CN111695824B - Method, device, equipment and computer storage medium for analyzing risk tail end customer - Google Patents
Method, device, equipment and computer storage medium for analyzing risk tail end customer Download PDFInfo
- Publication number
- CN111695824B CN111695824B CN202010549845.1A CN202010549845A CN111695824B CN 111695824 B CN111695824 B CN 111695824B CN 202010549845 A CN202010549845 A CN 202010549845A CN 111695824 B CN111695824 B CN 111695824B
- Authority
- CN
- China
- Prior art keywords
- risk
- tail end
- analysis
- model
- rule
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 35
- 238000004458 analytical method Methods 0.000 claims abstract description 261
- 238000012502 risk assessment Methods 0.000 claims abstract description 36
- 238000012549 training Methods 0.000 claims description 24
- 238000004590 computer program Methods 0.000 claims description 12
- 238000003066 decision tree Methods 0.000 claims description 10
- 238000012545 processing Methods 0.000 claims description 10
- 230000002159 abnormal effect Effects 0.000 claims description 9
- 238000010276 construction Methods 0.000 claims description 5
- 238000005457 optimization Methods 0.000 claims description 5
- 238000005516 engineering process Methods 0.000 abstract description 4
- 230000003993 interaction Effects 0.000 description 8
- 238000004891 communication Methods 0.000 description 6
- 238000012216 screening Methods 0.000 description 4
- 238000004422 calculation algorithm Methods 0.000 description 2
- 239000000284 extract Substances 0.000 description 2
- 238000001914 filtration Methods 0.000 description 2
- 230000005856 abnormality Effects 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0635—Risk analysis of enterprise or organisation activities
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
- G06Q10/047—Optimisation of routes or paths, e.g. travelling salesman problem
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/06—Buying, selling or leasing transactions
- G06Q30/0601—Electronic shopping [e-shopping]
- G06Q30/0609—Buyer or seller confidence or verification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
Landscapes
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Human Resources & Organizations (AREA)
- Strategic Management (AREA)
- Economics (AREA)
- Marketing (AREA)
- General Business, Economics & Management (AREA)
- Development Economics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Entrepreneurship & Innovation (AREA)
- Accounting & Taxation (AREA)
- Finance (AREA)
- Operations Research (AREA)
- Tourism & Hospitality (AREA)
- Quality & Reliability (AREA)
- Game Theory and Decision Science (AREA)
- Educational Administration (AREA)
- Technology Law (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention relates to the field of financial science and technology, and discloses a risk tail end customer analysis method, a risk tail end customer analysis device, risk tail end customer analysis equipment and a computer storage medium. The method comprises the following steps: when a risk analysis request is received, determining that the risk analysis request corresponds to a target client to be analyzed, and acquiring client information of the target client; inputting the client information into different tail end analysis models corresponding to preset different types of risk features, and analyzing the client information through the level rules of each level in the tail end analysis models to obtain tail end level scores of various risk features; and fusing the tail end level scores of various risk features to obtain a comprehensive risk score of the target client, and outputting prompt information according to the comprehensive risk score. According to the invention, targeted analysis is carried out on the characteristics of each guest group through the level rules of each level of the tail end analysis model, so that the accuracy of risk tail end customer analysis is improved.
Description
Technical Field
The present invention relates to the field of financial technology (Fintech), and in particular, to a method, an apparatus, a device, and a computer storage medium for analyzing a risk tail end customer.
Background
In recent years, internet financial technology (fittech) has been rapidly developed, and more technologies (big data, distributed, blockchain, artificial intelligence, etc.) are applied in the financial field.
In order to improve the safety of financial business, the data volume and business volume of the financial business are exponentially increased, and in order to improve the safety of the financial business, risk tail end customer analysis is needed, wherein the risk tail end customer analysis refers to analysis of customers with extreme risk performance on information related to users, and in order to eliminate the customers with extreme risk performance, an admission threshold is usually set for indexes related to direct risk characteristics in an approval process so as to screen tail end customers except the threshold.
Disclosure of Invention
The invention mainly aims to provide a risk tail end customer analysis method, a risk tail end customer analysis device, risk tail end customer analysis equipment and a risk tail end customer analysis computer storage medium, and aims to solve the technical problem that the tail end customer risk analysis lacks customer group pertinence, all customer groups are analyzed according to the same analysis rule, and the risk tail end customer identification is inaccurate.
In order to achieve the above object, the present invention provides a method for analyzing a risk tail end customer, the method for analyzing a risk tail end customer comprising the steps of:
when a risk analysis request is received, determining that the risk analysis request corresponds to a target client to be analyzed, and acquiring client information of the target client;
inputting the client information into different tail end analysis models corresponding to preset different types of risk features, and analyzing the client information through the level rules of each level in the tail end analysis models to obtain tail end level scores of various risk features;
and fusing the tail end level scores of various risk features to obtain a comprehensive risk score of the target client, and outputting prompt information according to the comprehensive risk score.
Optionally, before the step of determining that the risk analysis request corresponds to a target client to be analyzed and acquiring client information of the target client when the risk analysis request is received, the method includes:
acquiring a sample data set and constructing an initial tail end layering model;
taking any layer output of the initial tail end hierarchical model as a target variable, taking a predefined grouping dimension as an independent variable, constructing a multi-fork decision tree, dividing sample data in a sample data set, and forming a sample data subset;
Respectively training the initial tail end layering model through each sample data subset to obtain a guest group analysis sub-model;
and combining layering rules in each guest group analysis sub-model to obtain a tail end layering model.
Optionally, the step of combining the layering rules in each passenger group analysis sub-model to obtain a tail end layering model includes:
acquiring the outermost layering rules with the largest influence surface in each guest group analysis sub-model, and combining the outermost layering rules to serve as the outermost layer of the model;
searching a path with optimal performance, so that each guest group sub-model rule moves inwards by one layer;
and the method is repeated until the model is moved to the innermost layer of each guest group analysis sub-model, and the optimal path formed by iteration forms the hierarchy of the final model to obtain the tail end hierarchical model with guest group difference.
Optionally, the steps of repeating the steps until moving to the innermost layer of each guest group analysis sub-model, iterating the formed optimal path to form a hierarchy of a final model, and obtaining a tail end hierarchical model with guest group differences, including:
acquiring a hierarchy rule corresponding to each hierarchy movement of each guest group analysis sub-model and an influence surface of the hierarchy rule;
Comparing the influence surfaces of the hierarchy movement corresponding hierarchy rules of the guest group analysis submodels;
if the difference between the influence surface of the current hierarchical rule of a target guest group analysis sub-model and the influence surface of other guest groups except the target guest group analysis sub-model is larger than a preset difference threshold, stopping moving the target guest group analysis sub-model in the iteration, and exhausting all paths of which the other guest group analysis sub-models except the target guest group analysis sub-model move inwards for one layer, so as to select a path with optimal performance;
if the target guest group analysis submodel with the abnormal influence surface of the hierarchical rule does not exist, all submodels are exhausted to move inwards a layer of paths, and the paths with optimal performance are selected from the paths.
