CN117934137A - Bad asset recovery prediction method, device and equipment based on model fusion - Google Patents
Bad asset recovery prediction method, device and equipment based on model fusion Download PDFInfo
- Publication number
- CN117934137A CN117934137A CN202410102914.2A CN202410102914A CN117934137A CN 117934137 A CN117934137 A CN 117934137A CN 202410102914 A CN202410102914 A CN 202410102914A CN 117934137 A CN117934137 A CN 117934137A
- Authority
- CN
- China
- Prior art keywords
- prediction
- model
- error
- prediction result
- characteristic data
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000011084 recovery Methods 0.000 title claims abstract description 84
- 238000000034 method Methods 0.000 title claims abstract description 55
- 230000004927 fusion Effects 0.000 title claims abstract description 30
- 238000012545 processing Methods 0.000 claims description 17
- 238000004590 computer program Methods 0.000 claims description 16
- 230000015654 memory Effects 0.000 claims description 15
- 238000010606 normalization Methods 0.000 claims description 11
- 230000006870 function Effects 0.000 claims description 10
- 238000010276 construction Methods 0.000 claims description 4
- 230000008569 process Effects 0.000 description 11
- 238000004064 recycling Methods 0.000 description 9
- 238000004891 communication Methods 0.000 description 8
- 238000010586 diagram Methods 0.000 description 7
- 238000013473 artificial intelligence Methods 0.000 description 5
- 238000004422 calculation algorithm Methods 0.000 description 4
- 238000005457 optimization Methods 0.000 description 4
- 230000008901 benefit Effects 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 230000007774 longterm Effects 0.000 description 3
- 230000003287 optical effect Effects 0.000 description 3
- 238000013528 artificial neural network Methods 0.000 description 2
- 210000004027 cell Anatomy 0.000 description 2
- 230000003993 interaction Effects 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000006403 short-term memory Effects 0.000 description 2
- 238000012549 training Methods 0.000 description 2
- 230000000007 visual effect Effects 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 1
- 239000000654 additive Substances 0.000 description 1
- 230000000996 additive effect Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 238000003491 array Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000001413 cellular effect Effects 0.000 description 1
- 238000010219 correlation analysis Methods 0.000 description 1
- 125000004122 cyclic group Chemical group 0.000 description 1
- 238000003066 decision tree Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 239000011521 glass Substances 0.000 description 1
- 239000004973 liquid crystal related substance Substances 0.000 description 1
- 230000007787 long-term memory Effects 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 230000006386 memory function Effects 0.000 description 1
- 210000002569 neuron Anatomy 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 238000004321 preservation Methods 0.000 description 1
- 230000002265 prevention Effects 0.000 description 1
- 230000000306 recurrent effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 230000001953 sensory effect Effects 0.000 description 1
- 210000000352 storage cell Anatomy 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The embodiment of the invention discloses a bad asset recovery prediction method, device and equipment based on model fusion. The method comprises the following steps: constructing a bad asset recovery prediction model, wherein the bad asset recovery prediction model comprises a first prediction model and a second prediction model; acquiring characteristic data of bad assets, and determining a first input parameter according to the characteristic data; inputting the first input parameters into a first prediction model to obtain a first prediction result; determining a second input parameter according to the first prediction result and time sequence data in the characteristic data; inputting the second input parameters into a second prediction model to obtain a second prediction result; acquiring a first prediction error and a second prediction error corresponding to the first prediction model and the second prediction model respectively; and determining a final prediction result of the recovery of the bad asset according to the first prediction result, the second prediction result, the first prediction error and the second prediction error. According to the method, prediction is performed through multi-model fusion, so that the prediction accuracy can be improved.
Description
Technical Field
The invention relates to the technical field of artificial intelligence, in particular to a bad asset recovery prediction method, device and equipment based on model fusion.
Background
In bad asset disposal, bad asset securitization can significantly improve the liquidity and disposal benefits of bad assets, and is expected to obtain higher transfer prices. Reasonably assessing pricing for bad assets directly affects bank benefits.
However, the estimated recovery rate of the bad asset cannot be accurately predicted only by the conventional technology at present. Therefore, how to accurately predict the securitization recovery amount of the bad assets by the artificial intelligence technology has great exploratory significance.
The inventor finds that the bad asset data has the characteristics of structuring, time sequence and the like in the research, and when the artificial intelligence technology is adopted for prediction, the application range of the model is difficult to ensure by adopting a single model, and the bad asset is accurately predicted. Therefore, a bad asset recovery prediction method based on model fusion is provided.
Disclosure of Invention
The invention provides a bad asset recovery prediction method, device and equipment based on model fusion, which are used for processing multi-type bad asset data and improving the estimation accuracy and robustness of bad assets.
