CN113421116A - User recall analysis method, device, equipment and storage medium - Google Patents
User recall analysis method, device, equipment and storage medium Download PDFInfo
- Publication number
- CN113421116A CN113421116A CN202110692005.5A CN202110692005A CN113421116A CN 113421116 A CN113421116 A CN 113421116A CN 202110692005 A CN202110692005 A CN 202110692005A CN 113421116 A CN113421116 A CN 113421116A
- Authority
- CN
- China
- Prior art keywords
- user
- data
- recall
- loss
- preset
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000004458 analytical method Methods 0.000 title claims abstract description 75
- 238000003860 storage Methods 0.000 title claims abstract description 23
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 64
- 238000007781 pre-processing Methods 0.000 claims abstract description 11
- 230000006399 behavior Effects 0.000 claims description 39
- 230000008451 emotion Effects 0.000 claims description 34
- 238000012544 monitoring process Methods 0.000 claims description 25
- 238000012163 sequencing technique Methods 0.000 claims description 19
- 238000012360 testing method Methods 0.000 claims description 13
- 238000001914 filtration Methods 0.000 claims description 11
- 230000002159 abnormal effect Effects 0.000 claims description 9
- 238000012549 training Methods 0.000 claims description 9
- 230000000694 effects Effects 0.000 claims description 8
- 230000002996 emotional effect Effects 0.000 claims description 7
- 238000000605 extraction Methods 0.000 claims description 6
- 238000012545 processing Methods 0.000 claims description 6
- 238000007635 classification algorithm Methods 0.000 claims description 5
- 238000004140 cleaning Methods 0.000 claims description 4
- 238000005516 engineering process Methods 0.000 abstract description 2
- 238000013473 artificial intelligence Methods 0.000 abstract 1
- 238000000034 method Methods 0.000 description 23
- 230000008569 process Effects 0.000 description 12
- 239000000284 extract Substances 0.000 description 8
- 238000010586 diagram Methods 0.000 description 6
- 239000003814 drug Substances 0.000 description 3
- 230000008859 change Effects 0.000 description 2
- 201000010099 disease Diseases 0.000 description 2
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 2
- 230000036541 health Effects 0.000 description 2
- 230000007246 mechanism Effects 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 238000002360 preparation method Methods 0.000 description 2
- 230000008901 benefit Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000007621 cluster analysis Methods 0.000 description 1
- 238000010835 comparative analysis Methods 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 229940079593 drug Drugs 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000008520 organization Effects 0.000 description 1
- 230000002085 persistent effect Effects 0.000 description 1
- 239000000955 prescription drug Substances 0.000 description 1
- 229940126532 prescription medicine Drugs 0.000 description 1
- 230000002265 prevention Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000001052 transient effect Effects 0.000 description 1
Images
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
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0201—Market modelling; Market analysis; Collecting market data
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
Abstract
The invention relates to the field of artificial intelligence, and discloses a user recall analysis method, a device, equipment and a storage medium, which are used for improving the user recall probability. The user recall analysis method comprises the following steps: acquiring initial user data, preprocessing the initial user data to obtain preprocessed data, and extracting target user data from the preprocessed data; calling a preset loss probability prediction model, and predicting the loss probability of target user data to obtain the user loss probability; calling a preset clustering algorithm, analyzing target user data and user loss probability to obtain user loss reason information, and determining a user recall strategy according to the user loss reason information; and acquiring user loss data and a user recall result, and evaluating and optimizing the user recall strategy to obtain an optimized user recall strategy. In addition, the invention also relates to a block chain technology, and the optimized user recall strategy can be stored in the block chain node.
Description
Technical Field
The invention relates to the field of gradient spanning trees, in particular to a user recall analysis method, device, equipment and storage medium.
Background
For the current internet enterprises, the research on the loss reason of the users and the establishment of a loss prevention strategy have important significance, and the loss of old users is reduced along with the reduction of user traffic dividends and higher cost, so that the method has a very obvious effect on improving the unit economic benefit of the enterprises.
Most enterprises in the market do not have a perfect user recall analysis mechanism at present, most enterprises in the market analyze based on statistical means, and a lost user group is indirectly judged by means of a single angle or a plurality of isolated angles, but users are difficult to recall once lost, so that the function of early warning is not achieved, the analysis means in the prior art is simple, only the lost results of the users are analyzed, the reason of the user loss is not found, and the recall probability of the lost users is low.
Disclosure of Invention
The invention provides a user recall analysis method, a device, equipment and a storage medium, which are used for calling a preset loss probability prediction model to predict loss probability, calling a preset clustering algorithm to analyze target user data and user loss probability, determining user loss reason information, determining a user recall strategy according to the user loss reason information and improving the user recall probability.
The first aspect of the present invention provides a user recall analysis method, including: acquiring initial user data, preprocessing the initial user data to obtain preprocessed data, and extracting target user data from the preprocessed data, wherein the target user data comprises user behavior data and user emotion data; calling a preset loss probability prediction model, and predicting the loss probability of the target user data to obtain the user loss probability; calling a preset clustering algorithm, analyzing the target user data and the user loss probability to obtain user loss reason information, and determining a user recall strategy according to the user loss reason information; and obtaining user loss data and user recall results based on the user recall strategy, and evaluating and optimizing the user recall strategy based on the user loss data and the user recall results to obtain the optimized user recall strategy.
