WO2019201001A1 - Fund forecasting method and device, and electronic device - Google Patents

Abstract

Disclosed are a fund forecasting method and device, and an electronic device. The forecasting method comprises: obtaining a target fund characteristic of a day to be forecasted, the target fund characteristic comprising a time attribute characteristic, a fund periodicity characteristic, a fund increment characteristic, and a fund increase rate characteristic of the day to be forecasted; and inputting the target fund characteristic into a trained neural network to forecast the corresponding target fund amount. According to the technical solution, the characteristics of the fund to be forecasted are comprehensively and accurately reflected by means of the time attribute characteristic, the fund periodicity characteristic, the fund increment characteristic, and the fund increase rate characteristic. The target fund amount is more accurately forecasted by means of a forecasting function obtained by the training of a neural network algorithm. Therefore, the technical problem in the prior art of low fund forecasting accuracy is solved, and the fund forecasting accuracy is increased.

Description

Method, device and electronic device for predicting funds

Cross-reference to related applications

The present application claims priority to Chinese Patent Application No. 201,810, 362, 560, filed on Apr. 20, 20, the entire disclosure of which is incorporated herein by reference. The manner of reference is incorporated herein.

Technical field

The present specification relates to the field of financial technology, and in particular, to a method, device and electronic device for predicting funds.

Background technique

With the continuous development of the social economy, the prediction of funds is one of the most common problems. For example, regarding the income of fund companies, there is often a need to know the amount of money per day as of today, and to predict the amount of money in the next few days (such as tomorrow, the day after tomorrow). The prior art mainly uses the method of variable analysis to predict the amount of funds in a certain day in the future, and only considers the linear relationship between the variable and the final result, and the prediction accuracy is low. There is a need for a new forecasting method to forecast funds to improve the accuracy of forecasts.

Summary of the invention

The embodiments of the present specification provide a method, a device, and an electronic device for predicting funds, which are used to solve the technical problem of low accuracy of fund prediction in the prior art, and improve the accuracy of prediction.

In a first aspect, an embodiment of the present specification provides a method for predicting funds, including:

Obtaining a target fund feature of a day to be predicted, wherein the target fund feature includes a time attribute feature of a day to be predicted, a periodicity feature of funds, a feature of capital increase, and a feature of a rate of increase of funds;

The target fund feature is input to the trained neural network, and the corresponding target amount of funds is predicted.

Optionally, the training method of the neural network includes:

Obtain historical capital data of the past n days, and the historical capital data of the day includes the amount of funds and the capital characteristics of the day, and n is an integer greater than 1. The capital characteristics include time attribute characteristics of the day, periodic characteristics of funds, and capital increase. Quantitative characteristics and capital increase rate characteristics;

The neural network training is performed based on the historical capital data of the past n days to obtain a trained neural network.

Optionally, the time attribute feature includes at least one of the following characteristics: whether the day is a weekend, whether it is a holiday, a day of the holiday, whether it is the beginning of the month, and whether it is the end of the month.

Optionally, the periodicity of the funds includes at least one of the following characteristics: the amount of funds yesterday, the amount of funds in the same period of last week, the amount of funds in the same period of last month, and the amount of funds in the same period of last year.

Optionally, the capital increase feature includes at least one of the following characteristics: yesterday, an increase in the amount of funds from the previous day, an increase in the amount of funds in the same period of last week, and an increase in the amount of funds in the same period of the previous month.

Optionally, the fund increase rate characteristic includes at least one of the following characteristics: a difference between yesterday's increment compared with the previous day and a previous day's larger previous day increment, yesterday's same day last week's increment, and the same day last week's same day last week's same day increment. value.

In a second aspect, the embodiment of the present specification provides a device for predicting funds, including:

An acquiring unit, configured to acquire a target fund feature of a day to be predicted, where the target fund feature includes a time attribute feature, a periodicity feature of funds, a fund increment feature, and a fund increase rate feature of a day to be predicted;

And a prediction unit, configured to input the target fund feature into the trained neural network, and predict a corresponding target amount of funds.