According to an aspect of the present invention, there is provided a bad asset recycling prediction method based on model fusion, the method comprising:
Constructing a bad asset recovery prediction model, wherein the bad asset recovery prediction model comprises a first prediction model and a second prediction model;
Acquiring characteristic data of bad assets, and determining a first input parameter according to the characteristic data; inputting the first input parameters into the first prediction model to obtain a first prediction result;
Determining a second input parameter according to the first prediction result and time sequence data in the characteristic data; inputting the second input parameters into the second prediction model to obtain a second prediction result;
Acquiring a first prediction error and a second prediction error which correspond to the first prediction model and the second prediction model respectively;
And determining a final prediction result of the recovery of the bad asset according to the first prediction result, the second prediction result, the first prediction error and the second prediction error.
Optionally, determining the first input parameter according to the feature data includes:
calculating the correlation coefficient of each characteristic data and the recovery amount;
and determining a first input parameter according to each characteristic data and the corresponding correlation coefficient.
Optionally, calculating a correlation coefficient between each of the feature data and the recovery amount includes:
And calculating the correlation coefficient of each characteristic data and the recovery amount by adopting the Pearson correlation coefficient.
Optionally, determining the first input parameter according to each of the feature data and the corresponding correlation coefficient includes:
Calculating the product of each characteristic data and the corresponding correlation coefficient;
And carrying out normalization processing on each product, and taking the normalization result as a first input parameter.
Optionally, obtaining a first prediction error and a second prediction error corresponding to the first prediction model and the second prediction model respectively includes:
acquiring historical characteristic data of the first prediction model and the second prediction model aiming at the historical bad asset, and predicting a first historical prediction result and a second historical prediction result;
and according to the historical characteristic data, the first historical prediction result and the second historical prediction result, adopting at least one error function of the following: the mean absolute error, the mean absolute percentage error, and the root mean square error determine a first prediction error and a second prediction error.
Optionally, determining a final prediction result of the bad asset recovery based on the first prediction result, the second prediction result, the first prediction error, and the second prediction error includes:
Calculating an error sum of the first prediction error and the second prediction error;
taking the ratio of the first prediction error to the error sum as a first model coefficient, and taking the ratio of the second prediction error to the error sum as a second model coefficient;
and taking the sum of the product of the first prediction result and the first model coefficient and the product of the second prediction result and the second model coefficient as the final prediction result of bad asset recovery.
Optionally, the first prediction model includes: XGBoost model;
The second predictive model includes: LSTM model.
According to another aspect of the present invention, there is provided a bad asset recycling prediction device based on model fusion, the device comprising:
the model construction module is used for constructing a bad asset recovery prediction model, wherein the bad asset recovery prediction model comprises a first prediction model and a second prediction model;
The first prediction result determining module is used for acquiring characteristic data of the bad asset and determining a first input parameter according to the characteristic data; inputting the first input parameters into the first prediction model to obtain a first prediction result;
The second prediction result determining module is used for determining a second input parameter according to the first prediction result and time sequence data in the characteristic data; inputting the second input parameters into the second prediction model to obtain a second prediction result;
The prediction error acquisition module is used for acquiring a first prediction error and a second prediction error which correspond to the first prediction model and the second prediction model respectively;
and the final prediction result determining module is used for determining a final prediction result of bad asset recovery according to the first prediction result, the second prediction result, the first prediction error and the second prediction error.
According to another aspect of the present invention, there is provided an electronic apparatus including:
At least one processor; and
A memory communicatively coupled to the at least one processor; wherein,
The memory stores a computer program executable by the at least one processor to enable the at least one processor to perform the model fusion-based bad asset recovery prediction method of any of the embodiments of the present invention.
According to another aspect of the present invention, there is provided a computer readable storage medium storing computer instructions for causing a processor to implement the model fusion-based bad asset recovery prediction method according to any one of the embodiments of the present invention when executed.
According to the technical scheme, the bad asset recovery prediction model is constructed, wherein the bad asset recovery prediction model comprises a first prediction model and a second prediction model; acquiring characteristic data of bad assets, and determining a first input parameter according to the characteristic data; inputting the first input parameters into a first prediction model to obtain a first prediction result; determining a second input parameter according to the first prediction result and time sequence data in the characteristic data; inputting the second input parameters into a second prediction model to obtain a second prediction result; acquiring a first prediction error and a second prediction error corresponding to the first prediction model and the second prediction model respectively; according to the first prediction result, the second prediction result, the first prediction error and the second prediction error, the final prediction result of the recovery of the bad asset is determined, the problem of prediction of the bad asset is solved, various types of bad asset data can be processed, and the prediction accuracy and the robustness of the bad asset are improved.