Optionally, in a first implementation manner of the first aspect of the present invention, the acquiring initial user data, preprocessing the initial user data to obtain preprocessed data, and extracting target user data from the preprocessed data, where the target user data includes user behavior data and user emotion data, and the method includes: acquiring initial user data, calling a preset interpolation algorithm, filling missing values in the initial user data, and performing abnormal value filtering and repeated value filtering on the filled initial user data to obtain preprocessed data; extracting user behavior data from the preprocessed data, wherein the user behavior data comprises behavior type data and behavior occurrence time data of a user within a preset time range; extracting user dialogue record data from the preprocessed data, calling a preset deep network model, and performing feature extraction on the user dialogue record data to obtain user emotion features; classifying the emotional characteristics of the users according to preset emotion types to obtain user emotion data, and combining the user behavior data and the user emotion data to obtain target user data.
Optionally, in a second implementation manner of the first aspect of the present invention, the invoking a preset churn probability prediction model to predict churn probability of the target user data, and obtaining the user churn probability includes: calling a preset loss probability prediction model, and predicting the loss probability of the target user data to obtain an initial loss probability; calling a preset extreme gradient lifting algorithm, training the loss probability prediction model based on the target user data to obtain a trained loss probability prediction model, calling the trained loss probability prediction model, and correcting the initial loss probability to obtain the user loss probability.
Optionally, in a third implementation manner of the first aspect of the present invention, the calling a preset extreme gradient lifting algorithm to train the attrition probability prediction model to obtain a trained attrition probability prediction model, calling the trained attrition probability prediction model to correct the initial attrition probability to obtain the user attrition probability includes: filtering and cleaning the target user data to obtain input parameters; calling a preset extreme gradient lifting algorithm, and performing classification tree structure training and prediction on the input parameters for preset times to obtain a prediction result; fitting the loss probability prediction model based on the prediction result to obtain a trained loss probability prediction model; and calling the trained loss probability prediction model, and correcting the initial loss probability to obtain the user loss probability.
Optionally, in a fourth implementation manner of the first aspect of the present invention, the invoking a preset clustering algorithm to analyze the target user data and the user churn probability to obtain user churn reason information, and determining a user recall policy according to the user churn reason information includes: calling a preset clustering algorithm to analyze the target user data to obtain user loss reason information; calling a preset data classification algorithm to classify the user loss reason information to obtain loss reason label information; acquiring initial user label information when the user loss probability reaches a preset threshold value, and updating the initial user label information through the loss reason label information to obtain target user label information; and determining a user recall strategy based on the target user label information, wherein the user recall strategy comprises issuing coupons and pushing messages.
Optionally, in a fifth implementation manner of the first aspect of the present invention, the acquiring user churn data and user recall results based on the user recall policy, and the evaluating and optimizing the user recall policy based on the user churn data and the user recall results to obtain an optimized user recall policy includes: acquiring user loss data and user recall results based on the user recall strategy, and monitoring the user loss data and the user recall results in real time within a preset time period to obtain monitoring results, wherein the user recall results are corresponding user recall numbers after the user recall strategy is implemented, and the user recall results are used for indicating user recall effects of the user recall strategy; analyzing the monitoring result to obtain an analysis result, calling a preset grouping test algorithm, and evaluating and optimizing the user recall strategy based on the analysis result to obtain an optimized user recall strategy.
Optionally, in a sixth implementation manner of the first aspect of the present invention, the analyzing the monitoring result to obtain an analysis result, invoking a preset grouping test algorithm, and evaluating and optimizing the user recall policy based on the analysis result, where the obtaining of the optimized user recall policy includes: comparing and analyzing preset conditions and the monitoring result to obtain an analysis result; if the analysis result is that the monitoring result does not meet the preset condition, grouping the users corresponding to the initial user data through a preset grouping test algorithm to obtain a plurality of user groups, wherein each user group corresponds to a user recall strategy; and sequencing the user loss data corresponding to each user group to obtain a sequencing result, and adjusting the user recall strategy of which the user loss data is greater than a preset threshold value in the sequencing result to obtain an optimized user recall strategy.
A second aspect of the present invention provides a user recall analysis apparatus comprising: the system comprises an acquisition module, a storage module and a processing module, wherein the acquisition module is used for acquiring initial user data, preprocessing the initial user data to obtain preprocessed data, and extracting target user data from the preprocessed data, and the target user data comprises user behavior data and user emotion data; the prediction module is used for calling a preset loss probability prediction model to predict the loss probability of the target user data to obtain the user loss probability; the determining module is used for calling a preset clustering algorithm, analyzing the target user data and the user loss probability to obtain user loss reason information, and determining a user recall strategy according to the user loss reason information; and the generating module is used for acquiring user loss data and user recall results based on the user recall strategy, and evaluating and optimizing the user recall strategy based on the user loss data and the user recall results to obtain the optimized user recall strategy.
A third aspect of the present invention provides a user recall analysis apparatus comprising: a memory and at least one processor, the memory having instructions stored therein; the at least one processor invokes the instructions in the memory to cause the user recall analysis apparatus to perform the user recall analysis method described above.
A fourth aspect of the present invention provides a computer-readable storage medium having stored therein instructions, which, when run on a computer, cause the computer to perform the user recall analysis method described above.
In the technical scheme provided by the invention, initial user data is obtained, the initial user data is preprocessed to obtain preprocessed data, target user data is extracted from the preprocessed data, and the target user data comprises user behavior data and user emotion data; calling a preset loss probability prediction model, and predicting the loss probability of the target user data to obtain the user loss probability; calling a preset clustering algorithm, analyzing the target user data and the user loss probability to obtain user loss reason information, and determining a user recall strategy according to the user loss reason information; and obtaining user loss data and user recall results based on the user recall strategy, and evaluating and optimizing the user recall strategy based on the user loss data and the user recall results to obtain the optimized user recall strategy. In the embodiment of the invention, the preset loss probability prediction model is called to predict the loss probability, the preset clustering algorithm is called to analyze the target user data and the user loss probability, the user loss reason information is determined, the user recall strategy is determined according to the user loss reason information, and the user recall probability is improved.