Optionally, the device further includes:

A training unit for performing neural network training by:

Obtain historical capital data of the past n days, and the historical capital data of the day includes the amount of funds and the capital characteristics of the day, and n is an integer greater than 1. The capital characteristics include time attribute characteristics of the day, periodic characteristics of funds, and capital increase. Quantitative characteristics and capital increase rate characteristics;

The neural network training is performed based on the historical capital data of the past n days to obtain a trained neural network.

Optionally, the time attribute feature includes at least one of the following characteristics: whether the day is a weekend, whether it is a holiday, a day of the holiday, whether it is the beginning of the month, and whether it is the end of the month.

Optionally, the periodicity of the funds includes at least one of the following characteristics: the amount of funds yesterday, the amount of funds in the same period of last week, the amount of funds in the same period of last month, and the amount of funds in the same period of last year.

Optionally, the capital increase feature includes at least one of the following characteristics: yesterday, an increase in the amount of funds from the previous day, an increase in the amount of funds in the same period of last week, and an increase in the amount of funds in the same period of the previous month.

Optionally, the fund increase rate characteristic includes at least one of the following characteristics: a difference between yesterday's increment compared with the previous day and a previous day's larger previous day increment, yesterday's same day last week's increment, and the same day last week's same day last week's same day increment. value.

In a third aspect, an embodiment of the present application provides a computer readable storage medium, where a computer program is stored, and when the program is executed by the processor, the following steps are implemented:

Obtaining a target fund feature of a day to be predicted, wherein the target fund feature package includes a time attribute feature, a periodicity of funds, a fund increment feature, and a fund increase rate characteristic of the day to be predicted;

The target fund feature is input to the trained neural network, and the corresponding target amount of funds is predicted.

In a fourth aspect, an embodiment of the present specification provides an electronic device including a memory and one or more programs, wherein one or more programs are stored in the memory and configured to be executed by one or more processors The one or more programs include instructions for performing the following operations:

Obtaining a target fund feature of a day to be predicted, wherein the target fund feature includes a time attribute feature of a day to be predicted, a periodicity feature of funds, a feature of capital increase, and a feature of a rate of increase of funds;

The target fund feature is input to the trained neural network, and the corresponding target amount of funds is predicted.

The above one or more technical solutions in the embodiments of the present specification have at least the following technical effects:

The embodiment of the present specification provides a method for predicting funds, by acquiring a target fund feature of a day to be predicted, the target fund feature package includes a time attribute feature, a periodicity of funds, a fund increment feature, and a day of the day to be predicted. The fund increase rate characteristic; input the target fund feature into the trained neural network, and predict the corresponding target fund amount. Due to the above prediction method, based on the characteristics of the time attribute characteristics, the periodicity of funds, the characteristics of capital increase, and the rate of increase of funds, the four characteristics of interaction are affected, and the target fund characteristics are input into the trained neural network. Corresponding target amount of funds, because the neural network fully considers the cross-relationship between the above characteristics, more accurately reflects the change process of funds, thereby obtaining more accurate prediction results, and solving the technology of lower prediction accuracy of funds in the prior art. The problem has increased the accuracy of the funding forecast.

DRAWINGS

In order to more clearly illustrate the technical solutions in the embodiments of the present specification, the drawings used in the embodiments or the description of the prior art will be briefly described below. Obviously, the drawings in the following description are the description of the present specification. For some embodiments, other drawings may be obtained from those of ordinary skill in the art without departing from the drawings.

FIG. 1 is a flowchart of a method for predicting funds according to an embodiment of the present disclosure;

2 is a schematic diagram of a cross relationship between capital features provided by an embodiment of the present specification;

FIG. 3 is a schematic diagram of a method for predicting funds according to an embodiment of the present specification;

FIG. 4 is a schematic diagram of an electronic device according to an embodiment of the present disclosure.

detailed description

The technical solutions in the embodiments of the present specification will be clearly and completely described in conjunction with the drawings in the embodiments of the present specification. It is a part of the embodiments of the present specification, and not all of the embodiments. All other embodiments obtained by a person of ordinary skill in the art based on the embodiments of the present specification without departing from the inventive scope are the scope of the present disclosure.

The embodiment of the present specification provides a method, a device, and an electronic device for predicting funds, which are used to solve the technical problem of low capital prediction accuracy in the prior art, and improve the accuracy of fund prediction.

The main implementation principles, specific implementation manners, and the corresponding beneficial effects that can be achieved by the technical solutions of the embodiments of the present specification are described in detail below with reference to the accompanying drawings.

The influencing factors of funds are the characteristics of funds, and the selection of capital characteristics is the key to the accuracy of fund forecasting. At the same time, the method of forecasting funds is also crucial, and the accuracy of the forecasted funds is different for different forecasting methods with the same capital characteristics. This specification combines forecasting methods and funding characteristics to optimize the forecast of the amount of funds.

The embodiment of the present specification provides a method for predicting funds. Referring to FIG. 1, the prediction method includes:

S110: Obtain a target fund feature of a day to be predicted, where the target fund feature includes a time attribute feature, a periodicity feature of funds, a fund increment feature, and a fund increase rate feature of the day to be predicted;

S120: Enter the target fund feature into the trained neural network, and predict the corresponding target fund amount.

In the specific implementation process, the training methods of the neural network include:

When obtaining historical fund data for the past n days, you can first obtain the amount of funds for n days per day, and then obtain the k characteristics of the day based on the amount of funds in the past n days. The larger the value of n is, the more accurate the prediction result is, but the larger the value of n is, the lower the prediction rate is. Specifically, n may take an intermediate value according to the computing capability of the electronic device and the prediction accuracy, such as n may take 90, 180, 365, and the like. The specific value of k is not limited in this specification. In order to improve the accuracy of the prediction, the capital feature provided in the embodiment of the present specification may include at least one of the following types of features, and the number of features in each type of feature is not limited:

The time attribute feature includes at least one of the following characteristics: whether the weekend is a weekend, whether it is a holiday, a day of the holiday, whether the beginning of the month (end), the day of the month, and the like.

The periodic characteristics of funds include at least one of the following characteristics: the amount of funds yesterday, the amount of funds in the same period last week, the amount of funds in the same period of last month, the amount of funds in the same period last year, and the average amount of funds in the last m days.

The characteristics of capital increase include at least one of the following characteristics: yesterday's increase in the amount of funds from the previous day, the increase in the amount of funds in the same period last week, the increase in the amount of funds in the same period of the previous month, and the increase in the amount of funds in the recent m-day.

The rate of increase in capital rate includes at least one of the following characteristics: the difference between yesterday's increment from the previous day and the previous day's larger previous day's increment, yesterday's increase from the same day last week, and the difference between the same day last week and the same day last week.

Among them, the periodic characteristics of funds, the characteristics of capital increments and the characteristics of capital increase rate belong to the characteristics of capital attributes. When acquiring the k funds features of the day, the daily capital attribute characteristics including the periodic feature, the incremental feature, and the increase rate feature in the past n days can be obtained as the k funds feature of the day, and the daily funds in the past n days can also be obtained. Attribute characteristics and time attribute characteristics are used as k funds characteristics per day. Further, the amount of money per day and the k funds characteristics per day are taken as historical capital data for each day.

After obtaining the historical fund data for the past n days, the neural network training is performed based on the historical fund data of the past n days to obtain a trained neural network. According to the historical fund data of the past n days, the neural network training can obtain the non-linear relationship between the capital characteristics of each day and the non-linear relationship between the amount of funds and the characteristics of k funds, and can obtain more accurate changes in funds. process. For example, if it is a holiday, the amount of funds in the same period will increase or decrease, and the increment of funds in the same period will increase or decrease.

Specifically, the historical fund data of the past n days can be used as a training sample for neural network training. The daily capital characteristics and funds in the past n days are a training sample <X, Y>, and X can be represented by a vector, including k Capital characteristics, the capital characteristics of the t-th sample can be expressed as <x _t1 , x _t2 ,..., x _tk >, Y represents the amount of funds on the day, and the amount of funds in the t-th sample can be expressed as y _t . Obtain a nonlinear function Y=f(X) between the amount of money per day and the characteristics of k funds. The nonlinear function Y=f(X) is used to characterize the relationship between the capital characteristics and the relationship between the amount of funds and the characteristics of funds. Nonlinear relationship. For example: Please refer to Figure 2 for a 3-layer neural network (more layers of neural network structure are similar), each sample has 3 features (ie k=3), and the nonlinear function Y=hw learned through neural network , b(X), through the nonlinear function Y=hw, b(X) to characterize the cross relationship between features and the nonlinear relationship between features and funds.

After the trained neural network is obtained, when the fund prediction is performed, S110 is executed to obtain the target fund characteristics of the day to be predicted. Specifically, the k target fund characteristics of the day to be predicted are obtained. The k target capital characteristics may be obtained according to the amount of funds in the past n days, including the time attribute characteristics of the day to be predicted, the periodicity characteristics of funds, the characteristics of capital increments, and the characteristics of capital increase rate. For example, the day to be predicted is the jth sample, and its capital characteristics <x _j1 , x _j2 ,..., x _jk > are obtained.

Then, S120 is executed to input the target fund feature into the trained neural network, and the corresponding target fund amount is predicted. Specifically, using the trained neural network Y=f(X), the k target capital features <x _j1 , x _j2 , . . . , x _jk > obtained by S110 are input into Y=f(X) for the target amount of funds. Y _j is predicted to obtain Y _j =f(x _j1 , x _j2 ,..., x _jk ). The neural network training method can more accurately describe the complex intersection and nonlinear relationship between features, so the prediction of the amount of funds can greatly improve the accuracy of fund prediction.

In the specific implementation process, when the fund prediction is performed, the time attribute feature or the fund attribute feature may be separately obtained to perform neural network training, and the non-linear relationship between the characteristic feature intersection relationship and the feature and the fund amount may be obtained. The linear function is the prediction function, from which the funds are predicted. Compared with the time attribute feature or the capital attribute feature, the above embodiment performs fund prediction according to the time attribute feature and the fund attribute feature, fully considering the cross-effect of the time attribute on each fund feature, and the prediction accuracy of the fund is greatly improved. improve.

Based on the method for predicting funds provided by the foregoing embodiment, the embodiment of the present specification further provides a prediction device. Referring to FIG. 3, the device includes:

The obtaining unit 31 is configured to acquire a target fund feature of a day to be predicted, where the target fund feature includes a time attribute feature, a fund periodic feature, a fund increment feature, and a fund increase rate feature of the day to be predicted;

The predicting unit 32 is configured to input the target capital feature into the trained neural network, and predict the corresponding target amount of funds.

As an optional implementation manner, the apparatus may further include: a training unit 33, configured to perform neural network training by the following method. Specifically, the historical fund data of the past n days is acquired, and the historical capital data of the day includes the amount of funds and the fund characteristics of the day, and n is an integer greater than 1, and the fund features include time attribute characteristics of the day and periodic characteristics of funds. , fund increment characteristics and capital increase rate characteristics; based on the historical n-day historical fund data for neural network training to obtain a well-trained neural network.

The time attribute feature includes at least one of the following characteristics: whether the day is a weekend, whether it is a holiday, a day of the holiday, whether it is the beginning of the month, and whether it is the end of the month. The periodic characteristics of the funds include at least one of the following characteristics: the amount of funds yesterday, the amount of funds in the same period last week, the amount of funds in the same period of last month, and the amount of funds in the same period of last year. The capital increase feature includes at least one of the following characteristics: yesterday's increase in the amount of funds from the previous day, the increase in the amount of funds in the same period last week, and the increase in the amount of funds in the same period of the previous month. The fund increase rate characteristic includes at least one of the following characteristics: a difference between yesterday's increment compared with the previous day and the previous day's larger previous day increment, yesterday's increase from the same day last week, and the difference between the same day last week and the same day last week.

With regard to the apparatus in the above embodiments, the specific manner in which the operations are performed by the respective units has been described in detail in the embodiments of the related methods, and will not be described in detail herein.

Please refer to FIG. 4 , which is a block diagram of an electronic device 700 for implementing a data query method, according to an exemplary embodiment. For example, electronic device 700 can be a computer, a database console, a tablet device, a personal digital assistant, and the like.

Referring to FIG. 4, electronic device 700 can include one or more of the following components: processing component 702, memory 704, power component 706, multimedia component 708, input/output (I/O) interface 710, and communication component 712.

Processing component 702 typically controls the overall operations of electronic device 700, such as operations associated with display, data communication, and recording operations. Processing component 702 can include one or more processors 720 to execute instructions to perform all or part of the steps described above. Moreover, processing component 702 can include one or more modules to facilitate interaction between component 702 and other components.

Memory 704 is configured to store various types of data to support operation at device 700. Examples of such data include instructions for any application or method operating on electronic device 700, contact data, phone book data, messages, pictures, videos, and the like. Memory 704 can be implemented by any type of volatile or non-volatile storage device, or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read only memory (EEPROM), erasable Programmable Read Only Memory (EPROM), Programmable Read Only Memory (PROM), Read Only Memory (ROM), Magnetic Memory, Flash Memory, Disk or Optical Disk.

Power component 706 provides power to various components of electronic device 700. Power component 706 can include a power management system, one or more power sources, and other components associated with generating, managing, and distributing power for electronic device 700.

The I/O interface 710 provides an interface between the processing component 702 and the peripheral interface module, which may be a keyboard, a click wheel, a button, or the like. These buttons may include, but are not limited to, a home button, a volume button, a start button, and a lock button.

Communication component 712 is configured to facilitate wired or wireless communication between electronic device 700 and other devices. The electronic device 700 can access a wireless network based on a communication standard, such as WiFi, 2G or 3G, or a combination thereof. In an exemplary embodiment, communication component 712 receives broadcast signals or broadcast associated information from an external broadcast management system via a broadcast channel. In an exemplary embodiment, the communication component 712 also includes a near field communication (NFC) module to facilitate short range communication. For example, the NFC module can be implemented based on radio frequency identification (RFID) technology, infrared data association (IrDA) technology, ultra-wideband (UWB) technology, Bluetooth (BT) technology, and other technologies.

In an exemplary embodiment, electronic device 700 may be implemented by one or more application specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), A gated array (FPGA), controller, microcontroller, microprocessor, or other electronic component implementation for performing the above methods.

In an exemplary embodiment, there is also provided a non-transitory computer readable storage medium comprising instructions, such as a memory 704 comprising instructions executable by processor 720 of electronic device 700 to perform the above method. For example, the non-transitory computer readable storage medium may be a ROM, a random access memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, and an optical data storage device.

A non-transitory computer readable storage medium, when instructions in the storage medium are executed by a processor of a mobile terminal, to enable an electronic device to perform a data query method, the method comprising:

Obtaining a target fund feature of a day to be predicted, the target fund feature including a time attribute feature of a day to be predicted, a fund periodic feature, a fund increment feature, and a fund increase rate feature; inputting the target fund feature into A well-trained neural network predicts the corresponding target amount of funds.

It is to be understood that the invention is not limited to the details of the details of The scope of the invention is limited only by the appended claims

The above are only the preferred embodiments of the present invention, and are not intended to limit the present invention. Any modifications, equivalents, improvements, etc., which are within the spirit and scope of the present invention, should be included in the protection of the present invention. Within the scope.

Claims (14)

A method of predicting funds, including: Obtaining a target fund feature of a day to be predicted, the target fund feature including a time attribute feature of the day to be predicted, a periodicity feature of funds, a feature of capital increase, and a feature of a rate of increase of funds; The target fund feature is input to the trained neural network, and the corresponding target amount of funds is predicted. The method of claim 1, the training method of the neural network comprising: Obtain historical capital data of the past n days, and the historical capital data of the day includes the amount of funds and the capital characteristics of the day, and n is an integer greater than 1. The capital characteristics include time attribute characteristics of the day, periodic characteristics of funds, and capital increase. Quantitative characteristics and capital increase rate characteristics; The neural network training is performed based on the historical capital data of the past n days to obtain a trained neural network. The method of claim 2, the time attribute feature comprising at least one of the following characteristics: whether the day is a weekend, whether it is a holiday, a day of the holiday, whether it is the beginning of the month, and whether it is the end of the month. The method of claim 2, wherein the capital periodicity feature comprises at least one of the following characteristics: the amount of funds yesterday, the amount of funds in the same period last week, the amount of funds in the same period of the previous month, and the amount of funds in the same period of last year. The method of claim 2, wherein the capital increase feature comprises at least one of the following: an increase in the amount of funds from yesterday, a decrease in the amount of funds in the same period of last week, and an increase in the amount of funds in the same period of the previous month. The method according to claim 2, wherein said capital increase rate characteristic comprises at least one of the following characteristics: yesterday, the difference between the previous day's increment and the previous day's larger day's day increment, and the yesterday's same day last week's increment compared with the previous day's same day. The difference between Zhou Tongtian increments. A forecasting device for funds, comprising: An acquiring unit, configured to acquire a target fund feature of a day to be predicted, where the target fund feature includes a time attribute feature of the day to be predicted, a periodicity feature of funds, a feature of capital increase, and a feature of a rate of increase of funds; And a prediction unit, configured to input the target fund feature into the trained neural network, and predict a corresponding target amount of funds. The device of claim 7 further comprising: A training unit for performing neural network training by: Obtain historical capital data of the past n days, and the historical capital data of the day includes the amount of funds and the capital characteristics of the day, and n is an integer greater than 1. The capital characteristics include time attribute characteristics of the day, periodic characteristics of funds, and capital increase. Quantitative characteristics and capital increase rate characteristics; The neural network training is performed based on the historical capital data of the past n days to obtain a trained neural network. The apparatus of claim 8, the time attribute feature comprising at least one of the following characteristics: whether the day is a weekend, whether it is a holiday, a day of the holiday, whether it is the beginning of the month, and whether it is the end of the month. The apparatus according to claim 8, wherein said capital periodicity characteristic comprises at least one of the following characteristics: the amount of funds yesterday, the amount of funds in the same period of last week, the amount of funds in the same period of the previous month, and the amount of funds in the same period of last year. The apparatus according to claim 8, wherein said capital increment feature comprises at least one of the following characteristics: yesterday, an increase in the amount of funds from the previous day, an increase in the amount of funds in the same period of the previous week, and an increase in the amount of funds in the same period of the previous month. The apparatus according to claim 8, wherein said capital increase rate characteristic comprises at least one of the following characteristics: yesterday, the difference between the previous day's increment and the previous day's larger day's day increment, and the yesterday's same day last week's increment compared with the previous day's same day. The difference between Zhou Tongtian increments. A computer readable storage medium having stored thereon a computer program, wherein the program, when executed by the processor, implements the following steps: Obtaining a target fund feature of a day to be predicted, the target fund feature including a time attribute feature of the day to be predicted, a periodicity feature of funds, a feature of capital increase, and a feature of a rate of increase of funds; The target fund feature is input to the trained neural network, and the corresponding target amount of funds is predicted. An electronic device, comprising: a memory, and one or more programs, wherein one or more programs are stored in the memory and configured to execute the one or more by one or more processors The program contains instructions for doing the following: Obtaining a target fund feature of a day to be predicted, wherein the target fund feature includes a time attribute feature of a day to be predicted, a periodicity feature of funds, a feature of capital increase, and a feature of a rate of increase of funds; The target fund feature is input to the trained neural network, and the corresponding target amount of funds is predicted.

Images