It should be understood that the description in this section is not intended to identify key or critical features of the embodiments of the invention or to delineate the scope of the invention. Other features of the present invention will become apparent from the description that follows.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly described below, and it is apparent that the drawings in the following description are only some embodiments of the present invention, and other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of a method for predicting bad asset recovery based on model fusion according to a first embodiment of the invention;
FIG. 2 is a schematic diagram of a bad asset recovery prediction model according to a first embodiment of the present invention;
FIG. 3 is a schematic structural diagram of a bad asset recycling prediction device based on model fusion according to a second embodiment of the present invention;
Fig. 4 is a schematic structural diagram of an electronic device implementing a model fusion-based bad asset recycling prediction method according to an embodiment of the present invention.
Detailed Description
In order that those skilled in the art will better understand the present invention, a technical solution in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in which it is apparent that the described embodiments are only some embodiments of the present invention, not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the present invention without making any inventive effort, shall fall within the scope of the present invention.
It should be noted that the terms "first," "second," and the like in the description and the claims of the present invention and the above figures are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged where appropriate such that the embodiments of the invention described herein may be implemented in sequences other than those illustrated or otherwise described herein. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.
Example 1
Fig. 1 is a flowchart of a model fusion-based bad asset recycling prediction method according to an embodiment of the present invention, where the method may be performed by a model fusion-based bad asset recycling prediction device, and the model fusion-based bad asset recycling prediction device may be implemented in hardware and/or software, and the model fusion-based bad asset recycling prediction device may be configured in an electronic device. As shown in fig. 1, the method includes:
and 110, constructing a bad asset recovery prediction model, wherein the bad asset recovery prediction model comprises a first prediction model and a second prediction model.
In an embodiment of the invention, the bad asset recovery predictive model may be a two-dimensional model. The multi-dimensional bad asset data may be analyzed by a two-dimensional model. The first prediction model and the second prediction model in the bad asset recovery prediction model may be generated by adopting a mode of independent training or simultaneous training, and the embodiment of the invention is not particularly limited. The first predictive model may process structured bad asset data. The second predictive model may augment the timing data based on the first predictive model. The structured and serialized bad asset data can be processed through the cooperation of the first prediction model and the second prediction model, so that the data processing range is increased, and the multi-type data can be processed; prediction accuracy, robustness and stability can also be improved by combining the two models.
Specifically, in an alternative implementation manner of the embodiment of the present invention, the first prediction model includes: a limit gradient lifting (Exterme Gradient Boosting, XGBoost) model; a second predictive model comprising: long Short-Term Memory (LSTM) model.
The XGBoost model is an integrated machine learning algorithm based on a decision tree, and takes Gradient Boost (Gradient Boost) as a framework. XGBoost developed from GBDT, which is also an optimization process using additive models and forward step algorithms to achieve learning, is different from GBDT. The differences are mainly in objective functions, optimization methods, missing value processing, overfitting prevention and prediction results. For example, the XGBoost penalty function adds a regularization term that uses regularization to control the complexity of the model, including the number of leaf nodes of the tree, the sum of squares of the weights of each leaf node (socre values of the leaf nodes). GBDT uses only the first derivative information in the optimization, XGBoost uses both the first and second derivative information in the optimization. XBGoost processes the missing values, and automatically selects the optimal default segmentation direction of the missing values through a learning model. XGBoost in addition to adding a regularization term to prevent overfitting, support a line sampling approach to prevent overfitting. XGBoost can achieve better results with less computational resources in the shortest time.
However, when using XGBoost to estimate bad assets, the inventors found that XGBoost is not suitable for processing sequence data with timing characteristics due to the inability of XGBoost to capture time dependencies in the data. Moreover, the model structure of XGBoost is relatively simple and may not be able to capture complex nonlinear relationships in the bad asset data. To solve the above problems, the inventors fused the LSTM model on the basis of XGBoost model.
LSTM is improved based on a cyclic neural network (recurrent neural network, RNN), improves the temporary memory function of RNN, and has long-term and short-term memory function. The cells in the LSTM network hidden layer are linear self-circulating storage cells, allowing long-term preservation of gradients. The LSTM comprises memory blocks from connected memory cells capable of storing time states, and the flow of information into and out of the memory blocks can be controlled by 3 gates. Thus, LSTM can address the problem in both long-term and short-term time series situations. By using a circulating structure and a gating mechanism of the LSTM model, long-term dependency in the sequence data is effectively processed and modeled, and time dimension data in securities asset data can be better fitted; and better model complex nonlinear relationships.