Iterative optimization is carried out until all sub-models are moved to the innermost layer, and the optimal paths of each iteration are combined into a tail end hierarchical model with guest group differences.
Optionally, the step of acquiring the sample data set and constructing an initial tail end hierarchical model includes:
constructing a risk feature set aiming at sample data in a sample data set, and converting each risk feature in the risk feature set into an initial variable value to form an initial variable set;
the quantiles of the preset duty ratio after exhausting all initial variable values in the initial variable set are respectively used as initial rule thresholds, and an initial rule set is determined according to all the initial rule thresholds;
Selecting initial variables which are associated with risk analysis from the initial variable set according to the initial rule set as modulus-entering variable values, and forming the modulus-entering variable values into a modulus-entering variable set;
the quantiles of the preset duty ratio after exhausting all the modulus variable values in the modulus variable set are respectively used as modulus rule threshold values, and a modulus rule set is determined according to all the modulus rule threshold values;
and training an initial model according to the sample data set and the modeling rule set to obtain an initial tail end layering model.
Optionally, the step of inputting the customer information to different tail end analysis models corresponding to preset different types of risk features, analyzing the customer information through a hierarchy rule of each hierarchy in the tail end analysis model, and obtaining tail end hierarchy scores of various risk features includes:
inputting the client information into a tail end analysis model corresponding to preset different types of risk features, and analyzing the client information through a hierarchy rule of each hierarchy in the tail end analysis model to obtain bad sample rates corresponding to each hierarchy;
and processing bad sample rates corresponding to the levels to obtain tail end level scores of various risk features, wherein the tail end level scores are as follows:
Optionally, the step of fusing the tail end level scores of the various risk features to obtain a comprehensive risk score of the target client and outputting prompt information according to the comprehensive risk score includes:
performing product operation on the scores of the tail end levels to obtain comprehensive scores of the target clients;
outputting prompt information that the target client is not a risk user when the comprehensive score is greater than a preset threshold value;
and outputting prompt information that the target client is a risk user when the comprehensive score is smaller than or equal to a preset threshold value.
In addition, to achieve the above object, the present invention also provides a risk tail end customer analysis apparatus including:
the request receiving module is used for determining that the risk analysis request corresponds to a target client to be analyzed when the risk analysis request is received, and acquiring client information of the target client;
the tail end hierarchy grading module is used for inputting the client information into different tail end analysis models corresponding to preset different types of risk features, analyzing the client information through hierarchy rules of all the levels in the tail end analysis models, and obtaining tail end hierarchy grading of all the risk features;
And the prompt output module is used for fusing the tail end level scores of various risk characteristics to obtain the comprehensive risk score of the target client, and outputting prompt information according to the comprehensive risk score.
In addition, to achieve the above object, the present invention also provides a risk tail end customer analysis apparatus including: the system comprises a memory, a processor and a risk end customer analysis corresponding computer program stored on the memory and capable of running on the processor, wherein the risk end customer analysis corresponding computer program realizes the steps of the risk end customer analysis method when being executed by the processor.
In addition, to achieve the above object, the present invention further provides a computer storage medium having stored thereon a computer program corresponding to the analysis of a risk end customer, which when executed by a processor, implements the steps of the method for analyzing a risk end customer as described above.
The invention provides a risk tail end customer analysis method, a risk tail end customer analysis device, risk tail end customer analysis equipment and a computer storage medium, wherein in the embodiment of the invention, when a risk analysis request is received, the risk analysis request is determined to correspond to a target customer to be analyzed, and customer information of the target customer is acquired; inputting the client information into different tail end analysis models corresponding to preset different types of risk features, and analyzing the client information through the level rules of each level in the tail end analysis models to obtain tail end level scores of various risk features; fusing the tail end level scores of various risk features to obtain a comprehensive risk score of the target client, and outputting prompt information according to the comprehensive risk score; in the embodiment of the invention, the tail end analysis model is preset, and the customer information analysis is performed through the level rules of each level in the tail end analysis model, so that the targeted analysis is performed for different customer groups in the risk tail end customer analysis process, and the accuracy of the risk tail end customer analysis is improved.
Drawings
FIG. 1 is a schematic diagram of a device architecture of a hardware operating environment according to an embodiment of the present invention;
fig. 2 is a flowchart of a first embodiment of a risk tail end customer analysis method according to the present invention.
The achievement of the objects, functional features and advantages of the present invention will be further described with reference to the accompanying drawings, in conjunction with the embodiments.
Detailed Description
It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention.
Referring to fig. 1, fig. 1 is a schematic device structure of a hardware running environment according to an embodiment of the present invention. In the embodiment of the invention, the risk tail end client analysis device may be a PC or a server, as shown in fig. 1, and the risk tail end client analysis device may include: a processor 1001, such as a CPU, a network interface 1004, a user interface 1003, a memory 1005, a communication bus 1002. Wherein the communication bus 1002 is used to enable connected communication between these components. The user interface 1003 may include a Display, an input unit such as a Keyboard (Keyboard), and the optional user interface 1003 may further include a standard wired interface, a wireless interface. The network interface 1004 may optionally include a standard wired interface, a wireless interface (e.g., WI-FI interface). The memory 1005 may be a high-speed RAM memory or a stable memory (non-volatile memory), such as a disk memory. The memory 1005 may also optionally be a storage device separate from the processor 1001 described above.
It will be appreciated by those skilled in the art that the device structure shown in fig. 1 is not limiting of the device and may include more or fewer components than shown, or may combine certain components, or a different arrangement of components.
As shown in fig. 1, a memory 1005, which is a computer storage medium, may include an operating network communication module, a user interface module, and a computer program for risk tail end customer analysis.
In the device shown in fig. 1, the network interface 1004 is mainly used for connecting to a background server, and performing data communication with the background server; the user interface 1003 is mainly used for connecting a client (user side) and performing data communication with the client; and the processor 1001 may be configured to invoke a corresponding computer program for risk end customer analysis stored in the memory 1005 and perform operations in the risk end customer analysis method described below.
Based on the above hardware structure, a first embodiment of the risk tail end customer analysis method of the present invention is provided, including:
when a risk analysis request is received, determining that the risk analysis request corresponds to a target client to be analyzed, and acquiring client information of the target client;
inputting the client information into different tail end analysis models corresponding to preset different types of risk features, and analyzing the client information through the level rules of each level in the tail end analysis models to obtain tail end level scores of various risk features;
And fusing the tail end level scores of various risk features to obtain a comprehensive risk score of the target client, and outputting prompt information according to the comprehensive risk score.
Before the first embodiment of the risk tail end customer analysis method of the present invention is executed, a tail end analysis model is pre-built, and the tail end analysis model building includes: 1. constructing an initial tail end layering model; and 2, constructing a tail end analysis model according to the initial tail end layering model.
The step of constructing the initial tail end layering model in the first embodiment of the risk tail end customer analysis method comprises the following steps:
and constructing a risk feature set aiming at sample data in the sample data set, and converting each risk feature in the risk feature set into an initial variable value to form an initial variable set.
The risk end customer analysis method in the present embodiment is applied to risk end customer analysis equipment in financial institutions (banking institutions, insurance institutions, securities institutions, etc.) in the financial industry.
The risk tail client analysis device acquires a sample data set (the sample data set refers to a set of sample data which is stored in advance, and a label is in the sample data), the risk tail client analysis device constructs a risk feature set for the sample data in the sample data set, namely, the risk tail client analysis device extracts risk features from each sample data in the sample data set, combines the extracted risk features to form the risk feature set, and converts each risk feature in the risk feature set into an initial variable value to form the initial variable set, namely, the risk tail client analysis device converts each risk feature into a vector or other kind of variable value.
And exhausting the quantiles of the preset duty ratio after each initial variable value in the initial variable set is used as initial rule thresholds respectively, and determining an initial rule set according to each initial rule threshold.
The risk tail end customer analysis device exhausts the fractional numbers of all initial variable values in the initial variable set and presets the fractional numbers of the fractional numbers (the preset fractional numbers refer to the preset fractional numbers of the risk tail end customers, the preset fractional numbers can be set according to specific scenes, for example, the preset fractional numbers can be set to be 10%) to be respectively used as initial rule thresholds, the risk tail end customer analysis device determines initial rule sets according to all the initial rule thresholds, namely, the risk tail end customer analysis device takes all the initial variables larger than or equal to the initial rule thresholds as initial rules, and the risk tail end customer analysis device takes all the initial rule sets and forms the initial rule sets.
In the embodiment, risk variables corresponding to all risk characteristics are analyzed by the risk tail end customer analysis equipment, so that risk analysis is more accurate and comprehensive.
And selecting initial variables which are associated with risk analysis from the initial variable set according to the initial rule set as modulus-entering variable values, and forming the modulus-entering variable values into the modulus-entering variable set.
The risk tail end customer analysis device selects the modulus-entering variable values from the initial variable set to form a modulus-entering variable set according to an initial rule set, and specifically comprises the following steps:
step a1, calculating the concentration lifting degree and the influence surface of each initial rule in the initial rule set hitting the negative example sample in the sample data set;
and a2, screening out target rules with highest concentration promotion degrees under different influence surfaces from the initial rule set, taking initial variable values corresponding to the target rules as modulus-in variable values, and forming the modulus-in variable values into a modulus-in variable set.
Namely, the risk tail end customer analysis equipment calculates the concentration lifting degree and the influence surface of negative examples in each initial rule hit sample data set in the initial rule set; the risk tail end customer analysis equipment screens out target rules with highest concentration lifting degrees under different influence surfaces from the initial rule set, takes variable values corresponding to the target rules as modulus-in variable values, and forms the modulus-in variable values into a modulus-in variable set. In the embodiment, the risk tail end customer analysis equipment selects the modeling variable according to the concentration lifting degree and the influence surface of the variable value, so that the model is simpler and more accurate.
And respectively taking the quantiles of the preset duty ratio after exhausting all the modulus-entering variable values in the modulus-entering variable set as modulus-entering rule thresholds, and determining a modulus-entering rule set according to all the modulus-entering rule thresholds.
The risk tail end customer analysis device exhaustively presets the duty ratio after the modulus variable values in the modulus variable set are all input (the preset duty ratio refers to the preset duty ratio of the risk tail end customer, the preset duty ratio can be set according to a specific scene, for example, the preset duty ratio can be set to 10%) and respectively serves as modulus rule threshold values, the risk tail end customer analysis device determines modulus rule sets according to the modulus rule threshold values, namely, the risk tail end customer analysis device takes the modulus variable values which are greater than or equal to the modulus rule threshold values as modulus rules, and the risk tail end customer analysis device takes the modulus rule sets and forms the modulus rule sets.
And training an initial model according to the sample data set and the modeling rule set to obtain an initial tail end layering model.
The risk tail end customer analysis device trains an initial model according to the sample data set and the modeling rule set to obtain an initial tail end layering model, and specifically comprises the following steps:
step b1, carrying out pairwise interaction on target rules with front concentration lifting degree and different variables in the modeling rule set in an intersection form to obtain an interaction rule set;
Step b2, selecting a training rule with highest concentration lifting degree from the modeling rule set and the interaction rule set, and outputting the training rule as a first layer of an initial model;
and b3, filtering and deleting the sample data hit by the training rule in the sample data set to obtain residual sample data, and forming a new sample data set by the residual sample data to perform iterative training to obtain an initial tail end layering model.
The risk tail end customer analysis equipment carries out pairwise interaction on target rules with front concentration lifting degree and different variables in the modeling rule set in an intersection form to obtain an interaction rule set; the risk tail end customer analysis equipment selects a training rule with highest concentration lifting degree from the modeling rule set and the interaction rule set, and outputs the training rule as a first layer of an initial model; then, the risk tail end customer analysis equipment filters and deletes the sample data hit in the training rule in the sample data set to obtain residual sample data, and the residual sample data is formed into a new sample data set to carry out iterative training to obtain a new training rule; and the risk tail end customer analysis equipment combines the training rules obtained each time with the training rules obtained before to obtain a layering rule combination, and when the concentration lifting degree of the layering rule combination is smaller than the preset concentration lifting degree or the influence surface is larger than the preset influence surface threshold value, training is terminated to obtain an initial tail end layering model.
Specifically, for easy understanding, a specific example of the construction of the initial tail end hierarchical model is given in this embodiment, and the input of the initial tail end hierarchical model is similar to the common supervised learning algorithm, and a sample set including the derivative index X and the positive example negative example label Y needs to be input. After the related data sources are derived according to a certain risk characteristic, the variables are subjected to risk homodromous and integer processing. That is, for the variable with smaller value and higher risk of default, the reverse processing is performed by the mode of 'variable maximum value-variable value', and the variable value is taken down to be an integer. After the preparation of the data set is completed, the algorithm flow can be entered, and the specific steps are as follows:
1. rule expansion and variable screening, namely, taking the quantiles of 10% of all variables after exhaustion as rule threshold values respectively to obtain exhaustion rule variable values > =threshold values of all variables, calculating the concentration lifting degree (Lift value) of a negative example sample of a rule hit sample and corresponding influence surfaces, selecting rules with better ranking of the Lift values under different influence surface levels (such as 0-1%, 1-2% and …), and screening out variables which can produce high-performance rules and have stronger service interpretation from being put into a model by combining experience judgment.
2. Rule exhaustion optimizing, exhaustively obtaining 10% quantiles after each modulus variable is used as rule threshold, and exhaustion rule of variable is 'variable value > =threshold', so that a single variable rule set is formed. And then, carrying out pairwise interaction on the rules which are ranked at the front and have different variables in the single-variable rule in an intersection form to obtain an interaction rule set.
3. And outputting the hierarchical rule, selecting a rule with the highest Lift from the two rule sets, outputting the rule as a rule of the first layer, and filtering and screening out samples hitting the rule in the data set. And (3) taking the rest data set as the input of the steps (2) and (3) again, repeating the exhaustive optimization and rule output of the rules, and repeating iteration until the Lift value of the selected rule is smaller than the required value or the influence surface is larger than the upper limit. Each iteration can output a rule, and the rule is combined with the rule output by each previous iteration to obtain the hierarchical rule combination of the iteration. Since the rule combination of each layer is expanded by the rule of the previous layer, the guest group division of the tail end of the risk characteristic can be realized, and the initial tail end layering model is obtained.
The step of constructing the tail end layering model in the first embodiment of the risk tail end client analysis method comprises the following steps:
And acquiring a sample data set and constructing an initial tail end layering model.
And taking any layer output of the initial tail end hierarchical model as a target variable, taking a predefined grouping dimension as an independent variable, constructing a multi-fork decision tree, and dividing sample data in a sample data set to form a sample data subset.
In this embodiment, the risk tail end client analysis device acquires a sample data set, builds an initial tail end hierarchical model, and refers to the embodiment described in detail herein, and in this embodiment, the risk tail end client analysis device uses any one layer of output of the initial tail end hierarchical model as a target variable, uses a predefined grouping dimension as an independent variable, builds a multi-branch decision tree, divides sample data in the sample data set to form a sample data subset, i.e., the risk tail end client analysis device divides the sample data in the sample data set according to the predefined grouping dimension to form a sample data subset, and builds a multi-branch decision tree, for example, because credit performance among different guest groups often has differences, and influence surfaces of the multi-branch decision tree under the same rule often have differences. Therefore, the risk tail end customer analysis equipment takes a certain layer rule of the initial model as a target variable, takes a grouping dimension (namely credit history, age and the like) as an independent variable, constructs a multi-branch decision tree, divides a sample into subsets with significant differences on rule influence surfaces, and therefore achieves guest grouping, and in order to improve the interpretability of grouping results, the number of guests is generally controlled to be 2-4. And the risk tail end customer analysis equipment divides sample data corresponding to the branch nodes in the multi-branch decision tree into sample data subsets with different influence surfaces.
And respectively training an initial tail end layering model through each sample data subset to obtain a guest group analysis sub-model.
The risk tail client analysis device trains the initial tail layering model through each sample data subset respectively to obtain a guest group analysis sub-model, for example, the guest group analysis sub-models of all guest groups can be applied to the whole guest groups after being combined. Since the Lift value of each group has reference meaning to the whole when the influence surface in each group is similar to the whole influence surface, the group results should be matched under the similar influence surface to obtain the final guest-dividing group tail end layering model, in particular,
and combining layering rules in each guest group analysis sub-model to obtain a tail end layering model.
Specifically, the method comprises the following steps:
acquiring the outermost layering rules with the largest influence surface in each guest group analysis sub-model, and combining the outermost layering rules to serve as the outermost layer of the model;
searching a path with optimal performance, so that each guest group sub-model rule moves inwards by one layer;
and the method is repeated until the model is moved to the innermost layer of each guest group analysis sub-model, and the optimal path formed by iteration forms the hierarchy of the final model to obtain the tail end hierarchical model with guest group difference.
The risk tail end customer analysis equipment acquires the outermost layering rules with the largest influence surface in each guest group analysis sub-model, and combines the outermost layering rules to be used as the outermost layer of the model; moving each guest group analysis sub-model inwards by one layer, combining layering rules of the guest group analysis sub-models as model sub-outer layers, and sequentially analogizing by a risk tail end customer analysis device until the guest group analysis sub-models move to the innermost layer of each guest group analysis sub-model, exhausting layering rule combinations formed by all moving layers to obtain a plurality of groups of guest group combination paths; and the risk tail end customer analysis equipment selects a target group combination path with highest performance from the multiple groups of group combination paths, takes the target group combination path as a hierarchical rule of the final model, and obtains a tail end hierarchical model.
The risk tail end client analysis device in this embodiment analytically waits until moving to the innermost layer of each guest group analysis sub-model, iterating the formed optimal path to form the hierarchy of the final model, and obtaining the tail end hierarchical model with guest group differences, including the steps of:
acquiring a hierarchy rule corresponding to each hierarchy movement of each guest group analysis sub-model and an influence surface of the hierarchy rule;
Comparing the influence surfaces of the hierarchy movement corresponding hierarchy rules of the guest group analysis submodels;
if the difference between the influence surface of the current hierarchical rule of a target guest group analysis sub-model and the influence surface of other guest groups except the target guest group analysis sub-model is larger than a preset difference threshold, stopping moving the target guest group analysis sub-model in the iteration, and exhausting all paths of which the other guest group analysis sub-models except the target guest group analysis sub-model move inwards for one layer, so as to select a path with optimal performance;
if the target guest group analysis submodel with the abnormal influence surface of the hierarchical rule does not exist, all submodels are exhausted to move inwards a layer of paths, and the paths with optimal performance are selected from the paths.
Iterative optimization is carried out until all sub-models are moved to the innermost layer, and the optimal paths of each iteration are combined into a tail end hierarchical model with guest group differences.
Namely, in the embodiment, the risk tail end customer analysis device acquires a hierarchy rule corresponding to each hierarchy movement of each guest group analysis sub-model and an influence surface of the hierarchy rule; comparing the influence surfaces of the hierarchy movement corresponding hierarchy rules of the guest group analysis submodels; if there is an abnormal influence surface of the hierarchical rule (the abnormal influence surface refers to an influence surface which accords with a preset influence surface abnormal condition, for example, if the influence surface after a certain guest group analysis submodel moves inwards is smaller than the influence surface of other guest group analysis submodels which moves inwards twice, that is, the abnormal influence surface refers to the excessively large difference of the influence surface between one guest group analysis submodel and other remaining guest group analysis submodels, at the moment, the influence surface abnormality is judged), the moving of the target guest group analysis submodel is stopped, and the other guest group analysis submodels except for the target guest group analysis submodel are moved until the target guest group analysis submodel moves to the innermost layer of each guest group analysis submodel, and a plurality of guest group combination paths are obtained through the layering rule combination formed by all moving layers; if the target guest group analysis sub-model with abnormal influence surface of the hierarchical rule does not exist, moving to the innermost layer of each guest group analysis sub-model, and exhausting layering rule combinations formed by all moving hierarchies to obtain a plurality of groups of guest group combination paths. And the risk tail end customer analysis equipment selects a target group combination path with highest performance from a plurality of groups of group combination paths, and takes the target group combination path as a hierarchical rule of a final model to obtain a tail end hierarchical model.
For example, 1, combining the outermost layer results with the largest influence surface of each group of models as the outermost layer rule of the final model; 2. and moving each guest group model inwards by one layer, exhausting all possible movement sequences, obtaining a plurality of groups of guest group combination paths, and selecting the path with the highest overall performance from the paths as the hierarchical rule of the final model. If the influence surface of a certain guest group after inward movement is smaller than the influence surface of other guest groups after inward movement twice, stopping inward movement of the guest group, and preventing overlarge difference of the influence surfaces among the guest groups; 3. and repeatedly executing the previous step until each guest group moves inwards to the innermost layer, and obtaining the tail end layering model after grouping and combining.
Referring to fig. 2, fig. 2 is a flowchart of a first embodiment of a risk end customer analysis method according to the present invention, where the risk end customer analysis method includes:
step S10, when a risk analysis request is received, determining that the risk analysis request corresponds to a target client to be analyzed, and acquiring client information of the target client.
The risk tail end customer analysis equipment receives the risk analysis request, the triggering mode of the risk analysis request is not particularly limited, when the risk tail end customer analysis equipment receives the risk analysis request, the risk tail end customer analysis equipment determines that the risk analysis request corresponds to a target customer to be analyzed, and obtains customer information of the target customer, wherein the customer information comprises a customer name, identity information, shopping information, housing information, financial information, working information and the like.
The risk tail end customer analysis device in this embodiment preprocesses customer information, removes interference information in the customer information, and then further analyzes the remaining customer information, thereby implementing risk tail end customer analysis, specifically:
and step S20, inputting the customer information into different tail end analysis models corresponding to preset different types of risk features, and analyzing the customer information through the level rules of each level in the tail end analysis models to obtain tail end level scores of various risk features.
The risk tail end customer analysis equipment inputs the preprocessed customer information into tail end analysis models corresponding to preset different risk features, and processes the customer information through the tail end analysis models corresponding to the different risk features to obtain risk feature information, namely, the network of each level in the tail end analysis models can perform feature existence, so that the customer information analysis is more accurate. The risk tail end customer analysis device analyzes risk characteristic information through a tail end analysis model, analyzes the customer information through a hierarchy rule of each hierarchy in the tail end analysis model, and obtains a tail end hierarchy score of each hierarchy, and specifically comprises the following steps:
Inputting the client information into a preset tail end analysis model, and analyzing the client information through the level rule of each level in the tail end analysis model to obtain bad sample rate corresponding to each level;
and processing the bad sample rate corresponding to each level to obtain the tail level score of each level rule, wherein the tail level score is as follows:
the risk tail end customer analysis equipment inputs customer information into a preset tail end analysis model, analyzes the customer information through the level rule of each level in the tail end analysis model, and obtains bad sample rate corresponding to each level; and processing bad sample rates corresponding to each level to obtain the tail end level score of each level rule, so that risk assessment is carried out according to the guest group, and the risk assessment is more accurate.
And step S30, fusing the tail end level scores of various risk features to obtain a comprehensive risk score of the target client, and outputting prompt information according to the comprehensive risk score.
The risk tail end customer analysis device fuses the grade scores of all the tail end levels to obtain the comprehensive risk score of the target customer, and outputs prompt information according to the comprehensive risk score, and specifically comprises the following steps:
performing product operation on the scores of the tail end levels to obtain comprehensive scores of the target clients;
Outputting prompt information that the target client is not a risk user when the comprehensive score is greater than a preset threshold value;
and outputting prompt information that the target client is a risk user when the comprehensive score is smaller than or equal to a preset threshold value.
Performing product operation on the scores of all tail end levels to obtain the comprehensive score of the target client; outputting prompt information that the target client is not a risk user when the comprehensive score is greater than a preset threshold (the preset threshold is set according to a specific scene, for example, the preset threshold is set to be 0.6); and outputting prompt information that the target client is a risk user when the comprehensive score is smaller than or equal to a preset threshold value.
In the embodiment, the tail end analysis model is preset, and customer information analysis is performed through the level rules of each level in the tail end analysis model, so that targeted analysis is performed for different customer groups in the risk tail end customer analysis process, and the accuracy of risk tail end customer analysis is improved. By adopting the analysis mode of the risk tail end customers in financial institutions such as banking institutions, the accuracy of analysis of the risk tail end customers is improved, so that the financial institutions can more comprehensively analyze the risk tail end customers, decisions are made, and the bad account rate of the financial institutions such as banking institutions is reduced.
Further, based on the first embodiment of the risk end customer analysis method of the present invention, a second embodiment of the risk end customer analysis method of the present invention is provided.
The present embodiment is a step subsequent to step S30 in the first embodiment, and differs from the first embodiment in that:
and when the risk of the target client is classified as a non-tail end client, recommending products according to the client information of the target client.
When the risk of the target client is classified as a non-tail client, the risk tail client analysis equipment carries out product recommendation according to the client information of the target client, namely, the risk tail client analysis equipment carries out product recommendation according to the client information, so that the risk of financial type products can be reduced, and the user requirements can be met.
Further, based on the above embodiment of the risk end customer analysis method of the present invention, a fourth embodiment of the risk end customer analysis method of the present invention is provided.
The present embodiment is a step subsequent to step S10 in the first embodiment, and differs from the above-described embodiment in that:
extracting a client identifier in client information, inquiring a preset blacklist database, and determining whether the preset blacklist database contains the client identifier;
If the preset blacklist database contains the client identification, outputting prompt information that the target client is a risk tail client;
and if the preset blacklist database does not contain the client identifier, executing the step of inputting the client information into a preset tail end analysis model, and processing the client information through the tail end analysis model to obtain risk characteristic information.
The risk tail end customer analysis equipment extracts a customer identification (the customer identification refers to information for uniquely identifying the customer identity, such as a customer identity card number) in the customer information, a preset blacklist database is arranged in the risk tail end customer analysis equipment, information of a trusted user is stored in the preset blacklist database, and the risk tail end customer analysis equipment inquires the preset blacklist database to determine whether the customer identification exists in the preset blacklist database; if a client identifier exists in a preset blacklist database, the risk tail client analysis equipment outputs prompt information that the target client is the risk tail client; if the preset blacklist database does not contain the client identifier, step S20 in the first embodiment is executed to input the client information into different tail end analysis models corresponding to preset different types of risk features, and the client information is analyzed through the level rules of each level in the tail end analysis models to obtain the tail end level scores of various risk features.
In the embodiment, before the customer information analysis is performed by using the tail end analysis model, the risk tail end customer analysis device pre-uses the blacklist database to remove data with excessively high risk, so that the processing amount of the tail end analysis model is reduced, and meanwhile, the analysis efficiency can be accelerated.
The present invention also provides a risk tail end customer analysis device, including:
the request receiving module is used for determining that the risk analysis request corresponds to a target client to be analyzed when the risk analysis request is received, and acquiring client information of the target client;
the tail end hierarchy grading module is used for inputting the client information into different tail end analysis models corresponding to preset different types of risk features, analyzing the client information through hierarchy rules of all the levels in the tail end analysis models, and obtaining tail end hierarchy grading of all the risk features;
and the prompt output module is used for fusing the tail end level scores of various risk characteristics to obtain the comprehensive risk score of the target client, and outputting prompt information according to the comprehensive risk score.
In one embodiment, the risk tail end customer analysis device comprises:
The model construction module is used for acquiring a sample data set and constructing an initial tail end layering model;
the data dividing module is used for taking any layer of output of the initial tail end layering model as a target variable, taking a predefined grouping dimension as an independent variable, constructing a multi-fork decision tree, dividing sample data in a sample data set, and forming a sample data subset;
the model training module is used for respectively training the initial tail end layering model through each sample data subset to obtain a guest group analysis sub-model;
and the model construction module is used for combining the layering rules in each guest group analysis sub-model to obtain a tail end layering model.
In one embodiment, the model building module includes:
the first combination unit is used for acquiring the outermost layering rules with the largest influence surface in each passenger group analysis sub-model, and combining the outermost layering rules to be used as the outermost layer of the model;
the model moving unit is used for searching a path with optimal performance and enabling each guest group sub-model rule to move inwards by one layer;
and the mobile exhaustion unit is used for analogizing until the mobile exhaustion unit moves to the innermost layer of each guest group analysis sub-model, and the optimal path formed by iteration forms the level of the final model to obtain the tail end layered model with guest group difference.
In an embodiment, the mobile exhaustive unit is further configured to:
acquiring a hierarchy rule corresponding to each hierarchy movement of each guest group analysis sub-model and an influence surface of the hierarchy rule;
comparing the influence surfaces of the hierarchy movement corresponding hierarchy rules of the guest group analysis submodels;
if the difference between the influence surface of the current hierarchical rule of a target guest group analysis sub-model and the influence surface of other guest groups except the target guest group analysis sub-model is larger than a preset difference threshold, stopping moving the target guest group analysis sub-model in the iteration, and exhausting all paths of which the other guest group analysis sub-models except the target guest group analysis sub-model move inwards for one layer, so as to select a path with optimal performance;
if the target guest group analysis submodel with the abnormal influence surface of the hierarchical rule does not exist, all submodels are exhausted to move inwards a layer of paths, and the paths with optimal performance are selected from the paths.
Iterative optimization is carried out until all sub-models are moved to the innermost layer, and the optimal paths of each iteration are combined into a tail end hierarchical model with guest group differences.
In one embodiment, the model building module includes:
the variable conversion unit is used for constructing a risk feature set aiming at sample data in the sample data set, and converting each risk feature in the risk feature set into an initial variable value to form an initial variable set;
The rule determining unit is used for exhausting the quantiles of the preset duty ratio after each initial variable value in the initial variable set to be respectively used as initial rule thresholds, and determining an initial rule set according to each initial rule threshold;
the variable selection unit is used for selecting initial variables which are associated with risk analysis from the initial variable set according to the initial rule set as modulus-entering variable values, and forming the modulus-entering variable values selected into a modulus-entering variable set;
the rule exhaustion unit is used for exhausting the quantiles of the preset duty ratio after each modulus-entering variable value in the modulus-entering variable set to be respectively used as modulus-entering rule thresholds, and determining a modulus-entering rule set according to each modulus-entering rule threshold;
and the initial training module is used for training an initial model according to the sample data set and the modeling rule set to obtain an initial tail end layering model.
In one embodiment, the tail end hierarchy scoring module comprises:
the information analysis unit is used for inputting the client information into a tail end analysis model corresponding to preset different types of risk features, and analyzing the client information through the level rules of each level in the tail end analysis model to obtain bad sample rates corresponding to each level;
The processing calculation unit is used for processing the bad sample rate corresponding to each level to obtain tail end level scores of various risk features, wherein the tail end level scores are as follows:
in one embodiment, the hint output module includes:
the scoring operation unit is used for performing product operation on the scores of the tail end levels to obtain the comprehensive scores of the target clients;
the first prompting unit is used for outputting prompting information that the target client is not a risk user when the comprehensive score is larger than a preset threshold value;
and the second prompting unit is used for outputting prompting information that the target client is a risk user when the comprehensive score is smaller than or equal to a preset threshold value.
The method implemented when the risk end customer analysis device is executed may refer to various embodiments of the risk end customer analysis method of the present invention, which are not described herein.
In the embodiment of the invention, when a risk analysis request is received, a target client to be analyzed corresponding to the risk analysis request is determined, and client information of the target client is acquired; inputting the client information into different tail end analysis models corresponding to preset different types of risk features, and analyzing the client information through the level rules of each level in the tail end analysis models to obtain tail end level scores of various risk features; fusing the tail end level scores of various risk features to obtain a comprehensive risk score of the target client, and outputting prompt information according to the comprehensive risk score; in the embodiment of the invention, the tail end analysis model is preset, and the customer information analysis is performed through the level rules of each level in the tail end analysis model, so that the targeted analysis is performed for different customer groups in the risk tail end customer analysis process, and the accuracy of the risk tail end customer analysis is improved.
The invention also provides a computer storage medium.
The computer storage medium of the invention stores a computer program corresponding to the analysis of the risk end customer, and the computer program corresponding to the analysis of the risk end customer realizes the steps of the analysis method of the risk end customer when being executed by a processor.
The method implemented when the computer program corresponding to the analysis of the risk end client running on the processor is executed may refer to various embodiments of the method for analyzing the risk end client of the present invention, which are not described herein again.
It should be noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or system. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article, or system that comprises the element.
The foregoing embodiment numbers of the present invention are merely for the purpose of description, and do not represent the advantages or disadvantages of the embodiments.
From the above description of the embodiments, it will be clear to those skilled in the art that the above-described embodiment method may be implemented by means of software plus a necessary general hardware platform, but of course may also be implemented by means of hardware, but in many cases the former is a preferred embodiment. Based on such understanding, the technical solution of the present invention may be embodied essentially or in a part contributing to the prior art in the form of a software product stored in a storage medium (e.g. ROM/RAM, magnetic disk, optical disk) as described above, comprising instructions for causing a terminal device (which may be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) to perform the method according to the embodiments of the present invention.
The foregoing description is only of the preferred embodiments of the present invention, and is not intended to limit the scope of the invention, but rather is intended to cover any equivalents of the structures or equivalent processes disclosed herein or in the alternative, which may be employed directly or indirectly in other related arts.
Claims (9)
1. A method for analyzing a risk tail end customer, the method comprising the steps of:
when a risk analysis request is received, determining that the risk analysis request corresponds to a target client to be analyzed, and acquiring client information of the target client;
inputting the client information into different tail end analysis models corresponding to preset different types of risk features, and analyzing the client information through the level rules of each level in the tail end analysis models to obtain tail end level scores of various risk features;
fusing the tail end level scores of various risk features to obtain a comprehensive risk score of the target client, and outputting prompt information according to the comprehensive risk score;
before the step of determining that the risk analysis request corresponds to a target client to be analyzed and acquiring client information of the target client when the risk analysis request is received, the method comprises the following steps:
acquiring a sample data set and constructing an initial tail end layering model;
taking any layer output of the initial tail end hierarchical model as a target variable, taking a predefined grouping dimension as an independent variable, constructing a multi-fork decision tree, dividing sample data in a sample data set, and forming a sample data subset;
Respectively training the initial tail end layering model through each sample data subset to obtain a guest group analysis sub-model;
and combining layering rules in each guest group analysis sub-model to obtain a tail end layering model.
2. The risk tail client analysis method of claim 1, wherein the step of combining the layering rules in each of the guest group analysis sub-models to obtain a tail end layering model comprises:
acquiring the outermost layering rules with the largest influence surface in each guest group analysis sub-model, and combining the outermost layering rules to serve as the outermost layer of the model;
searching a path with optimal performance, so that each guest group analysis sub-model rule moves inwards by one layer;
and the method is repeated until the model is moved to the innermost layer of each guest group analysis sub-model, and the optimal path formed by iteration forms the hierarchy of the final model to obtain the tail end hierarchical model with guest group difference.
3. The method for analyzing the tail end risk client according to claim 2, wherein the step of repeating the steps until the tail end risk client moves to the innermost layer of each group analysis sub-model, and forming the hierarchy of the final model by iterating the formed optimal paths to obtain the tail end hierarchical model with group differences comprises the steps of:
Acquiring a hierarchy rule corresponding to each hierarchy movement of each guest group analysis sub-model and an influence surface of the hierarchy rule;
comparing the influence surfaces of the hierarchy movement corresponding hierarchy rules of the guest group analysis submodels;
if the difference between the influence surface of the current hierarchical rule of a target guest group analysis sub-model and the influence surface of other guest groups except the target guest group analysis sub-model is larger than a preset difference threshold, stopping moving the target guest group analysis sub-model in the iteration, and exhausting all paths of which the other guest group analysis sub-models except the target guest group analysis sub-model move inwards for one layer, so as to select a path with optimal performance;
if the target guest group analysis submodel with the abnormal influence surface of the hierarchical rule does not exist, exhausting paths of all submodels moving inwards by one layer, and selecting a path with optimal performance from the paths;
iterative optimization is carried out until all sub-models are moved to the innermost layer, and the optimal paths of each iteration are combined into a tail end hierarchical model with guest group differences.
4. The method of risk tail end customer analysis of claim 1, wherein the step of obtaining a sample dataset and constructing an initial tail end hierarchical model comprises:
Constructing a risk feature set aiming at sample data in a sample data set, and converting each risk feature in the risk feature set into an initial variable value to form an initial variable set;
the quantiles of the preset duty ratio after exhausting all initial variable values in the initial variable set are respectively used as initial rule thresholds, and an initial rule set is determined according to all the initial rule thresholds;
selecting initial variables which are associated with risk analysis from the initial variable set according to the initial rule set as modulus-entering variable values, and forming the modulus-entering variable values into a modulus-entering variable set;
the quantiles of the preset duty ratio after exhausting all the modulus variable values in the modulus variable set are respectively used as modulus rule threshold values, and a modulus rule set is determined according to all the modulus rule threshold values;
and training an initial model according to the sample data set and the modeling rule set to obtain an initial tail end layering model.
5. The risk tail end customer analysis method according to claim 1, wherein the step of inputting the customer information into different tail end analysis models corresponding to preset different types of risk features, analyzing the customer information through hierarchical rules of each hierarchy in the tail end analysis models, and obtaining tail end hierarchy scores of the various risk features comprises the steps of:
Inputting the client information into a tail end analysis model corresponding to preset different types of risk features, and analyzing the client information through a hierarchy rule of each hierarchy in the tail end analysis model to obtain bad sample rates corresponding to each hierarchy;
and processing bad sample rates corresponding to the levels to obtain tail end level scores of various risk features, wherein the tail end level scores are as follows:
tail end。
6. The risk tail end customer analysis method according to any one of claims 1 to 5, wherein the step of fusing tail end level scores of various risk features to obtain a comprehensive risk score of the target customer, and outputting prompt information according to the comprehensive risk score includes:
performing product operation on the scores of the tail end levels to obtain comprehensive scores of the target clients;
outputting prompt information that the target client is not a risk user when the comprehensive score is greater than a preset threshold value;
and outputting prompt information that the target client is a risk user when the comprehensive score is smaller than or equal to a preset threshold value.
7. A risk tail end customer analysis device, the risk tail end customer analysis device comprising:
The request receiving module is used for determining that the risk analysis request corresponds to a target client to be analyzed when the risk analysis request is received, and acquiring client information of the target client;
the tail end hierarchy grading module is used for inputting the client information into different tail end analysis models corresponding to preset different types of risk features, analyzing the client information through hierarchy rules of all the levels in the tail end analysis models, and obtaining tail end hierarchy grading of all the risk features;
the prompt output module is used for fusing the tail end level scores of various risk characteristics to obtain a comprehensive risk score of the target client, and outputting prompt information according to the comprehensive risk score;
the risk tail end customer analysis device further comprises:
the model construction module is used for acquiring a sample data set and constructing an initial tail end layering model;
the data dividing module is used for taking any layer of output of the initial tail end layering model as a target variable, taking a predefined grouping dimension as an independent variable, constructing a multi-fork decision tree, dividing sample data in a sample data set, and forming a sample data subset;
the model training module is used for respectively training the initial tail end layering model through each sample data subset to obtain a guest group analysis sub-model;
And the model construction module is used for combining the layering rules in each guest group analysis sub-model to obtain a tail end layering model.
8. A risk tail end customer analysis device, the risk tail end customer analysis device comprising: memory, a processor and a risk end customer analysis corresponding computer program stored on the memory and executable on the processor, which risk end customer analysis corresponding computer program, when executed by the processor, implements the steps of the risk end customer analysis method according to any of claims 1 to 6.
9. A computer storage medium, wherein a computer program corresponding to the analysis of a risk end customer is stored on the computer storage medium, and the computer program corresponding to the analysis of a risk end customer, when executed by a processor, implements the steps of the method for analyzing a risk end customer according to any one of claims 1 to 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010549845.1A CN111695824B (en) | 2020-06-16 | 2020-06-16 | Method, device, equipment and computer storage medium for analyzing risk tail end customer |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010549845.1A CN111695824B (en) | 2020-06-16 | 2020-06-16 | Method, device, equipment and computer storage medium for analyzing risk tail end customer |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111695824A CN111695824A (en) | 2020-09-22 |
| CN111695824B true CN111695824B (en) | 2024-03-29 |
Family
ID=72481577
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010549845.1A Active CN111695824B (en) | 2020-06-16 | 2020-06-16 | Method, device, equipment and computer storage medium for analyzing risk tail end customer |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111695824B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112631686A (en) * | 2020-12-31 | 2021-04-09 | 招联消费金融有限公司 | Data processing method, data processing device, computer equipment and storage medium |
| CN113254510B (en) * | 2021-07-06 | 2021-09-28 | 平安科技（深圳）有限公司 | Method, device, equipment and storage medium for identifying business risk customer group |
| CN115221948A (en) * | 2022-06-24 | 2022-10-21 | 北京百度网讯科技有限公司 | Feature combination determination method, device, equipment, storage medium and program product |
| CN116415237B (en) * | 2023-03-03 | 2024-03-19 | 港珠澳大桥管理局 | Risk device identification method, apparatus, computer device and storage medium |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108876600A (en) * | 2018-08-20 | 2018-11-23 | 平安科技（深圳）有限公司 | Warning information method for pushing, device, computer equipment and medium |
| CN109165840A (en) * | 2018-08-20 | 2019-01-08 | 平安科技（深圳）有限公司 | Risk profile processing method, device, computer equipment and medium |
| CN109272396A (en) * | 2018-08-20 | 2019-01-25 | 平安科技（深圳）有限公司 | Customer risk method for early warning, device, computer equipment and medium |
| CN109949154A (en) * | 2018-12-17 | 2019-06-28 | 深圳平安综合金融服务有限公司 | Customer information classification method, device, computer equipment and storage medium |
| WO2019184119A1 (en) * | 2018-03-26 | 2019-10-03 | 平安科技（深圳）有限公司 | Risk model training method and apparatus, risk identification method and apparatus, device, and medium |
| CN110458412A (en) * | 2019-07-16 | 2019-11-15 | 阿里巴巴集团控股有限公司 | The generation method and device of risk monitoring and control data |
-
2020
- 2020-06-16 CN CN202010549845.1A patent/CN111695824B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2019184119A1 (en) * | 2018-03-26 | 2019-10-03 | 平安科技（深圳）有限公司 | Risk model training method and apparatus, risk identification method and apparatus, device, and medium |
| CN108876600A (en) * | 2018-08-20 | 2018-11-23 | 平安科技（深圳）有限公司 | Warning information method for pushing, device, computer equipment and medium |
| CN109165840A (en) * | 2018-08-20 | 2019-01-08 | 平安科技（深圳）有限公司 | Risk profile processing method, device, computer equipment and medium |
| CN109272396A (en) * | 2018-08-20 | 2019-01-25 | 平安科技（深圳）有限公司 | Customer risk method for early warning, device, computer equipment and medium |
| WO2020037942A1 (en) * | 2018-08-20 | 2020-02-27 | 平安科技（深圳）有限公司 | Risk prediction processing method and apparatus, computer device and medium |
| CN109949154A (en) * | 2018-12-17 | 2019-06-28 | 深圳平安综合金融服务有限公司 | Customer information classification method, device, computer equipment and storage medium |
| CN110458412A (en) * | 2019-07-16 | 2019-11-15 | 阿里巴巴集团控股有限公司 | The generation method and device of risk monitoring and control data |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111695824A (en) | 2020-09-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111695824B (en) | Method, device, equipment and computer storage medium for analyzing risk tail end customer | |
| CN110414555B (en) | Method and device for detecting abnormal sample | |
| Idris et al. | Intelligent churn prediction for telecom using GP-AdaBoost learning and PSO undersampling | |
| WO2023279696A1 (en) | Service risk customer group identification method, apparatus and device, and storage medium | |
| CN110310114B (en) | Object classification method, device, server and storage medium | |
| CN113011889B (en) | Account anomaly identification method, system, device, equipment and medium | |
| KR20190026529A (en) | SYSTEM AND METHOD FOR text classification | |
| Nivaashini et al. | Computational intelligence techniques for automatic detection of Wi-Fi attacks in wireless IoT networks | |
| CN115577357A (en) | Android malicious software detection method based on stacking integration technology | |
| Ali et al. | Improved differential evolution algorithm with decentralisation of population | |
| US20040002879A1 (en) | System and method for feature selection in decision trees | |
| JP4802176B2 (en) | Pattern recognition apparatus, pattern recognition program, and pattern recognition method | |
| CN109977977B (en) | Method for identifying potential user and corresponding device | |
| CN111753998A (en) | Model training method, device and equipment for multiple data sources and storage medium | |
| CN116304518A (en) | Heterogeneous graph convolution neural network model construction method and system for information recommendation | |
| CN111401675A (en) | Similarity-based risk identification method, device, equipment and storage medium | |
| CN111652525B (en) | Method, device, equipment and computer storage medium for analyzing risk tail end customer | |
| CN114997378A (en) | Inductive graph neural network pruning method, system, device and storage medium | |
| JP2015099616A (en) | Multiple class identifier generation device, data identification device, multiple class identifier generation method, data identification method, and program | |
| CN110990353B (en) | Log extraction method, log extraction device and storage medium | |
| CN114048796A (en) | Improved hard disk failure prediction method and device | |
| CN112580268A (en) | Method and device for selecting machine learning model based on business processing | |
| CN112529319A (en) | Grading method and device based on multi-dimensional features, computer equipment and storage medium | |
| Dumitrache et al. | Churn prepaid customers classified by HyperOpt techniques | |
| Panduro-Ramirez et al. | Implementation of machine learning techniques for predicting credit card customer action |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |