CN110796554B - User complaint early warning method and device, computer equipment and storage medium - Google Patents
User complaint early warning method and device, computer equipment and storage medium Download PDFInfo
- Publication number
- CN110796554B CN110796554B CN201910841818.9A CN201910841818A CN110796554B CN 110796554 B CN110796554 B CN 110796554B CN 201910841818 A CN201910841818 A CN 201910841818A CN 110796554 B CN110796554 B CN 110796554B
- Authority
- CN
- China
- Prior art keywords
- complaint
- training
- user
- probability
- target user
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; 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
- G06Q40/08—Insurance
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; 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
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/10—Services
- G06Q50/18—Legal services
- G06Q50/182—Alternative dispute resolution
Landscapes
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Tourism & Hospitality (AREA)
- Economics (AREA)
- Marketing (AREA)
- Strategic Management (AREA)
- Technology Law (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- Finance (AREA)
- Accounting & Taxation (AREA)
- Development Economics (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Human Resources & Organizations (AREA)
- Primary Health Care (AREA)
- Financial Or Insurance-Related Operations Such As Payment And Settlement (AREA)
Abstract
The invention provides a user complaint early warning method, a device, computer equipment and a storage medium, wherein the method comprises the following steps: acquiring unique identification information of a target user; acquiring user data corresponding to a target user from a preset user database according to the unique identification information; preprocessing the user data; predicting the final complaint probability of the current complaint of the target user according to the preprocessed user data; judging whether the final complaint probability is larger than a preset threshold value, and if so, sending an alarm. The invention can realize automatic and efficient complaint early warning.
Description
Technical Field
The present invention relates to the field of information technologies, and in particular, to a method and apparatus for early warning of user complaints, a computer device, and a storage medium.
Background
With the development of economies, the number of motor vehicles is increasing. Currently, motor vehicle insurance has become the biggest risk in China's property insurance business. Motor vehicle insurance has covered a large portion of the dangerous accidents of automobiles, and the chinese department of transportation has forced vehicle purchasers to purchase motor vehicle insurance to ensure that in the event of an automobile accident, the legal rights and interests of victims are ensured. With the rapid development of car insurance services, the number of complaints related to car insurance is correspondingly increased. In the current fierce competitive environment, not only is the generated user complaints required to be processed efficiently, but also the complaint tendency of the user is required to be evaluated scientifically and the complaint behaviors of the user are required to be pre-warned. At present, the complaint early warning methods in the market are all based on manual experience rules, and the efficiency and the accuracy are low.
Disclosure of Invention
Aiming at the defects of the prior art, the invention aims to provide a user complaint early warning method, a device, computer equipment and a storage medium so as to realize automatic and efficient complaint early warning.
In order to achieve the above purpose, the present invention provides a user complaint early warning method, comprising the following steps:
Acquiring unique identification information of a target user;
Acquiring user data corresponding to a target user from a preset user database according to the unique identification information;
preprocessing the user data;
predicting the final complaint probability of the current complaint of the target user according to the preprocessed user data;
Judging whether the final complaint probability is larger than a preset threshold value, and if so, sending an alarm notification.
In one embodiment of the invention, the preprocessing comprises: data cleaning, feature extraction and normalization.
In one embodiment of the present invention, the user data includes a personal characteristic, a vehicle characteristic, and a full-stage vehicle insurance service characteristic corresponding to the target user in a predetermined period of time, and further includes a time characteristic, a geographic characteristic, a personal characteristic, a vehicle characteristic, and a current-stage vehicle insurance service characteristic corresponding to the target user.
In one embodiment of the present invention, the step of predicting the final complaint probability of the current complaint of the target user according to the preprocessed target user feature includes:
Inputting personal characteristics, vehicle characteristics and full-stage vehicle insurance service characteristics corresponding to a target user in a preset time period into a basic complaint probability model corresponding to all service stages obtained through pre-training, and processing the basic complaint probability model to obtain basic complaint probability of the target user in each service stage of vehicle insurance;
inputting the time feature, the geographic feature, the personal feature, the vehicle feature and the vehicle risk service feature corresponding to the current stage of the target user into an initial complaint probability model corresponding to the current stage of training in advance for processing to obtain the initial complaint probability of the current complaint of the target user;
And carrying out weighted summation operation on the basic complaint probability and the initial complaint probability to obtain the final complaint probability of the current complaint of the target user.
In one embodiment of the present invention, the training steps of the basic complaint probability model are as follows:
acquiring a first training sample set, wherein the first training sample set comprises a plurality of first training samples and complaint information corresponding to each first training sample, and each first training sample respectively comprises personal characteristics, vehicle characteristics and full-stage vehicle insurance business characteristics of each training user in past years;
and training according to the first training sample set to obtain the basic complaint probability model.
In one embodiment of the present invention, the basic complaint probability model is a hybrid logistic regression model, and the expression of the hybrid logistic regression model is as shown in formula (1):
Wherein σ (·) represents a classification function, η (·) represents a fitting function, x represents a sample feature in the first training sample, { u1, …, um } is a weight of x in the classification function, { ω1, …, ωm } is a weight of x in the fitting function, p (y= 1|x) represents a probability of occurrence of an event y under a given sample feature x, y represents an event of complaint by the training user, g (·) is used to ensure that the model conforms to the definition of the probability function, m represents a hyper-parameter fragment, and m=3.