Drawings
FIG. 1 is a schematic diagram of an embodiment of a user recall analysis method according to an embodiment of the present invention;
FIG. 2 is a schematic diagram of another embodiment of a user recall analysis method according to an embodiment of the present invention;
FIG. 3 is a schematic diagram of an embodiment of a user recall analysis apparatus according to an embodiment of the present invention;
FIG. 4 is a schematic diagram of another embodiment of a user recall analysis apparatus according to an embodiment of the present invention;
fig. 5 is a schematic diagram of an embodiment of a user recall analysis apparatus in the embodiment of the present invention.
Detailed Description
The embodiment of the invention provides a user recall analysis method, a device, equipment and a storage medium, which are used for calling a preset loss probability prediction model to predict loss probability, calling a preset clustering algorithm to analyze target user data and user loss probability, determining user loss reason information, determining a user recall strategy according to the user loss reason information and improving the user recall probability.
The terms "first," "second," "third," "fourth," and the like in the description and in the claims, as well as in the drawings, if any, are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It will be appreciated that the data so used may be interchanged under appropriate circumstances such that the embodiments described herein may be practiced otherwise than as specifically illustrated or described herein. Furthermore, the terms "comprises," "comprising," or "having," and any variations thereof, are intended to cover non-exclusive inclusions, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed, but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.
For convenience of understanding, a detailed flow of an embodiment of the present invention is described below, and referring to fig. 1, an embodiment of a user recall analysis method according to an embodiment of the present invention includes:
101. the method comprises the steps of obtaining initial user data, preprocessing the initial user data to obtain preprocessed data, and extracting target user data from the preprocessed data, wherein the target user data comprise user behavior data and user emotion data.
It is to be understood that the executing entity of the present invention may be a user recalling the analysis apparatus, and may also be a terminal or a server, which is not limited herein. The embodiment of the present invention is described by taking a server as an execution subject.
The server acquires initial user data, preprocesses the initial user data to obtain preprocessed data, and extracts target user data from the preprocessed data, wherein the target user data comprises user behavior data and user emotion data. The execution process of preprocessing the initial user data may be: the method comprises the steps of carrying out missing value filling, abnormal value filtering and repeated value filtering on initial user data to obtain preprocessed data, extracting target user data from the preprocessed data by a server, wherein the target user data is mainly divided into two types, the first type is user behavior data, the user behavior data can be related data in the medical field, the second type is user emotion data, the user emotion data can be user dialogue recorded data in the inquiry process, and the server combines the user behavior data and the user emotion data to finally obtain the target user data.
102. And calling a preset loss probability prediction model to predict the loss probability of the target user data to obtain the user loss probability.
And calling a preset loss probability prediction model by the server to predict the loss probability of the target user data to obtain the user loss probability. The server predicts the loss probability of the target user data by calling a preset loss probability model to obtain an initial loss probability, and a preset extreme gradient boosting (XGB) algorithm is called to train the loss probability model to obtain a trained loss probability prediction model, the XGB algorithm trains input parameters for k times of classification and regression tree (CART), wherein, the residual errors of all the previous predictions are considered in the process of each prediction, the XGB algorithm is a stack of CART trees, the idea of the XGB algorithm is to continuously add trees to fit the residual errors of the previous prediction, when the training is finished to obtain k trees, the score of one sample is predicted, according to the characteristics of the sample, a corresponding leaf node is inquired in each tree, each leaf node corresponds to a score, and finally, the scores corresponding to each tree are added to form a predicted value (namely, the user loss probability).
103. And calling a preset clustering algorithm, analyzing the target user data and the user loss probability to obtain user loss reason information, and determining a user recall strategy according to the user loss reason information.
And the server calls a preset clustering algorithm to analyze the target user data and the user loss probability to obtain user loss reason information, and determines a user recall strategy according to the user loss reason information. The server determines the user loss reasons by carrying out cluster analysis on target user data, divides the user loss reasons into a plurality of categories or carries out more detailed reason subdivision, such as inquiry service quality, prescription medicine price, medicine purchasing logistics, health content quality and other reasons, and can also divide the reasons into more detailed parts by properly combining service characteristics, and determines a corresponding user recall strategy according to user loss reason information.
104. And obtaining user loss data and user recall results based on the user recall strategy, and evaluating and optimizing the user recall strategy based on the user loss data and the user recall results to obtain the optimized user recall strategy.
The server acquires user loss data and user recall results based on the user recall strategy, and evaluates and optimizes the user recall strategy based on the user loss data and the user recall results to obtain the optimized user recall strategy. For example: the server periodically monitors the change of user loss data within 1-3 months, obtains monitoring results according to different user loss reasons and corresponding recall strategy effects, performs comparative analysis on preset conditions and the monitoring results to obtain analysis results, and adopts a grouping test algorithm if the analysis results indicate that the monitoring results do not meet the preset conditions, wherein the preset conditions include that the user loss data are smaller than a preset loss data threshold value and the user recall number is larger than a preset recall number threshold value, and the grouping test algorithm specifically comprises the following steps: dividing users into N groups, wherein the 1 st group adopts a strategy 1, the K group adopts a strategy K, sequencing user loss data corresponding to each user group from large to small to obtain a sequencing result, comprehensively evaluating the rationality of a user recall strategy according to the sequencing result, indicating more user loss in the sequencing result, and adjusting the user recall strategy of which the user loss data in the server sequencing result is greater than a preset threshold value, such as adding a new recall strategy or modifying the original user recall strategy, and continuously and iteratively optimizing the user recall strategy.
In the embodiment of the invention, the preset loss probability prediction model is called to predict the loss probability, the preset clustering algorithm is called to analyze the target user data and the user loss probability, the user loss reason information is determined, the user recall strategy is determined according to the user loss reason information, and the user recall probability is improved.
Referring to fig. 2, another embodiment of a method for analyzing user recall according to an embodiment of the present invention includes:
201. the method comprises the steps of obtaining initial user data, preprocessing the initial user data to obtain preprocessed data, and extracting target user data from the preprocessed data, wherein the target user data comprise user behavior data and user emotion data.
The server acquires initial user data, preprocesses the initial user data to obtain preprocessed data, and extracts target user data from the preprocessed data, wherein the target user data comprises user behavior data and user emotion data. Specifically, the server acquires initial user data, calls a preset interpolation algorithm, fills missing values in the initial user data, and filters abnormal values and repeated values in the filled initial user data to obtain preprocessed data; the server extracts user behavior data from the preprocessed data, wherein the user behavior data comprises behavior type data and behavior occurrence time data of a user within a preset time range; the server extracts user dialogue record data from the preprocessed data, calls a preset deep network model, and performs feature extraction on the user dialogue record data to obtain user emotion features; the server classifies the emotional characteristics of the users according to preset emotional categories to obtain user emotional data, and combines the user behavior data and the user emotional data to obtain target user data.
In the data preparation stage, a server firstly acquires initial user data and preprocesses the initial user data, the preprocessing process mainly comprises data cleaning, namely processing a missing value, an abnormal value and a repeated value, the missing value processing comprises but is not limited to multiple interpolation, the abnormal value processing mainly adopts an abnormal value detection algorithm z-score to identify the abnormal value and delete the abnormal value, meanwhile, the repeated value is processed in a duplicate removal mode to obtain preprocessed data, the server extracts target user data from the preprocessed data, the target user data is mainly divided into two types, the first type is user behavior data, the user behavior data can be medical field related data which comprises medicine purchasing condition data, prescription data, inquiry times in the last 30 days, disease data and the like which can be specifically quantized or definite, the second type is user emotion data, the server extracts characteristics of user dialogue records by calling a deep network model, selecting four types of emotional conversations including happiness, calmness, anger, sadness and the like to obtain user emotion data, wherein the user emotion data can be recorded data of user conversations in the inquiry process, and the server combines the user behavior data and the user emotion data to finally obtain target user data.
202. And calling a preset loss probability prediction model to predict the loss probability of the target user data to obtain the user loss probability.
And calling a preset loss probability prediction model by the server to predict the loss probability of the target user data to obtain the user loss probability. Specifically, the server calls a preset loss probability prediction model to predict the loss probability of the target user data to obtain an initial loss probability; the server calls a preset extreme gradient lifting algorithm, trains the loss probability prediction model based on target user data to obtain a trained loss probability prediction model, calls the trained loss probability prediction model, and corrects the initial loss probability to obtain the user loss probability. The server selects an extreme gradient lifting XGB algorithm to perform model training, evaluation and optimization, target user data obtained in a data preparation stage is used as input parameters of the XGB, the XGB performs k times of classification tree structure CART training on the input parameters, residual errors of all previous times of prediction are considered in each prediction process, the XGB algorithm corresponds to a stack of CART trees, the algorithm idea is to continuously add trees to fit the residual errors of the previous times of prediction, when the k trees are obtained after the training is completed, the score of one sample is predicted, namely, according to the characteristics of the sample, a corresponding leaf node is inquired in each tree, each leaf node corresponds to one score, and finally, the score corresponding to each tree is added to form a predicted value, namely, the user loss probability in the embodiment, in the actual inquiry geographic position is missing, most of diseases are in sparse evidence-holding scenes such as inquiry categories and the like, compared with other classification tree algorithms, the XGB algorithm is adopted to predict the sparse matrix well in the embodiment, after the preset loss probability prediction model is trained to obtain the trained loss probability prediction model, the server corrects the initial loss probability through the trained loss probability prediction model to obtain the user loss probability.
203. And calling a preset clustering algorithm to analyze the target user data to obtain the user loss reason information.
And the server calls a preset clustering algorithm to analyze the target user data to obtain the user loss reason information. In the clustering algorithm, a data set is divided into different classes or clusters according to a specific standard, so that the similarity of data objects in the same cluster is as large as possible, and the difference of the data objects which are not in the same cluster is also as large as possible, that is, the data of the same class are gathered together as much as possible after clustering, and different data are separated as much as possible.
204. And calling a preset data classification algorithm to classify the loss reason information of the user to obtain the loss reason label information.
And calling a preset data classification algorithm by the server to classify the loss reason information of the user to obtain the loss reason label information. In this embodiment, a K-nearest neighbor algorithm is used to classify the loss causes of the users to obtain the loss cause label information, the algorithm classifies the loss causes by measuring distances between different feature values, where K is an integer no greater than 20, and selects a class with the largest occurrence frequency among the K most similar data as a class of new data (i.e., loss cause label information).
205. And acquiring initial user tag information when the user loss probability reaches a preset threshold value, and updating the initial user tag information through the loss reason tag information to obtain target user tag information.
The server obtains initial user label information when the user loss probability reaches a preset threshold value, and updates the initial user label information through the loss reason label information to obtain target user label information. In this embodiment, the server writes a loss reason tag to the user whose user loss probability reaches the preset threshold, (i.e., updates the initial user tag information through the loss reason tag information), so as to obtain target user tag information, where the target user tag information may include at least one of inquiry service quality tag information, prescription drug price tag information, drug purchase flow tag information, and health content quality tag information.
206. And determining a user recall strategy based on the target user label information, wherein the user recall strategy comprises issuing coupons and pushing messages.
The server determines a user recall strategy based on the target user label information, wherein the user recall strategy comprises issuing coupons and pushing messages. The server determines a user recall strategy for each loss reason information, such as issuing a coupon for a user with sensitive price, pushing operation activities of high-quality healthy content organization for a content user, reducing issuing of prompt information for a user with sensitive message pushing and the like, determines a plurality of corresponding user recall strategies for users containing a plurality of loss reason label information, and can be a combination of two or more user recall strategies.
207. And obtaining user loss data and user recall results based on the user recall strategy, and evaluating and optimizing the user recall strategy based on the user loss data and the user recall results to obtain the optimized user recall strategy.
The server acquires user loss data and user recall results based on the user recall strategy, and evaluates and optimizes the user recall strategy based on the user loss data and the user recall results to obtain the optimized user recall strategy. Specifically, the server acquires user loss data and user recall results based on the user recall strategy, and monitors the user loss data and the user recall results in real time within a preset time period to obtain monitoring results, wherein the user recall results are corresponding user recall numbers after the user recall strategy is implemented, and the user recall results are used for indicating the user recall effect of the user recall strategy; and the server analyzes the monitoring result to obtain an analysis result, calls a preset grouping test algorithm, evaluates and optimizes the user recall strategy based on the analysis result, and obtains the optimized user recall strategy.
For example: the server periodically monitors the change of user loss data within 1-3 months, and the effect of different user loss reasons and corresponding recall strategies to obtain a monitoring result, and compares and analyzes the preset condition with the monitoring result to obtain an analysis result, if the analysis result is that the monitoring result does not meet the preset condition, a grouping test algorithm is adopted, the preset condition comprises that the user loss data is smaller than a preset loss data threshold value and the user recall number is larger than a preset recall number threshold value, the user loss data in the embodiment refers to the number of user loss, and the grouping test algorithm specifically comprises: dividing users into N groups, wherein the 1 st group adopts a strategy 1, the K group adopts a strategy K, sequencing user loss data corresponding to each user group from large to small to obtain a sequencing result, comprehensively evaluating the rationality of a user recall strategy according to the sequencing result, indicating more user loss in the sequencing result, and adjusting the user recall strategy of which the user loss data in the server sequencing result is greater than a preset threshold value, such as adding a new recall strategy or modifying the original user recall strategy, and continuously and iteratively optimizing the user recall strategy.
In the embodiment of the invention, the preset loss probability prediction model is called to predict the loss probability, the preset clustering algorithm is called to analyze the target user data and the user loss probability, the user loss reason information is determined, the user recall strategy is determined according to the user loss reason information, and the user recall probability is improved.
With reference to fig. 3, the user recall analysis method in the embodiment of the present invention is described above, and a user recall analysis apparatus in the embodiment of the present invention is described below, where an embodiment of the user recall analysis apparatus in the embodiment of the present invention includes:
an obtaining module 301, configured to obtain initial user data, pre-process the initial user data to obtain pre-processed data, and extract target user data from the pre-processed data, where the target user data includes user behavior data and user emotion data;
the prediction module 302 is configured to invoke a preset loss probability prediction model, and predict loss probability of target user data to obtain user loss probability;
the determining module 303 is configured to invoke a preset clustering algorithm, analyze the target user data and the user loss probability to obtain user loss reason information, and determine a user recall strategy according to the user loss reason information;
the generating module 304 is configured to obtain user loss data and user recall results based on the user recall policy, and evaluate and optimize the user recall policy based on the user loss data and the user recall results to obtain an optimized user recall policy.
In the embodiment of the invention, the preset loss probability prediction model is called to predict the loss probability, the preset clustering algorithm is called to analyze the target user data and the user loss probability, the user loss reason information is determined, the user recall strategy is determined according to the user loss reason information, and the user recall probability is improved.
Referring to fig. 4, another embodiment of a user recall analysis apparatus according to an embodiment of the present invention includes:
an obtaining module 301, configured to obtain initial user data, pre-process the initial user data to obtain pre-processed data, and extract target user data from the pre-processed data, where the target user data includes user behavior data and user emotion data;
the prediction module 302 is configured to invoke a preset loss probability prediction model, and predict loss probability of target user data to obtain user loss probability;
the determining module 303 is configured to invoke a preset clustering algorithm, analyze the target user data and the user loss probability to obtain user loss reason information, and determine a user recall strategy according to the user loss reason information;
the determining module 303 specifically includes:
an analyzing unit 3031, configured to invoke a preset clustering algorithm to analyze target user data to obtain user loss cause information;
a classifying unit 3032, configured to invoke a preset data classification algorithm to classify the loss reason information of the user, so as to obtain loss reason label information;
an updating unit 3033, configured to obtain initial user tag information when the user churn probability reaches a preset threshold, and update the initial user tag information through churn cause tag information to obtain target user tag information;
a determining unit 3034, configured to determine a user recall policy based on the target user tag information, where the user recall policy includes issuing a coupon and pushing a message;
the generating module 304 is configured to obtain user loss data and user recall results based on the user recall policy, and evaluate and optimize the user recall policy based on the user loss data and the user recall results to obtain an optimized user recall policy.
Optionally, the obtaining module 301 includes:
the obtaining unit 3011 is configured to obtain initial user data, invoke a preset interpolation algorithm, fill a missing value in the initial user data, and perform outlier filtering and repeated value filtering on the filled initial user data to obtain preprocessed data;
a first extraction unit 3012, configured to extract user behavior data from the preprocessed data, where the user behavior data includes behavior type data and behavior occurrence time data of a user within a preset time range;
a second extraction unit 3013, configured to extract user session record data from the preprocessed data, invoke a preset deep network model, and perform feature extraction on the user session record data to obtain user emotion features;
and the merging unit 3014 is configured to classify the emotion characteristics of the users according to preset emotion categories to obtain user emotion data, and merge the user behavior data and the user emotion data to obtain target user data.
Optionally, the prediction module 302 includes:
the prediction unit 3021 is configured to invoke a preset loss probability prediction model, and predict loss probability of target user data to obtain an initial loss probability;
and the correcting unit 3022 is configured to invoke a preset extreme gradient lifting algorithm, train the loss probability prediction model based on the target user data to obtain a trained loss probability prediction model, and invoke the trained loss probability prediction model to correct the initial loss probability to obtain a user loss probability.
Optionally, the modifying unit 3022 may be further specifically configured to:
filtering and cleaning the target user data to obtain input parameters; calling a preset extreme gradient lifting algorithm, and performing classification tree structure training and prediction on input parameters for preset times to obtain a prediction result; fitting the loss probability prediction model based on the prediction result to obtain a trained loss probability prediction model; and calling the trained loss probability prediction model, and correcting the initial loss probability to obtain the user loss probability.
Optionally, the generating module 304 includes:
a monitoring unit 3041, configured to acquire user loss data and a user recall result based on a user recall policy, and monitor the user loss data and the user recall result in real time within a preset time period to obtain a monitoring result, where the user recall result is a corresponding user recall number after the user recall policy is implemented, and the user recall result is used to indicate a user recall effect of the user recall policy;
the optimizing unit 3042 is configured to analyze the monitoring result to obtain an analysis result, call a preset grouping test algorithm, and evaluate and optimize the user recall policy based on the analysis result to obtain an optimized user recall policy.
Optionally, the optimization unit 3042 may be further specifically configured to:
comparing and analyzing the preset conditions and the monitoring results to obtain analysis results; if the analysis result is that the monitoring result does not meet the preset condition, grouping the users corresponding to the initial user data through a preset grouping test algorithm to obtain a plurality of user groups, wherein each user group corresponds to a user recall strategy; and sequencing the user loss data corresponding to each user group to obtain a sequencing result, and adjusting the user recall strategy of which the user loss data is greater than a preset threshold value in the sequencing result to obtain an optimized user recall strategy.
In the embodiment of the invention, the preset loss probability prediction model is called to predict the loss probability, the preset clustering algorithm is called to analyze the target user data and the user loss probability, the user loss reason information is determined, the user recall strategy is determined according to the user loss reason information, and the user recall probability is improved.
Fig. 3 and 4 above describe the user recall analysis apparatus in the embodiment of the present invention in detail from the perspective of the modular functional entity, and the user recall analysis apparatus in the embodiment of the present invention is described in detail from the perspective of hardware processing.
Fig. 5 is a schematic structural diagram of a user recall analysis apparatus 500 according to an embodiment of the present invention, where the user recall analysis apparatus 500 may have a relatively large difference due to different configurations or performances, and may include one or more processors (CPUs) 510 (e.g., one or more processors) and a memory 520, one or more storage media 530 (e.g., one or more mass storage devices) for storing applications 533 or data 532. Memory 520 and storage media 530 may be, among other things, transient or persistent storage. The program stored on the storage medium 530 may include one or more modules (not shown), each of which may include a series of instruction operations for a user to recall the analysis apparatus 500. Still further, the processor 510 may be configured to communicate with the storage medium 530 to execute a series of instruction operations in the storage medium 530 on the user recall analysis device 500.
The user recall analysis apparatus 500 may also include one or more power supplies 540, one or more wired or wireless network interfaces 550, one or more input-output interfaces 560, and/or one or more operating systems 531, such as Windows Server, Mac OS X, Unix, Linux, FreeBSD, and the like. Those skilled in the art will appreciate that the user recalled analysis device configuration shown in FIG. 5 does not constitute a limitation of user recalled analysis devices and may include more or fewer components than shown, or some components may be combined, or a different arrangement of components.
The present invention also provides a user recall analysis apparatus, the computer apparatus comprising a memory and a processor, the memory having stored therein computer readable instructions, which, when executed by the processor, cause the processor to perform the steps of the user recall analysis method in the above embodiments.
The present invention also provides a computer readable storage medium, which may be a non-volatile computer readable storage medium, which may also be a volatile computer readable storage medium, having stored therein instructions, which, when run on a computer, cause the computer to perform the steps of the user recall analysis method.
The block chain is a novel application mode of computer technologies such as distributed data storage, point-to-point transmission, a consensus mechanism, an encryption algorithm and the like. A block chain (Blockchain), which is essentially a decentralized database, is a series of data blocks associated by using a cryptographic method, and each data block contains information of a batch of network transactions, so as to verify the validity (anti-counterfeiting) of the information and generate a next block. The blockchain may include a blockchain underlying platform, a platform product service layer, an application service layer, and the like.
It is clear to those skilled in the art that, for convenience and brevity of description, the specific working processes of the above-described systems, apparatuses and units may refer to the corresponding processes in the foregoing method embodiments, and are not described herein again.
The integrated unit, if implemented in the form of a software functional unit and sold or used as a stand-alone product, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present invention may be embodied in the form of a software product, which is stored in a storage medium and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: various media capable of storing program codes, such as a usb disk, a removable hard disk, a read-only memory (ROM), a Random Access Memory (RAM), a magnetic disk, or an optical disk.
The above-mentioned embodiments are only used for illustrating the technical solutions of the present invention, and not for limiting the same; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.
Claims (10)
1. A user recall analysis method, the user recall analysis method comprising:
acquiring initial user data, preprocessing the initial user data to obtain preprocessed data, and extracting target user data from the preprocessed data, wherein the target user data comprises user behavior data and user emotion data;
calling a preset loss probability prediction model, and predicting the loss probability of the target user data to obtain the user loss probability;
calling a preset clustering algorithm, analyzing the target user data and the user loss probability to obtain user loss reason information, and determining a user recall strategy according to the user loss reason information;
and obtaining user loss data and user recall results based on the user recall strategy, and evaluating and optimizing the user recall strategy based on the user loss data and the user recall results to obtain the optimized user recall strategy.
2. The user recall analysis method of claim 1, wherein the obtaining initial user data, preprocessing the initial user data to obtain preprocessed data, and extracting target user data from the preprocessed data, wherein the target user data comprises user behavior data and user emotion data, comprises:
acquiring initial user data, calling a preset interpolation algorithm, filling missing values in the initial user data, and performing abnormal value filtering and repeated value filtering on the filled initial user data to obtain preprocessed data;
extracting user behavior data from the preprocessed data, wherein the user behavior data comprises behavior type data and behavior occurrence time data of a user within a preset time range;
extracting user dialogue record data from the preprocessed data, calling a preset deep network model, and performing feature extraction on the user dialogue record data to obtain user emotion features;
classifying the emotional characteristics of the users according to preset emotion types to obtain user emotion data, and combining the user behavior data and the user emotion data to obtain target user data.
3. The user recall analysis method of claim 1, wherein the invoking a preset churn probability prediction model to predict churn probability for the target user data, and obtaining a user churn probability comprises:
calling a preset loss probability prediction model, and predicting the loss probability of the target user data to obtain an initial loss probability;
calling a preset extreme gradient lifting algorithm, training the loss probability prediction model based on the target user data to obtain a trained loss probability prediction model, calling the trained loss probability prediction model, and correcting the initial loss probability to obtain the user loss probability.
4. The user recall analysis method of claim 3, wherein the invoking a preset extreme gradient boost algorithm to train the attrition probability prediction model to obtain a trained attrition probability prediction model, invoking the trained attrition probability prediction model to correct the initial attrition probability to obtain a user attrition probability comprises:
filtering and cleaning the target user data to obtain input parameters;
calling a preset extreme gradient lifting algorithm, and performing classification tree structure training and prediction on the input parameters for preset times to obtain a prediction result;
fitting the loss probability prediction model based on the prediction result to obtain a trained loss probability prediction model;
and calling the trained loss probability prediction model, and correcting the initial loss probability to obtain the user loss probability.
5. The user recall analysis method according to claim 1, wherein the invoking a preset clustering algorithm to analyze the target user data and the user churn probability to obtain user churn cause information, and the determining a user recall policy according to the user churn cause information comprises:
calling a preset clustering algorithm to analyze the target user data to obtain user loss reason information;
calling a preset data classification algorithm to classify the user loss reason information to obtain loss reason label information;
acquiring initial user label information when the user loss probability reaches a preset threshold value, and updating the initial user label information through the loss reason label information to obtain target user label information;
and determining a user recall strategy based on the target user label information, wherein the user recall strategy comprises issuing coupons and pushing messages.
6. The user recall analysis method of any of claims 1-5 wherein the obtaining user churn data and user recall results based on the user recall policy, and the evaluating and optimizing the user recall policy based on the user churn data and the user recall results, and the obtaining an optimized user recall policy comprises:
acquiring user loss data and user recall results based on the user recall strategy, and monitoring the user loss data and the user recall results in real time within a preset time period to obtain monitoring results, wherein the user recall results are corresponding user recall numbers after the user recall strategy is implemented, and the user recall results are used for indicating user recall effects of the user recall strategy;
analyzing the monitoring result to obtain an analysis result, calling a preset grouping test algorithm, and evaluating and optimizing the user recall strategy based on the analysis result to obtain an optimized user recall strategy.
7. The user recall analysis method of claim 6, wherein the analyzing the monitoring results to obtain analysis results, invoking a preset grouping test algorithm, and based on the analysis results, evaluating and optimizing the user recall policy, wherein obtaining an optimized user recall policy comprises:
comparing and analyzing preset conditions and the monitoring result to obtain an analysis result;
if the analysis result is that the monitoring result does not meet the preset condition, grouping the users corresponding to the initial user data through a preset grouping test algorithm to obtain a plurality of user groups, wherein each user group corresponds to a user recall strategy;
and sequencing the user loss data corresponding to each user group to obtain a sequencing result, and adjusting the user recall strategy of which the user loss data is greater than a preset threshold value in the sequencing result to obtain an optimized user recall strategy.
8. A user recall analysis apparatus, the user recall analysis apparatus comprising:
the system comprises an acquisition module, a storage module and a processing module, wherein the acquisition module is used for acquiring initial user data, preprocessing the initial user data to obtain preprocessed data, and extracting target user data from the preprocessed data, and the target user data comprises user behavior data and user emotion data;
the prediction module is used for calling a preset loss probability prediction model to predict the loss probability of the target user data to obtain the user loss probability;
the determining module is used for calling a preset clustering algorithm, analyzing the target user data and the user loss probability to obtain user loss reason information, and determining a user recall strategy according to the user loss reason information;
and the generating module is used for acquiring user loss data and user recall results based on the user recall strategy, and evaluating and optimizing the user recall strategy based on the user loss data and the user recall results to obtain the optimized user recall strategy.
9. A user recall analysis apparatus, the user recall analysis apparatus comprising:
a memory and at least one processor, the memory having instructions stored therein;
the at least one processor invoking the instructions in the memory to cause the user recall analysis device to perform the user recall analysis method of any of claims 1-7.
10. A computer-readable storage medium having instructions stored thereon, wherein the instructions, when executed by a processor, implement a user recall analysis method according to any one of claims 1-7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110692005.5A CN113421116A (en) | 2021-06-22 | 2021-06-22 | User recall analysis method, device, equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110692005.5A CN113421116A (en) | 2021-06-22 | 2021-06-22 | User recall analysis method, device, equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113421116A true CN113421116A (en) | 2021-09-21 |
Family
ID=77716050
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110692005.5A Pending CN113421116A (en) | 2021-06-22 | 2021-06-22 | User recall analysis method, device, equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113421116A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113988955A (en) * | 2021-11-26 | 2022-01-28 | 中国银行股份有限公司 | Potential asset promotion client prediction method and device |
| CN115456223A (en) * | 2022-11-09 | 2022-12-09 | 深圳市闪盾能源科技有限公司 | Lithium battery echelon recovery management method and system based on full life cycle |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109583651A (en) * | 2018-12-03 | 2019-04-05 | 焦点科技股份有限公司 | A kind of method and apparatus for insuring electric business platform user attrition prediction |
| US20200134648A1 (en) * | 2017-07-24 | 2020-04-30 | Beijing Didi Infinity Technology And Development Co., Ltd. | Methods and systems for preventing user churn |
| CN111275503A (en) * | 2020-03-20 | 2020-06-12 | 京东数字科技控股有限公司 | Data processing method and device for acquiring lost user recall success rate |
| CN112884515A (en) * | 2021-02-22 | 2021-06-01 | 上海汽车集团股份有限公司 | User loss prediction method and device and computer storage medium |
-
2021
- 2021-06-22 CN CN202110692005.5A patent/CN113421116A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20200134648A1 (en) * | 2017-07-24 | 2020-04-30 | Beijing Didi Infinity Technology And Development Co., Ltd. | Methods and systems for preventing user churn |
| CN109583651A (en) * | 2018-12-03 | 2019-04-05 | 焦点科技股份有限公司 | A kind of method and apparatus for insuring electric business platform user attrition prediction |
| CN111275503A (en) * | 2020-03-20 | 2020-06-12 | 京东数字科技控股有限公司 | Data processing method and device for acquiring lost user recall success rate |
| CN112884515A (en) * | 2021-02-22 | 2021-06-01 | 上海汽车集团股份有限公司 | User loss prediction method and device and computer storage medium |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113988955A (en) * | 2021-11-26 | 2022-01-28 | 中国银行股份有限公司 | Potential asset promotion client prediction method and device |
| CN115456223A (en) * | 2022-11-09 | 2022-12-09 | 深圳市闪盾能源科技有限公司 | Lithium battery echelon recovery management method and system based on full life cycle |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107609708B (en) | User loss prediction method and system based on mobile game shop | |
| CN109598095B (en) | Method and device for establishing scoring card model, computer equipment and storage medium | |
| CN113657545B (en) | User service data processing method, device, equipment and storage medium | |
| CN104123592B (en) | Bank's backstage TPS transaction events trend forecasting method and system | |
| CN110992167A (en) | Bank client business intention identification method and device | |
| CN113421116A (en) | User recall analysis method, device, equipment and storage medium | |
| CN111784040B (en) | Optimization method and device for policy simulation analysis and computer equipment | |
| JP2014225175A (en) | Data analyzer and health care business support method | |
| CN111738297A (en) | Feature selection method, device, equipment and storage medium | |
| CN110910991B (en) | Medical automatic image processing system | |
| Gao et al. | Diabetic retinopathy classification using an efficient convolutional neural network | |
| CN114187979A (en) | Data processing, model training, molecular prediction and screening method and device thereof | |
| CN113420018A (en) | User behavior data analysis method, device, equipment and storage medium | |
| AU2021204470A1 (en) | Benefit surrender prediction | |
| CN112217908B (en) | Information pushing method and device based on transfer learning and computer equipment | |
| CN116610821B (en) | Knowledge graph-based enterprise risk analysis method, system and storage medium | |
| TWI747290B (en) | Customer grouping service customized marketing method and system | |
| CN112308623A (en) | High-quality client loss prediction method and device based on supervised learning and storage medium | |
| CN113824580A (en) | Network index early warning method and system | |
| CN108320788A (en) | Hospital business analysis method and device | |
| CN111445969A (en) | Sales prediction method and system capable of flexibly adapting to noise | |
| CN115938600A (en) | Mental health state prediction method and system based on correlation analysis | |
| CN113627997A (en) | Data processing method and device, electronic equipment and storage medium | |
| CN115295115A (en) | Sodium valproate blood concentration prediction method and device based on deep learning | |
| CN114334062A (en) | Disease abnormity early warning method, equipment and medium based on medical history |
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 |