is trained based on an autocorrelation model.

Further, the recommendation module 64 includes:

a number obtaining unit 641 adapted to obtain a fund number included in each current candidate fund combination;

a density calculating unit 642 adapted to calculate an occurrence density of each fund number in all current candidate fund combinations, wherein the occurrence density is a proportion of the fund number in all fund numbers;

a combination determination unit 643, which determines a recommended fund combination according to the occurrence density.

The embodiment also provides a computer device, such as a smart phone, a tablet computer, a notebook computer, a desktop computer, a rack server, a blade server, a tower server or a rack server (including an independent server or a server cluster composed of a plurality of servers) capable of executing programs, and the like. The computer device 70 of the present embodiment includes at least, but is not limited to: a memory 71, a processor 72, which may be communicatively coupled to each other via a system bus, as shown in FIG. 7. It is noted that fig. 7 only shows a computer device 70 having components 71-72, but it is to be understood that not all of the shown components are required to be implemented, and that more or fewer components may be implemented instead.

In this embodiment, the memory 71 (i.e., a readable storage medium) includes a flash memory, a hard disk, a multimedia card, a card-type memory (e.g., SD or DX memory, etc.), a Random Access Memory (RAM), a Static Random Access Memory (SRAM), a read-only memory (ROM), an electrically erasable programmable read-only memory (EEPROM), a programmable read-only memory (PROM), a magnetic memory, a magnetic disk, an optical disk, and the like. In some embodiments, the storage 71 may be an internal storage unit of the computer device 70, such as a hard disk or a memory of the computer device 70. In other embodiments, the memory 71 may also be an external storage device of the computer device 70, such as a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) Card, a Flash memory Card (Flash Card), or the like, provided on the computer device 70. Of course, the memory 71 may also include both internal and external storage devices of the computer device 70. In this embodiment, the memory 71 is generally used for storing an operating system and various application software installed on the computer device 70, such as the program codes of the portfolio investment prediction device 60 of the first embodiment. Further, the memory 71 may also be used to temporarily store various types of data that have been output or are to be output.

Processor 72 may be a Central Processing Unit (CPU), controller, microcontroller, microprocessor, or other data Processing chip in some embodiments. The processor 72 generally serves to control the overall operation of the computer device 70. In this embodiment, the processor 72 is configured to execute the program codes stored in the memory 71 or process data, for example, to execute the portfolio investment prediction device 60, so as to implement the portfolio investment prediction method of the first embodiment.

The present embodiment also provides a computer-readable storage medium, such as a flash memory, a hard disk, a multimedia card, a card-type memory (e.g., SD or DX memory, etc.), a Random Access Memory (RAM), a Static Random Access Memory (SRAM), a read-only memory (ROM), an electrically erasable programmable read-only memory (EEPROM), a programmable read-only memory (PROM), a magnetic memory, a magnetic disk, an optical disk, a server, an App application mall, etc., on which a computer program is stored, which when executed by a processor implements corresponding functions. The computer readable storage medium of the present embodiment is used for storing a portfolio investment prediction device 60, and when being executed by a processor, the computer readable storage medium implements the portfolio investment prediction method of the first embodiment.

The above-mentioned serial numbers of the embodiments of the present invention are merely for description and do not represent the merits of the embodiments.

Any process or method descriptions in flow charts or otherwise described herein may be understood as representing modules, segments, or portions of code which include one or more executable instructions for implementing specific logical functions or steps of the process, and alternate implementations are included within the scope of the preferred embodiment of the present invention in which functions may be executed out of order from that shown or discussed, including substantially concurrently or in reverse order, depending on the functionality involved, as would be understood by those reasonably skilled in the art of the present invention.

It will be understood by those skilled in the art that all or part of the steps carried by the method for implementing the above embodiments may be implemented by hardware related to instructions of a program, which may be stored in a computer readable medium, and when executed, the program includes one or a combination of the steps of the method embodiments.

In the description herein, references to the description of the term "one embodiment," "some embodiments," "an example," "a specific example" or "some examples" or the like are intended to mean that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the invention. In this specification, the schematic representations of the terms used above do not necessarily refer to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

Through the above description of the embodiments, those skilled in the art will clearly understand that the method of the above embodiments can be implemented by software plus a necessary general hardware platform, and certainly can also be implemented by hardware, but in many cases, the former is a better implementation manner.

The above description is only a preferred embodiment of the present invention, and not intended to limit the scope of the present invention, and all modifications of equivalent structures and equivalent processes, which are made by using the contents of the present specification and the accompanying drawings, or directly or indirectly applied to other related technical fields, are included in the scope of the present invention.

Claims

1. A method for forecasting a portfolio investment for a monetary fund, comprising:

obtaining current operation characteristics of a target merchant in a plurality of different time periods, and respectively taking each current operation characteristic, the predicted income and the predicted liquidity as decision factors of a random forest model to obtain candidate fund combinations of the target merchant in the different time periods; wherein every two adjacent time intervals contain a crossing time interval;

and determining a recommended fund combination according to the occurrence density of different funds contained in the plurality of candidate fund combinations.

2. The method of forecasting the portfolio investment of monetary funds of claim 1 wherein the step of obtaining historical performance data for a plurality of funds and determining the forecasted revenue for each fund using a revenue forecasting model comprises:

3. The method of forecasting the combined investment of a monetary fund according to claim 2, wherein the step of inputting the historical performance data into a profit forecasting model to output a forecasted profit for each fund over a preset time period comprises:

4. The method of forecasting the combined investment of a money fund according to claim 1, wherein the step of obtaining historical operating characteristics of a target merchant and determining the forecasted liquidity of the merchant using a liquidity forecasting model comprises:

5. The method of forecasting the portfolio investment of monetary funds according to claim 1, wherein the step of obtaining a plurality of candidate fund combinations and determining the recommended fund combination based on the density of occurrence of the different funds contained in the plurality of fund combinations comprises:

acquiring fund numbers contained in each current candidate fund combination;

6. The method of forecasting the portfolio investment of monetary funds of claim 1, wherein the training process of the profit forecasting model includes:

and adding the operation sample data into the training sample to obtain a new training sample, training the profit prediction model by using the new sample data, and testing the profit prediction model by using the test sample.

7. The method of forecasting the portfolio investment of a monetary fund according to claim 1, wherein the training process of the predictive fund flow model comprises:

training the liquidity forecasting model with the new sample data.

8. A combinatorial investment prediction apparatus for a monetary fund, comprising:

the random forest module is suitable for obtaining current operation characteristics of a target merchant in a plurality of different time periods, and respectively taking each current operation characteristic, the predicted income and the predicted liquidity as decision factors of a random forest model so as to obtain candidate fund combinations of the target merchant in the plurality of different time periods; wherein every two adjacent time intervals contain a crossing time interval;

and the recommending module is used for determining the recommended fund combination according to the occurrence density of different funds contained in the candidate fund combinations.

9. A computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the steps of the method of any of claims 1 to 7 are implemented by the processor when executing the computer program.

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.