In the model of bad asset recovery prediction fused with XGBoost and LSTM, the difficulty is to solve the fusion of related information between the two models and to promote the fitting degree of complex nonlinear systems. In order to solve the above-mentioned problems, fig. 2 is a schematic structural diagram of a bad asset recycling prediction model according to a first embodiment of the present invention.
Step 120, obtaining characteristic data of bad assets, and determining a first input parameter according to the characteristic data; and inputting the first input parameters into the first prediction model to obtain a first prediction result.
Characteristic data of bad assets include, but are not limited to: the amount of the credential and the factors affecting the amount of the recovery. Wherein the factors affecting the recovery amount include: multi-dimensional information such as asset quality, expiration time, region, principal, interest, and academic.
As shown in fig. 2, determining the first input parameter according to the feature data may be performing vectorization processing, correlation analysis, normalization processing, and the like on the feature data. Specifically, in an optional implementation manner of the embodiment of the present invention, determining the first input parameter according to the feature data includes: calculating the correlation coefficient of each characteristic data and the recovery amount; and determining a first input parameter according to each characteristic data and the corresponding correlation coefficient.
The calculation of the correlation coefficient between each feature data and the recovery amount may be a correlation between the feature data and the recovery amount determined according to the history data, and a factor having a large influence on the recovery amount may be obtained. Therefore, the input parameters of the model can be adjusted according to the correlation, so that the prediction accuracy is improved. The correlation coefficient may be determined by a correlation algorithm.
Optionally, calculating a correlation coefficient between each feature data and the recovery amount includes: and calculating the correlation coefficient of each characteristic data and the recovery amount by adopting the Pearson correlation coefficient. The pearson correlation coefficient can measure the wireless correlation between two variables and the correlation degree. In embodiments of the present invention, the pearson correlation coefficient may be used to measure the correlation between the recovery amount and other characteristic data.
The pearson correlation coefficient between variables X, Y is: Wherein ρ is the correlation strength of the two, and a positive value indicates positive correlation between the characteristic data and the recovery amount, and a negative value indicates negative correlation between the characteristic data and the recovery amount. ρ can be determined by dividing the covariance of the two variables by the standard deviation product of the two variables.
In an optional implementation manner of the embodiment of the present invention, determining the first input parameter according to each feature data and the corresponding correlation coefficient includes: calculating the product of each characteristic data and the corresponding correlation coefficient; and carrying out normalization processing on each product, and taking the normalization result as a first input parameter.
The contribution rates of different characteristic data to recovery amount prediction are greatly different from the correlation of different dimensions in securitized asset data. Before XGBoost the first prediction, each characteristic data is multiplied by a corresponding correlation coefficient, and then input into the model for prediction. For n pieces of feature data h= (h 1,h2,...,hn), its correlation coefficient with the predicted value is α= (α 1,α2,...,αn), the prediction model input data is: x i＝αihi.
Because the numerical units of different characteristic data are different, the data are normalized, and the value of the different characteristic data is uniformly between 0 and 1. Normalization of the input data can avoid neuron saturation and increase accuracy of model prediction. In the embodiment of the invention, the data can be processed by adopting maximum and minimum normalization, and the formula is as followsWhere x i is the feature data correlation processed data, x max is the feature data correlation processed maximum, x min is the feature data correlation processed minimum, and y i is the normalized data. The normalized result may be input to the XGBoost model as a first input parameter.
Step 130, determining a second input parameter according to the first prediction result and the time sequence data in the characteristic data; and inputting the second input parameters into a second prediction model to obtain a second prediction result.
As shown in fig. 2, the second input parameter may be to add one-dimensional time series data based on the first prediction result. The time sequence data may be an expiration time in the feature data. Therefore, XGBoost and LSTM can be combined to process the structured data and the sequence data simultaneously, the application range of the model is expanded, and more types of data can be processed.
And 140, acquiring a first prediction error and a second prediction error which correspond to the first prediction model and the second prediction model respectively.
The first prediction error and the second prediction error may be prediction errors generated by the first prediction model and the second prediction model when predicting the historical feature data, respectively. Optionally, obtaining a first prediction error and a second prediction error corresponding to the first prediction model and the second prediction model respectively includes: acquiring historical characteristic data of a first prediction model and a second prediction model aiming at historical bad assets, and predicting a first historical prediction result and a second historical prediction result; according to the historical characteristic data, the first historical prediction result and the second historical prediction result, adopting at least one error function of the following: the mean absolute error, the mean absolute percentage error, and the root mean square error determine a first prediction error and a second prediction error.
The mean absolute error is given byThe mean absolute percentage error is given by/>The root mean square error is given byWherein X obj,i is the actual recovery amount for the history bad asset i, X model, i is the prediction result for the history bad asset i, and N is the history bad asset number.
In the embodiment of the invention, any one of the average absolute error, the average absolute percentage error and the root mean square error can be used for determining the first prediction error and the second prediction error, or a plurality of items can be used for comprehensively determining the first prediction error and the second prediction error.
And step 150, determining a final prediction result of the recovery of the bad asset according to the first prediction result, the second prediction result, the first prediction error and the second prediction error.
The final prediction result may be determined taking into account the influence of the first prediction error and the second prediction error on the first prediction result and the second prediction result. For example, a model weight coefficient may be set based on the first prediction error and the second prediction error, and a final prediction result of the bad asset recovery may be determined based on the model weight coefficient, the first prediction result, and the second prediction result.
Specifically, in the embodiment of the invention, the prediction results of the two models can be weighted and combined by taking the inverse error method into consideration to obtain the final prediction result. Optionally, determining a final prediction result of the bad asset recovery based on the first prediction result, the second prediction result, the first prediction error, and the second prediction error includes: calculating an error sum of the first prediction error and the second prediction error; taking the ratio of the first prediction error to the error sum as a first model coefficient, and taking the ratio of the second prediction error to the error sum as a second model coefficient; and taking the sum of the product of the first prediction result and the first model coefficient and the product of the second prediction result and the second model coefficient as the final prediction result of bad asset recovery.
The final prediction result may be expressed as h tl＝β1hx+β2hL. Where h x is the first predicted result, h L is the second predicted result, and h tl is the final predicted result. Beta 1 is the first model coefficient and,L 1 is the first prediction error and l 2 is the second prediction error. Beta 2 is the second model coefficient,/>By the error reciprocal method, the two models are combined to reduce the overall error, so that the prediction precision is improved.
According to the technical scheme, a bad asset recovery prediction model is built, wherein the bad asset recovery prediction model comprises a first prediction model and a second prediction model; acquiring characteristic data of bad assets, and determining a first input parameter according to the characteristic data; inputting the first input parameters into a first prediction model to obtain a first prediction result; determining a second input parameter according to the first prediction result and time sequence data in the characteristic data; inputting the second input parameters into a second prediction model to obtain a second prediction result; acquiring a first prediction error and a second prediction error corresponding to the first prediction model and the second prediction model respectively; according to the first prediction result, the second prediction result, the first prediction error and the second prediction error, the final prediction result of the recovery of the bad asset is determined, the problem of prediction of the bad asset is solved, various types of bad asset data can be processed, and the prediction accuracy and the robustness of the bad asset are improved; based on the recovery prediction model of artificial intelligence, manual operation is reduced, labor and time cost are saved, and prediction precision and efficiency of disposal pricing recovery are improved.
Specifically, XGBoost and LSTM in combination can process both structured data and sequence data, thereby expanding the application range of the model and processing more types of data. The advantages of the two models can be fully exerted by combining XGBoost with LSTM, so that the prediction accuracy is improved. XGBoost can process structured data, extract features and classify or regress, while LSTM can process sequence data, capture timing features in the sequence, thereby improving prediction accuracy. By combining XGBoost with LSTM, the robustness of the model can be improved, more abnormal conditions and data missing problems can be handled, and therefore the reliability and stability of the model are improved. XGBoost provides a feature importance assessment method, which can help a user to know the influence degree of each feature on a prediction result and improve the interpretability of a model, while an LSTM can display the time sequence features of model learning in a visual mode and help the user to understand the learning process and the prediction basis of the model.
In the technical scheme of the embodiment of the invention, the acquisition, storage, application and the like of the related bad asset data all conform to the regulations of related laws and regulations and do not violate the popular regulations of the public order.
Example two
Fig. 3 is a schematic structural diagram of a bad asset recovery prediction device based on model fusion according to a second embodiment of the present invention. As shown in fig. 3, the apparatus includes: the model construction module 310, the first prediction result determination module 320, the second prediction result determination module 330, the prediction error acquisition module 340 and the final prediction result determination module 350. Wherein:
A model construction module 310, configured to construct a bad asset recovery prediction model, where the bad asset recovery prediction model includes a first prediction model and a second prediction model;
A first prediction result determining module 320, configured to obtain feature data of the bad asset, and determine a first input parameter according to the feature data; inputting the first input parameters into a first prediction model to obtain a first prediction result;
A second prediction result determining module 330, configured to determine a second input parameter according to the first prediction result and the time sequence data in the feature data; inputting the second input parameters into a second prediction model to obtain a second prediction result;
A prediction error obtaining module 340, configured to obtain a first prediction error and a second prediction error corresponding to the first prediction model and the second prediction model, respectively;
The final prediction result determining module 350 is configured to determine a final prediction result of the recovery of the bad asset according to the first prediction result, the second prediction result, the first prediction error, and the second prediction error.
