CN111126676A

CN111126676A - Method, device and equipment for predicting company operation risk

Info

Publication number: CN111126676A
Application number: CN201911234903.5A
Authority: CN
Inventors: 张�杰; 于皓; 李犇; 吴信东
Original assignee: Beijing Mininglamp Software System Co ltd
Current assignee: Beijing Mininglamp Software System Co ltd
Priority date: 2019-12-05
Filing date: 2019-12-05
Publication date: 2020-05-08

Abstract

A method, apparatus, device and computer-readable storage medium for company operational risk prediction, wherein the method comprises: obtaining company operation historical data, and training a prediction model according to the company operation historical data; the input data of the prediction model comprises current risk release data, current business increment data, previous-period risk stock data and previous-period risk performance data, and the output data of the prediction model comprises current risk stock data and current-period risk performance data; and acquiring current company operation data, inputting the current company operation data into the prediction model, and taking current risk performance data output by the prediction model as a prediction result.

Description

Method, device and equipment for predicting company operation risk

Technical Field

This document relates to the field of risk prediction, and more particularly, to a method, apparatus, device, and computer-readable storage medium for company operations risk prediction.

Background

In the financial field, it is very important to predict the operational risk of a company. The related art predicts a future value given historical data by regarding each business risk index as time-series data. The input data is a macroscopic operation index, and the output is also a macroscopic index. The purpose of the predictive algorithm is to learn and analyze the periodicity and trend of the index.

Common prediction methods for time series are:

the traditional method comprises the following steps: simple averaging, moving averaging, exponential smoothing, etc.

The modern method comprises the following steps: auto regression Model (AR), Moving average Model (MA), Auto regression Moving average Model (ARMA), differential Auto regression Moving average Model (ARIMA), and the like.

The input and output data of the prediction method are the same dimension and the same business meaning, and the macroscopic index cannot be predicted through more detailed microscopic data, so the accuracy is not high enough.

Disclosure of Invention

The application provides a method, a device, equipment and a computer readable storage medium for forecasting company operation risk, so as to forecast macroscopic operation risk through fine-grained microscopic sales and after-sales data.

The embodiment of the application provides a method for predicting company operation risk, which comprises the following steps:

obtaining company operation historical data, and training a prediction model according to the company operation historical data; the input data of the prediction model comprises current risk release data, current business increment data, previous-period risk stock data and previous-period risk performance data, and the output data of the prediction model comprises current risk stock data and current-period risk performance data;

and acquiring current company operation data, inputting the current company operation data into the prediction model, and taking current risk performance data output by the prediction model as a prediction result.

In one embodiment, the predictive model includes a neural network, and the training of the predictive model based on the company operation history data includes:

and taking the previous-period risk performance data and the current-period business increment data as inputs, taking the risk increment data in the current business increment as outputs, and training the neural network.

In one embodiment, the prediction model further comprises a classifier, the training of the prediction model according to the company operation history data further comprises:

and taking the risk increment data and the current risk stock data as input, taking the current risk performance data as output, and training the classifier in a machine learning mode.

In one embodiment, the current stage risk inventory data is the vector addition of the previous stage risk inventory data, the current stage risk release data and the risk increment data.

The embodiment of the present application further provides a device for predicting company operation risk, including:

the data acquisition module is used for acquiring company operation historical data and company operation current data;

the model training module is used for training a prediction model according to the company operation historical data; the input data of the prediction model comprises current risk release data, current business increment data, previous-period risk stock data and previous-period risk performance data, and the output data of the prediction model comprises current risk stock data and current-period risk performance data;

and the risk prediction module is used for inputting the current company operation data into the prediction model and taking the current risk performance data output by the prediction model as a prediction result.

In one embodiment, the predictive model includes a neural network, and the model training module is configured to:

In an embodiment, the prediction model further comprises a classifier, and the model training module is configured to:

taking the risk increment data and the current risk stock data as input, taking the current risk performance data as output, and training the classifier in a machine learning mode;

the current-stage risk stock data is vector addition of the previous-stage risk stock data, the current-stage risk release data and the risk increment data.

In one embodiment, the predictive model comprises:

the forgetting unit is used for carrying out vector addition on the previous-period risk stock data and the current-period risk release data and sending the obtained released risk data to the integrating unit;

the memory unit comprises a neural network, the input of the neural network is previous-stage risk expression data and current-stage business increment data, the output of the neural network is risk increment data in current business increments, and the memory unit respectively sends the risk increment data to the integration unit and the prediction unit;

the integration unit is used for carrying out vector addition on the released risk data and the risk increment data and sending the obtained current risk stock data to the prediction unit;

and the input of the classifier is the risk increment data and the current risk stock data, and the output of the classifier is current risk performance data.

The embodiment of the present application further provides a device for predicting company operation risk, including: a memory, a processor and a computer program stored on the memory and executable on the processor, the processor when executing the program implementing the method of company business risk prediction.

The embodiment of the application also provides a computer-readable storage medium, which stores computer-executable instructions for executing the method for predicting the company operation risk.

Compared with the related art, the method comprises the following steps: obtaining company operation historical data, and training a prediction model according to the company operation historical data; the input data of the prediction model comprises current risk release data, current business increment data, previous-period risk stock data and previous-period risk performance data, and the output data of the prediction model comprises current risk stock data and current-period risk performance data; and acquiring current company operation data, inputting the current company operation data into the prediction model, and taking current risk performance data output by the prediction model as a prediction result. According to the method and the device, the macroscopic risk index is predicted through the microscopic data, the periodic risk, the trend of the macroscopic trend and the sporadic aperiodic risk can be automatically learned in a data-driven mode, and therefore the predicted risk index is more accurate.

Additional features and advantages of the application will be set forth in the description which follows, and in part will be obvious from the description, or may be learned by the practice of the application. Other advantages of the present application may be realized and attained by the instrumentalities and combinations particularly pointed out in the specification and the drawings.

Drawings

The accompanying drawings are included to provide an understanding of the present disclosure and are incorporated in and constitute a part of this specification, illustrate embodiments of the disclosure and together with the examples serve to explain the principles of the disclosure and not to limit the disclosure.

FIG. 1 is a flow chart of a method for company operational risk prediction in an embodiment of the present application;

FIG. 2 is a schematic diagram of a company operation risk prediction device according to an embodiment of the present application;

FIG. 3 is a schematic diagram of a company operation risk prediction device according to an embodiment of the present application;

fig. 4 is a schematic diagram showing the composition of a prediction model according to an application example of the present application.

Detailed Description

The present application describes embodiments, but the description is illustrative rather than limiting and it will be apparent to those of ordinary skill in the art that many more embodiments and implementations are possible within the scope of the embodiments described herein. Although many possible combinations of features are shown in the drawings and discussed in the detailed description, many other combinations of the disclosed features are possible. Any feature or element of any embodiment may be used in combination with or instead of any other feature or element in any other embodiment, unless expressly limited otherwise.

The present application includes and contemplates combinations of features and elements known to those of ordinary skill in the art. The embodiments, features and elements disclosed in this application may also be combined with any conventional features or elements to form a unique inventive concept as defined by the claims. Any feature or element of any embodiment may also be combined with features or elements from other inventive aspects to form yet another unique inventive aspect, as defined by the claims. Thus, it should be understood that any of the features shown and/or discussed in this application may be implemented alone or in any suitable combination. Accordingly, the embodiments are not limited except as by the appended claims and their equivalents. Furthermore, various modifications and changes may be made within the scope of the appended claims.

Further, in describing representative embodiments, the specification may have presented the method and/or process as a particular sequence of steps. However, to the extent that the method or process does not rely on the particular order of steps set forth herein, the method or process should not be limited to the particular sequence of steps described. Other orders of steps are possible as will be understood by those of ordinary skill in the art. Therefore, the particular order of the steps set forth in the specification should not be construed as limitations on the claims. Further, the claims directed to the method and/or process should not be limited to the performance of their steps in the order written, and one skilled in the art can readily appreciate that the sequences may be varied and still remain within the spirit and scope of the embodiments of the present application.

With the continuous development of internet and big data technology, many companies face the problem of digital transformation, how to accurately predict macroscopic operation risk indexes according to microscopic data inside the companies and timely take corresponding coping strategies becomes a very important and urgent problem to be solved. For example, in the loan business of financial institutions, various risk factors often have lagged nature with different lengths in time, and how to predict the operation risk indexes (such as total loan balance, accumulated bad account rate, overdue rate, profit margin and the like) in a future period of time according to the data (such as loan amount, credit score, repayment balance, overdue days and the like) of each loan client every day is very important for future operation decisions (such as adjustment of sales rhythm, asset structure and wind control strategy).

The embodiment of the application provides a method for predicting macro operation risk through fine-grained micro sales and after-sales data. The periodicity and the development trend of the indexes can be learned, and the potential influence of a shorter term can be learned.

As shown in fig. 1, a method for predicting company operation risk according to an embodiment of the present application includes:

step 101, obtaining company operation historical data, and training a prediction model according to the company operation historical data; the input data of the prediction model comprises current risk release data, current business increment data, previous-period risk stock data and previous-period risk performance data, and the output data of the prediction model comprises current-period risk stock data and current-period risk performance data.

After the operation historical data of the company is obtained, the data can be firstly cleaned, and then a prediction model can be trained according to the cleaned data.

The data cleaning is to filter out unnecessary data, perform quality verification and format conversion on the data needing further processing, and split and summarize the data according to dimensions of customers, orders, products, channels and the like.

The current risk release data refers to data such as overdue data, bad accounts data and clearing data which are newly generated in the current time period, wherein the term refers to a time period which can be a working day, a week, a month and the like.

The current service increment data refers to newly added deposit data aggregated in the current time period, and the deposit data includes but is not limited to: the identity of the newly added order, the product attributes of the newly added order (which may include total, deadline, rate, monthly payment amount, payment method), the customer identity of the newly added order, the customer attributes of the newly added order (which may include age, gender, marital status, education level, fixed asset size, monthly stable income, income liability ratio), the mortgage, or the warranty method.

The previous-period risk performance data refers to risk indexes of the previous period, such as total deposit balance, accumulated bad account rate, overdue rate, profit margin and the like.

The previous-period risk stock data refers to the total risk stock accumulated in the previous-period history, such as the total deposit data and the like, and can be represented by low-dimensional continuous vectors.

The current inventory risk data refers to the total inventory risk accumulated so far.

The current stage risk performance data refers to a current stage risk indicator.

In one embodiment, the prediction model includes a neural network, and in step 101, the neural network is trained by taking the previous-period risk performance data and the current-period business increment data as inputs and taking the risk increment data in the current business increment as outputs.

The Neural network can adopt a Deep Neural Network (DNN) and comprises an input layer, a hidden layer and an output layer, wherein the first layer is the input layer, the last layer is the output layer, and the middle layers are all hidden layers. The layers are all connected, namely any neuron of the ith layer is necessarily connected with any neuron of the (i + 1) th layer. Since the deep neural network can be implemented by using related technologies, it is not described herein.

In an embodiment, the prediction model further includes a classifier, and in step 101, the classifier is trained in a machine learning manner by taking the risk increment data and the current risk stock data as inputs and taking the current risk performance data as outputs.

The classifier based on machine learning can adopt a nearest neighbor algorithm KNN, a Bayesian classifier, a logistic regression algorithm, a decision tree, a support vector machine and the like.

And 102, acquiring current company operation data, inputting the current company operation data into the prediction model, and taking current-term risk expression data output by the prediction model as a prediction result.

Similar to the processing in step 101, after the current data of the company is acquired, data can be first cleaned, and then a prediction result can be acquired according to the cleaned data.

According to the method and the device, the macroscopic risk index is predicted through the microscopic data, the periodic risk, the trend of the macroscopic trend and the sporadic aperiodic risk can be automatically learned in a data-driven mode, and therefore the predicted risk index is more accurate.

As shown in fig. 2, an embodiment of the present application further provides a device for predicting company operational risk, including:

the data acquisition module 21 is used for acquiring company operation historical data and company operation current data;

the model training module 22 is used for training a prediction model according to the company operation historical data; the input data of the prediction model comprises current risk release data, current business increment data, previous-period risk stock data and previous-period risk performance data, and the output data of the prediction model comprises current risk stock data and current-period risk performance data;

and the risk prediction module 23 is used for inputting the current company management data into the prediction model and taking the current risk performance data output by the prediction model as a prediction result.

Referring to fig. 3, the model training module 22 may include a historical database and a cyclic training module, the historical database includes company operation historical data after data cleaning by the data acquisition module 21.

Risk prediction module 23 may include a prediction module and a prediction model. The internal structures of the loop training module and the prediction module are consistent, and the difference is that the loop training module is used for batch training and the prediction module is used for single prediction.

The prediction model is explained below.

Referring to fig. 4, wherein the input data of the prediction model includes:

1. current risk release data, i.e. current release risk data: the overdue, bad account and clearing data newly generated in the current time period are shown by y in FIG. 4_tRepresents; ("period" refers to a period of time, which may be a weekday, a week, a month, etc.).

2. Current business increment data: the newly added deposit data after the summary in the current time period is represented by x in FIG. 4_tAnd (4) showing. The deposit data includes, but is not limited to: identification of newly added order, product attribute (total amount, deadline, rate, monthly payment amount, payment mode) of newly added order, customer identification of newly added order, customer attribute (age, gender, marital status, education level, fixed asset scale, month, etc.) of newly added orderStable income, income-debt ratio), mortgage or guaranty.

3. Previous stage risk performance data: risk index of the previous stage, in h_t-1Representing, for example, total deposit balance, cumulative bad account rate, overdue rate, profit margin, etc.;

4. previous period risk stock data: total accumulated historical risk inventory, using C_t-1And (4) showing.

The output data of the prediction model includes:

1. current risk performance data: by using h_tAnd (4) showing.

2. Current risk stock data: accumulation of Total Risk inventory Up to now, with C_tAnd (4) showing.

Referring to fig. 4, the prediction model may include: a forgetting unit 41, a memory unit 42, an integration unit 43, and a prediction unit 44. Wherein:

and the forgetting unit 41 is used for performing vector addition on the previous-period risk stock data and the current-period risk release data and sending the obtained released risk data to the integrating unit.

The forgetting unit 41 adds the risk data newly released at the present time to the risk stock data C of the previous time in a vector addition manner_t-1And updating the state of the system, and forgetting the clients who release risks, such as overdue, bad account and clear.

The memory unit 42 includes a neural network, inputs of the neural network are previous-stage risk performance data and current-stage business increment data, outputs of the neural network are risk increment data in current business increments, and the memory unit sends the risk increment data to the integration unit and the prediction unit respectively.

The newly added payment information and the updated payment state (such as surplus amount in the loan, paid amount, maximum overdue days, maximum overdue amount and the like) of the monthly payment behavior of the client in the loan can be input into the memory unit, the memory unit is realized through a deep neural network, and the network structure is obtained through training of a large amount of historical data, so that the memory unit has a selective memory function.

And the integrating unit 43 is configured to perform vector addition on the released risk data and the risk increment data, and send the obtained current risk stock data to the predicting unit.

The integration unit 43 adds the outputs of the forgetting unit 41 and the memory unit 42 together to generate the current risk stock C_t。

And the prediction unit 44 comprises a classifier, wherein the input of the classifier is the risk increment data and the current risk stock data, and the output of the classifier is the current risk performance data.

The prediction unit 44 may be implemented by a classifier trained by a machine learning method, and may predict the current risk performance according to the risk stock and the output of the memory unit 41.

In summary, the embodiment of the present application provides a scheme for predicting a macroscopic risk indicator through microscopic data, where a prediction model obtained by the scheme has long-term memory and short-term memory, the long-term memory retains risk stock information, the short-term memory filters newly-added risk increments from a currently-input service increment signal, and periodic risks, trends of macroscopic trends, and sporadic aperiodic risks can be automatically learned in a data-driven manner, so that the predicted risk indicator is more accurate.

In this embodiment, the storage medium may include, but is not limited to: a U-disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a removable hard disk, a magnetic or optical disk, and other various media capable of storing program codes.

It will be understood by those of ordinary skill in the art that all or some of the steps of the methods, systems, functional modules/units in the devices disclosed above may be implemented as software, firmware, hardware, and suitable combinations thereof. In a hardware implementation, the division between functional modules/units mentioned in the above description does not necessarily correspond to the division of physical components; for example, one physical component may have multiple functions, or one function or step may be performed by several physical components in cooperation. Some or all of the components may be implemented as software executed by a processor, such as a digital signal processor or microprocessor, or as hardware, or as an integrated circuit, such as an application specific integrated circuit. Such software may be distributed on computer readable media, which may include computer storage media (or non-transitory media) and communication media (or transitory media). The term computer storage media includes volatile and nonvolatile, removable and non-removable media implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules or other data, as is well known to those of ordinary skill in the art. Computer storage media includes, but is not limited to, RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, Digital Versatile Disks (DVD) or other optical disk storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other medium which can be used to store the desired information and which can accessed by a computer. In addition, communication media typically embodies computer readable instructions, data structures, program modules or other data in a modulated data signal such as a carrier wave or other transport mechanism and includes any information delivery media as known to those skilled in the art.

Claims

1. A method for risk prediction for a company, comprising:

2. The method of claim 1, wherein the predictive model comprises a neural network, and wherein training a predictive model based on the company operational history data comprises:

3. The method of claim 2, wherein the predictive model further comprises a classifier, the training of the predictive model based on the company operational history data further comprising:

4. The method of claim 3,

5. An apparatus for risk prediction in a company, comprising:

6. The apparatus of claim 5, wherein the predictive model comprises a neural network, and wherein the model training module is configured to:

7. The apparatus of claim 6, wherein the predictive model further comprises a classifier, and wherein the model training module is configured to:

8. The apparatus of claim 5, wherein the predictive model comprises:

9. An apparatus for company operational risk prediction, comprising: memory, processor and computer program stored on the memory and executable on the processor, characterized in that the processor implements the method according to any of claims 1 to 4 when executing the program.

10. A computer-readable storage medium storing computer-executable instructions for performing the method of any one of claims 1-4.