In one embodiment of the present invention, the training steps of the initial complaint probability model corresponding to the current stage are as follows:
acquiring a second training sample set, wherein the second training sample set comprises a plurality of second training samples and complaint information corresponding to each second training sample, and each second training sample respectively comprises complaint time characteristics, complaint geographic characteristics, personal characteristics, vehicle characteristics and current-stage vehicle insurance service characteristics of each training user in the past year;
And training the initial complaint probability model corresponding to the current stage according to the second training sample set.
In order to achieve the above object, the present invention further provides a user complaint early warning device, which includes:
the identification information acquisition module is used for acquiring the unique identification information of the target user;
The user data acquisition module is used for acquiring user data corresponding to a target user from a preset user database according to the unique identification information;
The preprocessing module is used for preprocessing the user data;
The complaint probability prediction module predicts the final complaint probability of the current complaint of the target user according to the preprocessed user data;
And the early warning module is used for judging whether the final complaint probability is larger than a preset threshold value, and if so, sending an alarm notification.
In one embodiment of the invention, the preprocessing comprises: data cleaning, feature extraction and normalization.
In one embodiment of the present invention, the user data includes a personal characteristic, a vehicle characteristic, and a full-stage vehicle insurance service characteristic corresponding to the target user in a predetermined period of time, and further includes a time characteristic, a geographic characteristic, a personal characteristic, a vehicle characteristic, and a current-stage vehicle insurance service characteristic corresponding to the target user.
In one embodiment of the present invention, the complaint probability prediction module includes:
The basic complaint probability prediction unit is used for inputting the personal characteristics, the vehicle characteristics and the full-stage vehicle insurance service characteristics corresponding to the target user in a preset time period into basic complaint probability models corresponding to all service stages obtained through pre-training, and processing the basic complaint probability models to obtain basic complaint probability of the target user in each service stage of the vehicle insurance;
The initial complaint probability prediction unit is used for inputting the time feature, the geographic feature, the personal feature, the vehicle feature and the vehicle risk service feature corresponding to the current stage of the target user into the initial complaint probability model corresponding to the current stage obtained through pre-training for processing, and obtaining the initial complaint probability of the current complaint of the target user;
and the weighting unit is used for carrying out weighted summation operation on the basic complaint probability and the initial complaint probability to obtain the final complaint probability of the current complaint of the target user.
In one embodiment of the present invention, the user complaint warning device further includes: the basic complaint probability model training module is used for:
acquiring a first training sample set, wherein the first training sample set comprises a plurality of first training samples and complaint information corresponding to each first training sample, and each first training sample respectively comprises personal characteristics, vehicle characteristics and full-stage vehicle insurance business characteristics of each training user in past years;
and training according to the first training sample set to obtain the basic complaint probability model.
In one embodiment of the present invention, the basic complaint probability model is a hybrid logistic regression model, and the expression of the hybrid logistic regression model is as shown in formula (1):
Wherein σ (·) represents a classification function, η (·) represents a fitting function, x represents a sample feature in the first training sample, { u 1,…,um } is a weight of x in the classification function, { w 1,…,wm } is a weight of x in the fitting function, p (y= 1|x) represents a probability of occurrence of an event y given the sample feature x, y represents an event of complaint by the training user, g (·) is used to ensure that the model conforms to the definition of the probability function, m represents a hyper-parameter tile, and m=3 is taken.
In one embodiment of the present invention, the user complaint warning device further includes: the initial complaint probability model training module is used for:
acquiring a second training sample set, wherein the second training sample set comprises a plurality of second training samples and complaint information corresponding to each second training sample, and each second training sample respectively comprises complaint time characteristics, complaint geographic characteristics, personal characteristics, vehicle characteristics and current-stage vehicle insurance service characteristics of each training user in the past year;
And training the initial complaint probability model corresponding to the current stage according to the second training sample set.
To achieve the above object, the present invention also provides a computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, which processor implements the steps of the aforementioned method when executing the computer program.
In order to achieve the above object, the present invention also provides a computer-readable storage medium having stored thereon a computer program which, when executed by a processor, implements the steps of the aforementioned method.
By adopting the technical scheme, the invention has the following beneficial effects:
According to the invention, the final complaint probability of the current complaint of the target user is automatically predicted, and the alarm notification is automatically sent when the final complaint probability is larger than the preset threshold value, so that automatic and efficient complaint early warning is realized, differentiated services are conveniently carried out on the user by service personnel according to the alarm notification, timely security care is actively carried out on the user in advance, the complaint treatment is changed from original 'after-compensation' to 'before-control', and the vehicle insurance company is changed from passive to active when facing the complaint problem, so that the complaint rate of the user is greatly reduced, and the competitive power of the insurance company is enhanced.
Drawings
FIG. 1 is a flow chart of one embodiment of a method of the present invention for early warning of customer complaints;
FIG. 2 is a schematic diagram of one embodiment of a GBDT model employed in the present invention;
FIG. 3 is a method of predicting a final complaint probability of a user's current complaint in accordance with the present invention;
FIG. 4 is a block diagram illustrating one embodiment of a customer complaint warning device of the present invention;
FIG. 5 is a hardware architecture diagram of one embodiment of a computer device of the present invention.
Detailed Description
The present invention will be described in further detail with reference to the drawings and examples, in order to make the objects, technical solutions and advantages of the present invention more apparent. 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. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Example 1
Referring to fig. 1, the embodiment provides a user complaint early warning method, which can be used in the fields of vehicle insurance and the like, as shown in fig. 1, and includes the following steps:
S1, obtaining unique identification information of a target user. In the present embodiment, the unique identification information can be acquired, for example, by the following means: when a user incoming call is received, the incoming call mobile phone number is automatically read, and unique identification information such as a certificate number corresponding to the incoming call mobile phone number is matched from a user database.
S2, acquiring user data corresponding to the target user from a preset user database according to the unique identification information of the target user. In this embodiment, the user data includes a personal feature, a vehicle feature, and a full-stage vehicle risk service feature corresponding to the target user in a predetermined period (the last year), and further includes a time feature, a geographic feature, a personal feature, a vehicle feature, and a current-stage vehicle risk service feature corresponding to the target user. Specifically, the personal characteristics include name, document number, cell phone number, age, gender, and driving age; the vehicle characteristics comprise license plate numbers, vehicle types and vehicle ages of the insuring vehicles; the full-stage vehicle insurance business features comprise business features corresponding to each business stage of the vehicle insurance; the traffic characteristics of the current stage of traffic insurance comprise the traffic characteristics corresponding to the traffic stage of the current position of the vehicle insurance; the time characteristics include: whether to night or holiday; the buried feature includes a GPS location address of the user.
In general, a vehicle insurance may include the following four business phases: an investigation loss assessment stage, a product distribution stage, a marketing exhibition stage and a bill pay-out stage. Wherein, the business characteristics corresponding to the investigation and loss assessment stage comprise: corresponding to the time of the vehicle insurance company, the time of setting up a case, etc.; the business characteristics corresponding to the product distribution stage comprise: the product type, the product price, the delivery time and the like of the products delivered by the corresponding car insurance company; the business characteristics corresponding to the marketing exhibition stage comprise: the call times and call time of the corresponding user, the value added service type provided by the vehicle insurance company, the call times and call time of the corresponding user to the vehicle insurance company and the like are generated; the business characteristics corresponding to the order-issuing payment stage comprise: corresponding to the payment mode and payment amount of the user, corresponding to the order-giving duration and order-giving accuracy of the car insurance company, and the like.
S3, preprocessing the user data. In the present embodiment, the preprocessing includes a data cleaning process, a digitizing process, and a normalizing process.
The cleaning process means deleting data with low association degree with the probability of complaints in the user data, such as name, license number, mobile phone number, payment mode and the like.
The feature extraction processing refers to representing non-numeric features among the user features as numeric features, for example, for gender features, if gender is male, it is denoted as 1, otherwise, it is denoted as 0; for the time feature, if the current time of the incoming call of the user is night, the current time is expressed as 1, otherwise, the current time is expressed as 0; for the bill outlet accuracy characteristics, if the bill outlet accuracy characteristics are accurately expressed as 1, otherwise, the bill outlet accuracy characteristics are expressed as 0, and for the vehicle type, the underground characteristics, the product type, the value added service type and the like, clustering can be respectively carried out according to a preset rule, and different categories are expressed by different numbers.
Normalization refers to the adjustment of all features that have undergone a cleaning process and a digitizing process to the same dimension. Normalization is performed because the user features are of a large order of magnitude, e.g., the price of the product may be thousands, tens of thousands, and the driving age may be only decades at most, and performing a large number of calculations is time consuming and the results of the calculations may be abnormally large when training the model later. In addition, the weight distribution is uneven, the weight obtained by a large number is likely to be larger, however, the large number is probably not the most critical factor for determining the prediction result, and the result becomes the most important factor because of large value, so that the prediction is inaccurate, all the characteristics are changed to be in the range of 0 to 1 by adjusting through normalization processing, and the dimension and the adjustment direction are unified. The normalization processing process for the user characteristics specifically comprises the following steps: firstly, acquiring the mean value and standard deviation of all user characteristics; and then, subtracting the mean value from each user characteristic value, and dividing the mean value by the standard deviation of the user characteristic value to obtain a normalization result.
S4, predicting the final complaint probability of the current complaint of the target user according to the preprocessed user data. Specifically, the basic complaint probability model corresponding to all phases and the initial complaint probability model corresponding to a single phase obtained through pre-training are predicted.
In this embodiment, the training process of the basic complaint probability model corresponding to all the stages is as follows:
firstly, a first training sample set is obtained, wherein the first training sample set comprises a plurality of first training samples and complaint information corresponding to each first training sample. Each first training sample includes three categories of personal characteristics, vehicle characteristics and full-stage vehicle insurance service characteristics (namely, investigation and damage assessment, product distribution, marketing canvasing and vehicle insurance service characteristics in a single payment stage) of each training user in the past year.
And then, training according to the first training sample set to obtain a basic complaint probability model corresponding to all the business stages of the vehicle insurance of each user, wherein the basic complaint probability model can be used for calculating the basic probability of the complaint event of the user in all the business stages of the vehicle insurance.
In the present embodiment, the aforementioned basic complaint probability model is preferably an MLR (Mixed Logistic Regression, hybrid logistic regression) model. The MLR model can be regarded as a natural generalization of the LR (logistic regression) model, which uses a divide-and-conquer approach to fit a nonlinear classification surface of a high-dimensional space with a piecewise linear model, and the expression is as follows:
Wherein σ (·) represents a classification function, η (·) represents a fitting function, x represents a sample feature in the first training sample, { u 1,…,um } is a weight of x in the classification function, { w 1,…,wm } is a weight of x in the fitting function, p (y= 1|x) represents a probability of occurrence of an event y given the sample feature x, y represents an event of complaint by the training user, g (·) is used to ensure that the model conforms to the definition of the probability function, and m represents a hyper-parametric slice.
In this embodiment, taking σ (·) as the softmax function, η (·) as the sigmoid function, g (x) =x, the MLR model can be converted into the following form:
Further, since in the present application, the sample features of the first training sample include three parts of the personal feature, the vehicle feature, and the full-stage vehicle risk service feature, m=3 is taken.
And (3) when the basic complaint probability model is trained, iterating and updating the parameter { u 1,…,um}、{w1,…,wm } in the (2) by using the first training sample set until the training of the basic complaint probability model is finished when the loss function of the MLR model meets the preset condition.
In this embodiment, the training process of the initial complaint probability model corresponding to a single stage (i.e. the investigation and damage assessment, product distribution, marketing canvasing or single payment stage) is as follows:
And acquiring a second training sample set, and training according to the second training sample set to obtain an initial complaint probability model corresponding to each user in the current business stage of the vehicle insurance (namely, the exploration and damage assessment, product distribution, marketing canvasing or single payment stage). The second training sample set includes a plurality of second training samples and complaint information corresponding to each second training sample, where each second training sample includes complaint time characteristics, complaint geographic characteristics, personal characteristics, vehicle characteristics and vehicle insurance service characteristics (i.e. investigation damage, product distribution, marketing canvasing or vehicle insurance service characteristics corresponding to a single payment stage) of each training user in the past year.
In this embodiment, the initial complaint probability model corresponding to a single stage is preferably in the form of GBDT (gradient-lifted decision tree) model combined with the LR model. During training, firstly, original sample characteristics in a second training sample set, namely time characteristics, geographic characteristics, personal characteristics, vehicle characteristics and corresponding stage car insurance service characteristics of a training user in the past year, are input into a GBTD model for processing, and then the processing result of the GBTD model is input into an LR model for training. Specifically, in an embodiment of the present application, the GBDT model may be trained according to the original sample features in the second training sample set to construct a GBDT model with N trees, where N is a positive integer, and then, an association relationship between the original sample features may be established according to the N trees in the GBDT model, and feature combination may be performed on the original sample features according to the association relationship to generate the cross-combination feature.
In one embodiment of the present application, a specific implementation process for generating the cross-combination feature may include: and sequentially passing the original sample features in the second training sample through N trees in the GBDT model until each original sample feature is distributed to a leaf node of a certain tree, and combining the original sample features corresponding to paths passing from the root node to the leaf node of each tree for each tree in the GBDT model to generate a cross combination feature.
In the present application GBDT is an iterative decision tree algorithm consisting of a number of decision trees, each decision tree containing a number of nodes, each node corresponding to a feature, the conclusions of all trees being accumulated as a composite result. Taking GBDT models in fig. 2 as an example, 2 decision numbers are shared in the graphs, after two trees are traversed, the features in a sample x fall onto leaf nodes of the two trees respectively, each leaf node corresponds to one-dimensional features, and then the cross combination features corresponding to the samples are obtained by traversing the trees. For example: in the two trees, the nodes A, B, C respectively correspond to the characteristics of whether the gender of the user is male, whether the age is greater than 30 years and whether the driving age is less than 2 years, the left tree is provided with three leaf nodes, the right tree is provided with two leaf nodes, and the comprehensive cross combination characteristic is a five-dimensional vector. For example, for input x, it is assumed that he falls on the first node of the left tree, then code [1, 0], and falls on the second node of the right tree, then code [0,1], so the overall code is [1,0,0,0,1], and the overall code is the cross-combination feature of x.
After GBDT model generates cross-combined features, the cross-combined features are then input to LR model for processing.
Specifically, the expression of the LR model is as follows:
where P (y= 1|x; θ) represents the probability of occurrence of event y given sample feature x, θ represents the weight of feature x, where x is the cross-combined feature output by the GBDT model, and y represents the event that trains the user to complain.
When the LR model is trained, the weight in the LR model is iteratively updated by utilizing the cross combination characteristic output by the GBDT model until the loss function of the LR model meets the preset condition, and the initial probability complaint model corresponding to a single stage is obtained.
Returning to step S3, predicting the final complaint probability of the current complaint of the target user specifically by:
S41, substituting the personal characteristics, the vehicle characteristics and the full-stage vehicle insurance business characteristics corresponding to the target user in a preset time period (preferably the last year) into the basic complaint probability model corresponding to all stages obtained through training in advance for processing, and obtaining the basic complaint probability Q1 of complaint of the target user in each business stage.
S42, firstly, obtaining the time feature, the geographic feature, the personal feature, the vehicle feature and the vehicle insurance service feature of the current stage corresponding to the target user. For example, assuming that the target user currently experiences a vehicle insurance investigation and damage assessment stage, determining the current service stage as the investigation and damage assessment stage, and taking the vehicle insurance service characteristics of the current stage as the service characteristics corresponding to the investigation and damage assessment stage currently experienced by the target user, namely the time of emergence, the time of case setting and the like of the corresponding vehicle insurance company. And substituting the time feature, the geographic feature, the personal feature, the vehicle feature and the vehicle risk service feature of the current stage corresponding to the target user into an initial complaint probability model corresponding to the current service stage obtained through pre-training to process, so as to obtain the initial complaint probability Q2 of complaint of the target user in the current service stage.
For example, if the current service stage is determined to be a exploration and damage assessment stage, substituting the time feature, the geographic feature, the personal feature, the vehicle feature and the vehicle risk service feature corresponding to the target user currently into an initial complaint probability model corresponding to the exploration and damage assessment stage, which is obtained through pre-training; similarly, when the current service stage is determined to be the product delivery stage, substituting the time feature, the geographic feature, the personal feature, the vehicle feature and the vehicle risk service feature of the product delivery stage corresponding to the current target user into the initial complaint probability model corresponding to the product delivery stage obtained through pre-training; when the determined current business stage is a marketing exhibition stage, substituting the current corresponding time feature, geographic feature, personal feature, vehicle feature and traffic risk business feature of the target user in the marketing exhibition stage into an initial complaint probability model corresponding to a marketing canvasing stage which is obtained through training in advance; when the current service stage is determined to be the order-issuing payment stage, substituting the time feature, the geographic feature, the personal feature, the vehicle feature and the order-issuing payment stage vehicle risk service feature corresponding to the current target user into the initial complaint probability model corresponding to the order-issuing payment stage, which is obtained through training in advance.
S43, in order to obtain an accurate prediction result, in the embodiment, the initial complaint probability Q2 of complaints of the target user in the current business stage and the basic complaint probability Q1 of complaints of the target user in each business stage are subjected to weighted summation operation, and the weighted summation result is used as the final complaint probability of the current complaints of the target user. For example, assuming that the predicted basic complaint probability of the target user is 40%, the weight occupied by the basic complaint probability is preset to 15%, the initial complaint probability of the current complaint of the target user is 60%, and the weight occupied by the initial complaint probability is preset to 85%, the final complaint probability of the complaint of the target user in the current business stage of the vehicle insurance is 60% ×20% +80% ×80% =76%.
S5, judging whether the predicted final complaint probability is larger than a preset threshold value. The preset threshold value can be adjusted according to specific needs
And S6, if the predicted final complaint probability is greater than a preset threshold, sending an alarm notification. For example, assuming that the final complaint probability of the current complaint of the target user is 76% and the preset threshold is 50%, a complaint alarm message is sent to remind the staff of the car insurance company.
In summary, when a user calls, the invention can predict the probability of user complaints and send out early warning when the probability of complaints is larger, so that staff can actively pacify and care the user in time in advance, and the complaint treatment is changed from the original 'post-compensation' to 'pre-control', so that an insurance company is changed from passive to active when facing the complaint problem, thereby greatly reducing the complaint rate of the user and enhancing the competitiveness of the insurance company.
It should be noted that, for simplicity of description, the present embodiment is shown as a series of acts, but it should be understood by those skilled in the art that the present invention is not limited by the order of acts, as some steps may be performed in other order or simultaneously in accordance with the present invention. Further, those skilled in the art will appreciate that the embodiments described in the specification are presently preferred embodiments, and that the acts are not necessarily required for the present invention.
Example two
As shown in fig. 4, the present embodiment provides a user complaint early warning device 10, which includes the following modules:
an identification information acquisition module 11, configured to acquire unique identification information of a target user;
The user data obtaining module 12 is configured to obtain user data corresponding to a target user from a preset user database according to the unique identification information;
A preprocessing module 13, configured to preprocess the user data;
A complaint probability prediction module 14 for predicting a final complaint probability of the current complaint of the target user based on the preprocessed user data;
And the early warning module 25 is used for judging whether the final complaint probability is greater than a preset threshold value, and if so, sending an alarm notification.
In this embodiment, the preprocessing 13 includes: data cleaning, feature extraction and normalization.
In this embodiment, the user data includes a personal feature, a vehicle feature, and a full-stage vehicle risk service feature corresponding to the target user in a predetermined period of time, and further includes a time feature, a geographic feature, a personal feature, a vehicle feature, and a current-stage vehicle risk service feature corresponding to the target user.
In this embodiment, the complaint probability prediction module 14 includes:
The basic complaint probability prediction unit is used for inputting the personal characteristics, the vehicle characteristics and the full-stage vehicle insurance service characteristics corresponding to the target user in a preset time period into basic complaint probability models corresponding to all service stages obtained through pre-training, and processing the basic complaint probability models to obtain basic complaint probability of the target user in each service stage of the vehicle insurance;
The initial complaint probability prediction unit is used for inputting the time feature, the geographic feature, the personal feature, the vehicle feature and the vehicle risk service feature corresponding to the current stage of the target user into the initial complaint probability model corresponding to the current stage obtained through pre-training for processing, and obtaining the initial complaint probability of the current complaint of the target user;
and the weighting unit is used for carrying out weighted summation operation on the basic complaint probability and the initial complaint probability to obtain the final complaint probability of the current complaint of the target user.
In this embodiment, the user complaint early warning device further includes: the basic complaint probability model training module is used for:
acquiring a first training sample set, wherein the first training sample set comprises a plurality of first training samples and complaint information corresponding to each first training sample, and each first training sample respectively comprises personal characteristics, vehicle characteristics and full-stage vehicle insurance business characteristics of each training user in past years;
and training according to the first training sample set to obtain the basic complaint probability model.
In this embodiment, the basic complaint probability model is a hybrid logistic regression model, and the expression of the hybrid logistic regression model is as shown in formula (1):
wherein σ (·) represents a classification function, η (·) represents a fitting function, x represents a sample feature in the first training sample, { u 1,…,um } is a weight of x in the classification function, { w 1,…,wm } is a weight of x in the fitting function, p (y= 1|x) represents a probability of occurrence of an event y given the sample feature x, y represents an event of complaint by the training user, g (·) is used to ensure that the model conforms to the definition of the probability function, m represents a hyper-parameter tile, and m=3 is taken.
In this embodiment, the user complaint early warning device further includes: the initial complaint probability model training module is used for:
acquiring a second training sample set, wherein the second training sample set comprises a plurality of second training samples and complaint information corresponding to each second training sample, and each second training sample respectively comprises complaint time characteristics, complaint geographic characteristics, personal characteristics, vehicle characteristics and current-stage vehicle insurance service characteristics of each training user in the past year;
And training the initial complaint probability model corresponding to the current stage according to the second training sample set.
Those skilled in the art will also appreciate that the embodiments described in the specification are all preferred embodiments and that the modules referred to are not necessarily essential to the invention.
Example III
The invention also provides a computer device, such as a smart phone, a tablet computer, a notebook computer, a desktop computer, a rack-mounted server, a blade server, a tower server or a cabinet server (comprising independent servers or a server cluster formed by a plurality of servers) and the like which can execute programs. The computer device 20 of the present embodiment includes at least, but is not limited to: a memory 21, a processor 22, which may be communicatively coupled to each other via a system bus, as shown in fig. 5. It should be noted that fig. 5 only shows a computer device 20 having components 21-22, but it should be understood that not all of the illustrated components are required to be implemented, and that more or fewer components may be implemented instead.
In the present embodiment, the memory 21 (i.e., readable storage medium) includes a flash memory, a hard disk, a multimedia card, a card memory (e.g., SD or DX memory, etc.), a Random Access Memory (RAM), a Static Random Access Memory (SRAM), a read-only memory (ROM), an electrically erasable programmable read-only memory (EEPROM), a programmable read-only memory (PROM), a magnetic memory, a magnetic disk, an optical disk, and the like. In some embodiments, the memory 21 may be an internal storage unit of the computer device 20, such as a hard disk or memory of the computer device 20. In other embodiments, the memory 21 may also be an external storage device of the computer device 20, such as a plug-in hard disk provided on the computer device 20, a smart memory card (SMART MEDIA CARD, SMC), a Secure Digital (SD) card, a flash memory card (FLASH CARD), or the like. Of course, the memory 21 may also include both internal storage units of the computer device 20 and external storage devices. In this embodiment, the memory 21 is generally used for storing an operating system and various application software installed on the computer device 20, such as the program code of the user complaint early warning device 10 of the second embodiment. Further, the memory 21 may be used to temporarily store various types of data that have been output or are to be output.
Processor 22 may be a central processing unit (Central Processing Unit, CPU), controller, microcontroller, microprocessor, or other data processing chip in some embodiments. The processor 22 is generally used to control the overall operation of the computer device 20. In this embodiment, the processor 22 is configured to execute the program code or process data stored in the memory 21, for example, execute the user complaint early warning device 10, so as to implement the user complaint early warning method of the first embodiment.
Example IV
The present invention also provides a computer readable storage medium such as a flash memory, a hard disk, a multimedia card, a card memory (e.g., SD or DX memory, etc.), a Random Access Memory (RAM), a Static Random Access Memory (SRAM), a read-only memory (ROM), an electrically erasable programmable read-only memory (EEPROM), a programmable read-only memory (PROM), a magnetic memory, a magnetic disk, an optical disk, a server, an App application store, etc., on which a computer program is stored that when executed by a processor performs a corresponding function. The computer readable storage medium of the present embodiment is used for storing the user complaint early warning device 10, and when executed by the processor, implements the user complaint early warning method of the first embodiment.
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.
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 (5)
1. The user complaint early warning method is characterized by comprising the following steps of:
Acquiring unique identification information of a target user;
Acquiring user data corresponding to a target user from a preset user database according to the unique identification information, wherein the user data comprises personal characteristics, vehicle characteristics and full-stage vehicle insurance service characteristics corresponding to the target user in a preset time period, and further comprises time characteristics, geographic characteristics, personal characteristics, vehicle characteristics and current-stage vehicle insurance service characteristics corresponding to the target user;
preprocessing the user data;
predicting the final complaint probability of the current complaint of the target user according to the preprocessed user data;
Judging whether the final complaint probability is larger than a preset threshold value, if so, sending an alarm notification;
The step of predicting the final complaint probability of the current complaint of the target user according to the preprocessed target user characteristics comprises the following steps:
Inputting personal characteristics, vehicle characteristics and full-stage vehicle insurance service characteristics corresponding to a target user in a preset time period into basic complaint probability models corresponding to all service stages obtained through pre-training, and processing the basic complaint probability of the target user complaint in each service stage;
inputting the time feature, the geographic feature, the personal feature, the vehicle feature and the vehicle risk service feature corresponding to the current stage of the target user into an initial complaint probability model corresponding to the current stage of training in advance for processing to obtain the initial complaint probability of the current complaint of the target user;
Performing weighted summation operation on the basic complaint probability and the initial complaint probability to obtain the final complaint probability of the current complaint of the target user;
the basic complaint probability model is a mixed logistic regression model, and the expression of the mixed logistic regression model is shown as formula (1):
（1）
wherein, Representing a classification function,/>Representing a fitting function, x representing sample features in the first training sample,/>, andFor the weight of x in the classification function,/>For fitting the weight of x in the function,/>Representing the probability of occurrence of event y given sample feature x, y representing the event that trains the user to complain,/>The method is used for ensuring that the model accords with the definition of a probability function, m represents super-parameter slicing, and m=3 is taken;
the initial complaint probability model adopts a form of combining GBDT models and LR models;
the basic complaint probability model is trained as follows:
acquiring a first training sample set, wherein the first training sample set comprises a plurality of first training samples and complaint information corresponding to each first training sample, and each first training sample respectively comprises personal characteristics, vehicle characteristics and full-stage vehicle insurance business characteristics of each training user in past years;
Training according to the first training sample set to obtain the basic complaint probability model;
The training steps of the initial complaint probability model corresponding to the current stage are as follows:
acquiring a second training sample set, wherein the second training sample set comprises a plurality of second training samples and complaint information corresponding to each second training sample, and each second training sample respectively comprises complaint time characteristics, complaint geographic characteristics, personal characteristics, vehicle characteristics and current-stage vehicle insurance service characteristics of each training user in the past year;
And training the initial complaint probability model corresponding to the current stage according to the second training sample set.
2. The user complaint warning method of claim 1, wherein the preprocessing includes: data cleaning, feature extraction and normalization.
3. A user complaint warning device, comprising:
the identification information acquisition module is used for acquiring the unique identification information of the target user;
The user data acquisition module is used for acquiring user data corresponding to the target user from a preset user database according to the unique identification information, wherein the user data comprises personal characteristics, vehicle characteristics and full-stage vehicle insurance service characteristics corresponding to the target user in a preset time period, and also comprises time characteristics, geographic characteristics, personal characteristics, vehicle characteristics and current-stage vehicle insurance service characteristics corresponding to the target user;
The preprocessing module is used for preprocessing the user data;
The complaint probability prediction module is used for predicting the final complaint probability of the current complaint of the target user according to the preprocessed user data;
the early warning module is used for judging whether the final complaint probability is larger than a preset threshold value, and if so, sending an alarm notification; the complaint probability prediction module includes:
The basic complaint probability prediction unit is used for inputting the personal characteristics, the vehicle characteristics and the full-stage vehicle insurance service characteristics corresponding to the target user in a preset time period into basic complaint probability models corresponding to all service stages obtained through pre-training, and processing the basic complaint probability models to obtain basic complaint probability of the target user in each service stage of the vehicle insurance;
The initial complaint probability prediction unit is used for inputting the time feature, the geographic feature, the personal feature, the vehicle feature and the vehicle risk service feature corresponding to the current stage of the target user into the initial complaint probability model corresponding to the current stage obtained through pre-training for processing, and obtaining the initial complaint probability of the current complaint of the target user;
The weighting unit carries out weighted summation operation on the basic complaint probability and the initial complaint probability to obtain the final complaint probability of the current complaint of the target user;
the basic complaint probability model is a mixed logistic regression model, and the expression of the mixed logistic regression model is shown as formula (1):
（1）
wherein, Representing a classification function,/>Representing a fitting function, x representing sample features in the first training sample,/>, andFor the weight of x in the classification function,/>For fitting the weight of x in the function,/>Representing the probability of occurrence of event y given sample feature x, y representing the event that trains the user to complain,/>The method is used for ensuring that the model accords with the definition of a probability function, m represents super-parameter slicing, and m=3 is taken;
the initial complaint probability model adopts a form of combining GBDT models and LR models;
Wherein the user complaint early warning device also comprises a basic complaint probability model training module and an initial complaint probability model training module,
The basic complaint probability model training module is used for acquiring a first training sample set and training according to the first training sample set to obtain a basic complaint probability model, wherein the first training sample set comprises a plurality of first training samples and complaint information corresponding to each first training sample, and each first training sample respectively comprises personal characteristics, vehicle characteristics and full-stage vehicle insurance service characteristics of each training user in the past year;
The initial complaint probability model training module is used for acquiring a second training sample set and training the initial complaint probability model corresponding to the current stage according to the second training sample set, wherein the second training sample set comprises a plurality of second training samples and complaint information corresponding to each second training sample, and each second training sample respectively comprises complaint time characteristics, complaint geographic characteristics, personal characteristics, vehicle characteristics and vehicle insurance service characteristics of each training user in the past year.
4. A computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the processor implements the steps of the method according to any one of claims 1 to 2 when the computer program is executed by the processor.
5. A computer readable storage medium, on which a computer program is stored, characterized in that the computer program, when being executed by a processor, implements the steps of the method according to any one of claims 1 to 2.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910841818.9A CN110796554B (en) | 2019-09-06 | 2019-09-06 | User complaint early warning method and device, computer equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910841818.9A CN110796554B (en) | 2019-09-06 | 2019-09-06 | User complaint early warning method and device, computer equipment and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110796554A CN110796554A (en) | 2020-02-14 |
| CN110796554B true CN110796554B (en) | 2024-05-24 |
Family
ID=69427249
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910841818.9A Active CN110796554B (en) | 2019-09-06 | 2019-09-06 | User complaint early warning method and device, computer equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110796554B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112927091B (en) * | 2021-04-08 | 2023-11-10 | 泰康保险集团股份有限公司 | Complaint early warning method and device for annual gold insurance, computer equipment and medium |
| CN113240355B (en) * | 2021-07-12 | 2021-11-02 | 贝壳找房(北京)科技有限公司 | Training method and evaluation method of IM service evaluation model |
| CN114066049A (en) * | 2021-11-15 | 2022-02-18 | 国家电网有限公司客户服务中心 | A user complaint behavior prediction method, system, device and medium |
| CN115169588A (en) * | 2022-09-06 | 2022-10-11 | 深圳前海中电慧安科技有限公司 | Electrographic computation space-time trajectory vehicle code correlation method, device, equipment and storage medium |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109710766A (en) * | 2018-12-29 | 2019-05-03 | 云南电网有限责任公司电力科学研究院 | Method and device for early warning analysis of complaint tendency of work order data |
| CN109961248A (en) * | 2017-12-25 | 2019-07-02 | 顺丰科技有限公司 | Waybill complains prediction technique, device, equipment and its storage medium |
| CN109982367A (en) * | 2017-12-28 | 2019-07-05 | 中国移动通信集团四川有限公司 | Mobile terminal Internet access customer complaint prediction technique, device, equipment and storage medium |
-
2019
- 2019-09-06 CN CN201910841818.9A patent/CN110796554B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109961248A (en) * | 2017-12-25 | 2019-07-02 | 顺丰科技有限公司 | Waybill complains prediction technique, device, equipment and its storage medium |
| CN109982367A (en) * | 2017-12-28 | 2019-07-05 | 中国移动通信集团四川有限公司 | Mobile terminal Internet access customer complaint prediction technique, device, equipment and storage medium |
| CN109710766A (en) * | 2018-12-29 | 2019-05-03 | 云南电网有限责任公司电力科学研究院 | Method and device for early warning analysis of complaint tendency of work order data |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110796554A (en) | 2020-02-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110796554B (en) | User complaint early warning method and device, computer equipment and storage medium | |
| US20220284517A1 (en) | Automobile Monitoring Systems and Methods for Detecting Damage and Other Conditions | |
| CN110060144B (en) | Method for training credit model, method, device, equipment and medium for evaluating credit | |
| CN110704730B (en) | Product data pushing method and system based on big data and computer equipment | |
| CN108062674B (en) | Order fraud identification method, system, storage medium and electronic equipment based on GPS | |
| US11367141B1 (en) | Systems and methods for forecasting loss metrics | |
| US20150332411A1 (en) | Insurance Claims and Rate Evasion Fraud System Based Upon Vehicle History | |
| CN110659318A (en) | Big data based strategy pushing method and system and computer equipment | |
| CN110992072A (en) | Abnormal order prediction method and system | |
| CN110349039B (en) | Complaint risk assessment method, system, computer device and readable storage medium | |
| CN108416506A (en) | Customer risk level management approach, server and computer readable storage medium | |
| CN111881158A (en) | Management report data processing method and device, computer system and readable storage medium | |
| US20230342855A1 (en) | Systems and methods for detecting insurance claim fraud by using image data validation | |
| CN112598513B (en) | Method and device for identifying stockholder risk transaction behaviors | |
| CN113222624B (en) | Intelligent analysis method and system for preventing electricity stealing | |
| CN108416619B (en) | Consumption interval time prediction method and device and readable storage medium | |
| CN112990386A (en) | User value clustering method and device, computer equipment and storage medium | |
| CN115936895A (en) | Risk assessment method, device and equipment based on artificial intelligence and storage medium | |
| CN110674129A (en) | Abnormal event processing method, system, computer equipment and storage medium | |
| CN114840660B (en) | Business recommendation model training method, device, equipment and storage medium | |
| CN110097250B (en) | Product risk prediction method, device, computer equipment and storage medium | |
| CN114048330B (en) | Risk conduction probability knowledge graph generation method, apparatus, device and storage medium | |
| CN112036488A (en) | Event identification method, device and equipment based on image identification and storage medium | |
| CN111524614A (en) | Epidemic situation information notification system | |
| CN117437019A (en) | Credit card overdue risk prediction method, apparatus, device, medium and program product |
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 |