Optionally, the first prediction result determining module 320 includes:
a correlation coefficient determining unit for calculating a correlation coefficient of each feature data and the recovery amount;
and the first input parameter determining unit is used for determining the first input parameter according to each characteristic data and the corresponding correlation coefficient.
Optionally, the correlation coefficient determining unit is specifically configured to:
And calculating the correlation coefficient of each characteristic data and the recovery amount by adopting the Pearson correlation coefficient.
Optionally, the first input parameter determining unit is specifically configured to:
Calculating the product of each characteristic data and the corresponding correlation coefficient;
And carrying out normalization processing on each product, and taking the normalization result as a first input parameter.
Optionally, the prediction error obtaining module 340 includes:
The historical prediction data acquisition unit is used for acquiring the historical characteristic data of the first prediction model and the second prediction model aiming at the historical bad asset, and predicting a first historical prediction result and a second historical prediction result;
A prediction error determining unit, configured to use at least one error function according to the historical feature data, the first historical prediction result and the second historical prediction result, where the error function is as follows: the mean absolute error, the mean absolute percentage error, and the root mean square error determine a first prediction error and a second prediction error.
Optionally, the final prediction result determining module 350 is specifically configured to:
Calculating an error sum of the first prediction error and the second prediction error;
Taking the ratio of the first prediction error to the error sum as a first model coefficient, and taking the ratio of the second prediction error to the error sum as a second model coefficient;
and taking the sum of the product of the first prediction result and the first model coefficient and the product of the second prediction result and the second model coefficient as the final prediction result of bad asset recovery.
Optionally, the first prediction model includes: XGBoost model; a second predictive model comprising: LSTM model.
The bad asset recovery prediction device based on model fusion provided by the embodiment of the invention can execute the bad asset recovery prediction method based on model fusion provided by any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the execution method.
Example III
Fig. 4 shows a schematic diagram of the structure of an electronic device 10 that may be used to implement an embodiment of the invention. Electronic devices are intended to represent various forms of digital computers, such as laptops, desktops, workstations, personal digital assistants, servers, blade servers, mainframes, and other appropriate computers. Electronic equipment may also represent various forms of mobile devices, such as personal digital processing, cellular telephones, smartphones, wearable devices (e.g., helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions, are meant to be exemplary only, and are not meant to limit implementations of the inventions described and/or claimed herein.
As shown in fig. 4, the electronic device 10 includes at least one processor 11, and a memory, such as a Read Only Memory (ROM) 12, a Random Access Memory (RAM) 13, etc., communicatively connected to the at least one processor 11, in which the memory stores a computer program executable by the at least one processor, and the processor 11 may perform various appropriate actions and processes according to the computer program stored in the Read Only Memory (ROM) 12 or the computer program loaded from the storage unit 18 into the Random Access Memory (RAM) 13. In the RAM 13, various programs and data required for the operation of the electronic device 10 may also be stored. The processor 11, the ROM 12 and the RAM 13 are connected to each other via a bus 14. An input/output (I/O) interface 15 is also connected to bus 14.
Various components in the electronic device 10 are connected to the I/O interface 15, including: an input unit 16 such as a keyboard, a mouse, etc.; an output unit 17 such as various types of displays, speakers, and the like; a storage unit 18 such as a magnetic disk, an optical disk, or the like; and a communication unit 19 such as a network card, modem, wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information/data with other devices via a computer network, such as the internet, and/or various telecommunication networks.
The processor 11 may be a variety of general and/or special purpose processing components having processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a Central Processing Unit (CPU), a Graphics Processing Unit (GPU), various specialized Artificial Intelligence (AI) computing chips, various processors running machine learning model algorithms, digital Signal Processors (DSPs), and any suitable processor, controller, microcontroller, etc. The processor 11 performs the various methods and processes described above, such as a bad asset recovery prediction method based on model fusion.
In some embodiments, the model fusion-based bad asset recovery prediction method may be implemented as a computer program tangibly embodied on a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program may be loaded and/or installed onto the electronic device 10 via the ROM 12 and/or the communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the model fusion-based bad asset recovery prediction method described above may be performed. Alternatively, in other embodiments, the processor 11 may be configured to perform the model fusion-based bad asset recovery prediction method in any other suitable manner (e.g., by means of firmware).
Various implementations of the systems and techniques described here above may be implemented in digital electronic circuitry, integrated circuit systems, field Programmable Gate Arrays (FPGAs), application Specific Integrated Circuits (ASICs), application Specific Standard Products (ASSPs), systems On Chip (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and/or combinations thereof. These various embodiments may include: implemented in one or more computer programs, the one or more computer programs may be executed and/or interpreted on a programmable system including at least one programmable processor, which may be a special purpose or general-purpose programmable processor, that may receive data and instructions from, and transmit data and instructions to, a storage system, at least one input device, and at least one output device.
A computer program for carrying out methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general purpose computer, special purpose computer, or other programmable data processing apparatus, such that the computer programs, when executed by the processor, cause the functions/acts specified in the flowchart and/or block diagram block or blocks to be implemented. The computer program may execute entirely on the machine, partly on the machine, as a stand-alone software package, partly on the machine and partly on a remote machine or entirely on the remote machine or server.
In the context of the present invention, a computer-readable storage medium may be a tangible medium that can contain, or store a computer program for use by or in connection with an instruction execution system, apparatus, or device. The computer readable storage medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. Alternatively, the computer readable storage medium may be a machine readable signal medium. More specific examples of a machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.
To provide for interaction with a user, the systems and techniques described here can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to a user; and a keyboard and a pointing device (e.g., a mouse or a trackball) through which a user can provide input to the electronic device. Other kinds of devices may also be used to provide for interaction with a user; for example, feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form, including acoustic input, speech input, or tactile input.
The systems and techniques described here can be implemented in a computing system that includes a background component (e.g., as a data server), or that includes a middleware component (e.g., an application server), or that includes a front-end component (e.g., a user computer having a graphical user interface or a web browser through which a user can interact with an implementation of the systems and techniques described here), or any combination of such background, middleware, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: local Area Networks (LANs), wide Area Networks (WANs), blockchain networks, and the internet.
The computing system may include clients and servers. The client and server are typically remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other. The server can be a cloud server, also called a cloud computing server or a cloud host, and is a host product in a cloud computing service system, so that the defects of high management difficulty and weak service expansibility in the traditional physical hosts and VPS service are overcome.
It should be appreciated that various forms of the flows shown above may be used to reorder, add, or delete steps. For example, the steps described in the present invention may be performed in parallel, sequentially, or in a different order, so long as the desired results of the technical solution of the present invention are achieved, and the present invention is not limited herein.
The above embodiments do not limit the scope of the present invention. It will be apparent to those skilled in the art that various modifications, combinations, sub-combinations and alternatives are possible, depending on design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the scope of the present invention.
Claims (10)
1. The bad asset recovery prediction method based on model fusion is characterized by comprising the following steps of:
Constructing a bad asset recovery prediction model, wherein the bad asset recovery prediction model comprises a first prediction model and a second prediction model;
Acquiring characteristic data of bad assets, and determining a first input parameter according to the characteristic data; inputting the first input parameters into the first prediction model to obtain a first prediction result;
Determining a second input parameter according to the first prediction result and time sequence data in the characteristic data; inputting the second input parameters into the second prediction model to obtain a second prediction result;
Acquiring a first prediction error and a second prediction error which correspond to the first prediction model and the second prediction model respectively;
And determining a final prediction result of the recovery of the bad asset according to the first prediction result, the second prediction result, the first prediction error and the second prediction error.
2. The method of claim 1, wherein determining a first input parameter from the characteristic data comprises:
calculating the correlation coefficient of each characteristic data and the recovery amount;
and determining a first input parameter according to each characteristic data and the corresponding correlation coefficient.
3. The method of claim 2, wherein calculating a correlation coefficient of each of the characteristic data and the recovery amount comprises:
And calculating the correlation coefficient of each characteristic data and the recovery amount by adopting the Pearson correlation coefficient.
4. The method of claim 2, wherein determining the first input parameter based on each of the characteristic data and the corresponding correlation coefficient comprises:
Calculating the product of each characteristic data and the corresponding correlation coefficient;
And carrying out normalization processing on each product, and taking the normalization result as a first input parameter.
5. The method of claim 1, wherein obtaining first and second prediction errors corresponding to the first and second prediction models, respectively, comprises:
acquiring historical characteristic data of the first prediction model and the second prediction model aiming at the historical bad asset, and predicting a first historical prediction result and a second historical prediction result;
and according to the historical characteristic data, the first historical prediction result and the second historical prediction result, adopting at least one error function of the following: the mean absolute error, the mean absolute percentage error, and the root mean square error determine a first prediction error and a second prediction error.
6. The method of claim 1, wherein determining a final prediction result for the recovery of the bad asset based on the first prediction result, the second prediction result, the first prediction error, and the second prediction error comprises:
Calculating an error sum of the first prediction error and the second prediction error;
taking the ratio of the first prediction error to the error sum as a first model coefficient, and taking the ratio of the second prediction error to the error sum as a second model coefficient;
and taking the sum of the product of the first prediction result and the first model coefficient and the product of the second prediction result and the second model coefficient as the final prediction result of bad asset recovery.
7. The method of claim 1, wherein the first predictive model comprises: XGBoost model;
The second predictive model includes: LSTM model.
8. A bad asset recovery prediction device based on model fusion, comprising:
the model construction module is used for constructing a bad asset recovery prediction model, wherein the bad asset recovery prediction model comprises a first prediction model and a second prediction model;
The first prediction result determining module is used for acquiring characteristic data of the bad asset and determining a first input parameter according to the characteristic data; inputting the first input parameters into the first prediction model to obtain a first prediction result;
The second prediction result determining module is used for determining a second input parameter according to the first prediction result and time sequence data in the characteristic data; inputting the second input parameters into the second prediction model to obtain a second prediction result;
The prediction error acquisition module is used for acquiring a first prediction error and a second prediction error which correspond to the first prediction model and the second prediction model respectively;
and the final prediction result determining module is used for determining a final prediction result of bad asset recovery according to the first prediction result, the second prediction result, the first prediction error and the second prediction error.
9. An electronic device, the electronic device comprising:
At least one processor; and
A memory communicatively coupled to the at least one processor; wherein,
The memory stores a computer program executable by the at least one processor to enable the at least one processor to perform the model fusion-based bad asset recovery prediction method of any of claims 1-7.
10. A computer readable storage medium storing computer instructions for causing a processor to implement the model fusion-based bad asset recovery prediction method of any one of claims 1-7 when executed.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410102914.2A CN117934137A (en) | 2024-01-24 | 2024-01-24 | Bad asset recovery prediction method, device and equipment based on model fusion |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410102914.2A CN117934137A (en) | 2024-01-24 | 2024-01-24 | Bad asset recovery prediction method, device and equipment based on model fusion |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117934137A true CN117934137A (en) | 2024-04-26 |
Family
ID=90755284
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410102914.2A Pending CN117934137A (en) | 2024-01-24 | 2024-01-24 | Bad asset recovery prediction method, device and equipment based on model fusion |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117934137A (en) |
-
2024
- 2024-01-24 CN CN202410102914.2A patent/CN117934137A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113705628B (en) | Determination method and device of pre-training model, electronic equipment and storage medium | |
| CN116307215A (en) | Load prediction method, device, equipment and storage medium of power system | |
| CN115456167B (en) | Lightweight model training method, image processing device and electronic equipment | |
| CN114399101A (en) | TCN-BIGRU-based gas load prediction method and device | |
| CN113408808A (en) | Training method, data generation method, device, electronic device and storage medium | |
| CN116468967B (en) | Sample image screening method and device, electronic equipment and storage medium | |
| CN117333076A (en) | Evaluation method, device, equipment and medium based on mixed expert model | |
| CN116303786B (en) | Block chain financial big data management system based on multidimensional data fusion algorithm | |
| CN117934137A (en) | Bad asset recovery prediction method, device and equipment based on model fusion | |
| CN116933896B (en) | Super-parameter determination and semantic conversion method, device, equipment and medium | |
| CN117094452B (en) | Drought state prediction method, and training method and device of drought state prediction model | |
| CN116307159A (en) | Load prediction method and device, electronic equipment and storage medium | |
| CN117635342A (en) | Investment portfolio optimization method, device, equipment and storage medium | |
| CN116542362A (en) | Load prediction method and device, electronic equipment and storage medium | |
| CN118100151A (en) | Power grid load prediction method, device, equipment and storage medium | |
| CN117217777A (en) | Evaluation method, device, equipment and medium based on contrast learning | |
| CN116151854A (en) | User type determining method, device, equipment and storage medium | |
| CN117153421A (en) | Data monitoring method and device based on neural network algorithm | |
| CN116960957A (en) | Vehicle charging load prediction method and device, electronic equipment and storage medium | |
| CN115600129A (en) | Information identification method and device, electronic equipment and storage medium | |
| CN116665823A (en) | Crystal structure prediction method, crystal structure prediction device, electronic equipment and storage medium | |
| CN117690181A (en) | Iris image processing method and device, electronic equipment and storage medium | |
| CN116111562A (en) | Method, device, equipment and storage medium for determining reliability of power distribution network | |
| CN116011570A (en) | XAI model consistency training method, device, equipment and storage medium | |
| CN117195120A (en) | Detection method, detection device, detection equipment and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |