A terminal switching prediction method, device, and storage medium

Abstract

An embodiment of the present application provides a method, device, and storage medium for predicting terminal replacement, which relates to the field of communication technologies and solves the technical problem of low accuracy in predicting terminal replacement in the prior art. The method for predicting terminal replacement includes: obtaining first terminal information and first service information of the terminal to be predicted; determining whether the terminal to be predicted will replace the device according to the first terminal information, the first service information, and a pre-trained terminal replacement prediction model; the pre-trained terminal replacement prediction model is a model obtained by training a training terminal including the terminal to be predicted according to an unsupervised learning algorithm and a supervised learning algorithm.

Description

Technical Field

[0001] The present invention relates to the field of communication technologies, and in particular, to a method and device for predicting terminal replacement and a storage medium. Background Art

[0002] With the development of China's mobile network and the popularization of the fourth-generation mobile communication technology (4G) and the fifth-generation mobile communication technology (5G), the smartphone market has developed rapidly.

[0003] The market share of mobile phone terminal replacement has now become huge, and the replacement frequency of users has increased significantly. How operators use the massive user service data they have to accurately market mobile phones to users has become the strategic center for operators to expand the market.

[0004] Existing methods for predicting terminal replacement usually predict whether a user will replace the terminal at a certain future time based on the user's historical call behavior and traffic behavior data. However, according to the market research results, there is a weak correlation between the user's historical call behavior and traffic behavior and whether the user will replace the terminal at a certain future time, resulting in low accuracy of terminal replacement prediction. Summary of the Invention

[0005] Embodiments of the present application provide a method and device for predicting terminal replacement and a storage medium, which solve the technical problem of low accuracy of existing terminal replacement prediction.

[0006] To achieve the above object, the embodiments of the present application adopt the following technical solutions:

[0007] In a first aspect, a method for predicting terminal replacement is provided, including:

[0008] Obtaining first terminal information and first service information of a terminal to be predicted;

[0009] Determining whether the terminal to be predicted will replace the terminal according to the first terminal information, the first service information, and a pre-trained terminal replacement prediction model; the pre-trained terminal replacement prediction model is a model obtained by training a training terminal including the terminal to be predicted according to an unsupervised learning algorithm and a supervised learning algorithm.

[0010] Optionally, the terminal replacement prediction further includes:

[0011] Obtaining second terminal information and second service information of a training terminal; the training terminal includes a target type terminal and a non-target type terminal;

[0012] Determine the first feature data of the training terminal; the first feature data includes the feature data of the second terminal information and the feature data of the second service information; the first feature data includes the feature data of the target type terminals and the feature data of the non-target type terminals;

[0013] Train multiple unsupervised clustering learning models according to the unsupervised learning algorithm and the first feature data;

[0014] Based on the first feature data and multiple unsupervised clustering learning models, determine the non-target type terminals that meet the preset conditions;

[0015] Remove the feature data of the non-target type terminals that meet the preset conditions from the first feature data to obtain the second feature data;

[0016] Train a supervised classification learning model according to the supervised learning algorithm and the second feature data, and determine the supervised classification learning model as the terminal replacement prediction model.

[0017] Optionally, the first feature data includes: labeled feature data, continuous feature data, and discrete feature data;

[0018] Determine the first feature data of the training terminal, including:

[0019] Remove the abnormal data in the second terminal information and the second service information to obtain the data to be processed; the data to be processed includes the labeled data to be processed, the continuous data to be processed, and the discrete data to be processed;

[0020] Perform feature engineering processing on the data to be processed to obtain the feature data of the data to be processed; the feature data of the data to be processed includes: labeled feature data, the continuous feature data to be processed, and the discrete feature data to be processed;

[0021] Perform normalization processing on the continuous feature data to be processed to obtain continuous feature data;

[0022] Perform dummy variable conversion processing on the discrete feature data to be processed to obtain discrete feature data.

[0023] Optionally, the labeled feature data includes the network type feature data of the training terminal; the multiple unsupervised clustering learning models include multiple classification clusters;

[0024] Based on the first feature data and multiple unsupervised clustering learning models, determine the non-target type terminals that meet the preset conditions, including:

[0025] According to the network type feature data, determine the first proportion of the target type terminals in the training terminal;

[0026] Determine the second proportion of the target class terminals in each classification cluster among multiple classification clusters;

[0027] From the second proportions, select the classification clusters corresponding to the second proportions that are greater than the first proportion and determine them as the first classification cluster set;

[0028] Determine the first non-target class terminal set according to the network type feature data;

[0029] From the first classification cluster set, select the classification clusters to which each non-target class terminal in the first non-target class terminal set belongs to obtain the second classification cluster set;

[0030] Sum the second proportions corresponding to each classification cluster in the second classification cluster set to obtain the heterogeneous target value of each non-target class terminal;

[0031] From the heterogeneous target values of each non-target class terminal, select the non-target class terminals corresponding to the heterogeneous target values that are greater than the preset heterogeneous target value to obtain the second non-target class terminal set;

[0032] Determine the third non-target class terminal set according to the second non-target class terminal set, the first proportion, and the preset removal threshold, and determine the non-target class terminals in the third non-target class terminal set as the non-target class terminals that meet the preset conditions.

[0033] Optionally, determining the third non-target class terminal set according to the second non-target class terminal set, the first proportion, and the preset removal threshold includes:

[0034] Determine the numerical set; the numerical set includes: the first numerical value, the second numerical value, and the third numerical value; the first numerical value is the number of non-target class terminals in the second non-target class terminal set; the second numerical value is the product of the number of non-target class terminals in the first non-target class terminal set and the first proportion; the third numerical value is the product of the number of non-target class terminals in the first non-target class terminal set and the preset removal threshold;

[0035] When the first numerical value is the smallest numerical value in the numerical set, determine the second non-target class terminal set as the third non-target class terminal set;

[0036] When the second numerical value is the smallest numerical value in the numerical set, determine the product of the number of non-target class terminals in the second non-target class terminal set and the reciprocal of the second numerical value as the first value to be removed, and from the second non-target class terminal set, select the non-target class terminals of the first value to be removed in descending order of the heterogeneous target value to determine the third non-target class terminal set;

[0037] When the third value is the minimum value in the value set, the product of the number of non-target terminals in the second non-target terminal set and the reciprocal of the third value is determined as the second value to be removed, and non-target terminals with the second value to be removed are selected from the second non-target terminal set in descending order of heterogeneous target values to determine the third non-target terminal set.

[0038] Optionally, determining whether a terminal to be predicted will change devices according to the first terminal information, the first service information, and a pre-trained terminal device change prediction model includes:

[0039] Determining the third feature data of the terminal to be predicted; the third feature data includes the feature data of the first terminal information and the feature data of the first service information;

[0040] Selecting target feature data from the third feature data; the target feature data includes the continuous feature data and discrete feature data of the terminal to be predicted;

[0041] Inputting the target feature data into the terminal device change prediction model to obtain an initial probability value;

[0042] When the initial probability value is greater than the preset probability value, determining whether the terminal to be predicted will change devices in the target time period according to the initial probability value and the third feature data.

[0043] Optionally, determining whether the terminal to be predicted will change devices in the target time period according to the initial probability value and the third feature data includes:

[0044] Determining a target probability value according to the initial probability value and the third feature data;

[0045] The initial probability value, the third feature data, and the target probability value satisfy the following formula:

[0046]

[0047] Where Y1 is the target probability value, Y2 is the initial probability value, a is the feature data of the number of days the terminal to be predicted has been used; b is the number of terminals the user corresponding to the terminal to be predicted has owned within n days; n is a positive integer;

[0048] When the target probability value is greater than or equal to the preset probability value, determining that the terminal to be predicted will change devices in the target time period;

[0049] When the target probability value is less than the preset probability value, determining that the terminal to be predicted will not change devices in the target time period.

[0050] In a second aspect, a terminal device change prediction device is provided, and the terminal device change prediction device includes: an acquisition unit and a processing unit;

[0051] An acquisition unit for acquiring first terminal information and first service information of a terminal to be predicted;

[0052] A processing unit for determining whether the terminal to be predicted will change devices according to the first terminal information, the first service information, and a pre-trained terminal device change prediction model; the pre-trained terminal device change prediction model is a model obtained by training training terminals including the terminal to be predicted according to an unsupervised learning algorithm and a supervised learning algorithm.

[0053] Optionally, the acquisition unit is further configured to acquire second terminal information and second service information of the training terminals; the training terminals include target type terminals and non-target type terminals;

[0054] The processing unit is further configured to determine first feature data of the training terminals; the first feature data includes feature data of the second terminal information and feature data of the second service information; the first feature data includes feature data of target type terminals and feature data of non-target type terminals;

[0055] The processing unit is further configured to train a plurality of unsupervised clustering learning models according to the unsupervised learning algorithm and the first feature data;

[0056] The processing unit is further configured to determine non-target type terminals that meet a preset condition based on the first feature data and the plurality of unsupervised clustering learning models;

[0057] The processing unit is further configured to remove the feature data of the non-target type terminals that meet the preset condition from the first feature data to obtain second feature data;

[0058] The processing unit is further configured to train a supervised classification learning model according to the supervised learning algorithm and the second feature data, and determine the supervised classification learning model as the terminal device change prediction model.

[0059] Optionally, the first feature data includes: labeled feature data, continuous feature data, and discrete feature data;

[0060] The processing unit is specifically configured to:

[0061] Remove abnormal data from the second terminal information and the second service information to obtain data to be processed; the data to be processed includes labeled data to be processed, continuous data to be processed, and discrete data to be processed;

[0062] Perform feature engineering processing on the data to be processed to obtain feature data of the data to be processed; the feature data of the data to be processed includes: labeled feature data, continuous feature data to be processed, and discrete feature data to be processed;

[0063] Perform normalization processing on the continuous feature data to be processed to obtain continuous feature data;

[0064] Perform dummy variable conversion processing on the discrete feature data to be processed to obtain discrete feature data.

[0065] Optionally, the label-type feature data includes network type feature data of the training terminal; the multiple unsupervised clustering learning models include multiple classification clusters;

[0066] The processing unit is specifically configured to:

[0067] Determine the first proportion of the target class terminals in the training terminal according to the network type feature data;

[0068] Determine the second proportion of the target class terminals in each of the multiple classification clusters;

[0069] Select, from the second proportions, the classification clusters corresponding to the second proportions that are greater than the first proportion and determine them as the first classification cluster set;

[0070] Determine the first non-target class terminal set according to the network type feature data;

[0071] Select, from the first classification cluster set, the classification clusters to which each non-target class terminal in the first non-target class terminal set belongs to obtain the second classification cluster set;

[0072] Sum the second proportions corresponding to each classification cluster in the second classification cluster set to obtain the heterogeneous target values of each non-target class terminal;

[0073] Select, from the heterogeneous target values of each non-target class terminal, the non-target class terminals corresponding to the heterogeneous target values that are greater than the preset heterogeneous target value to obtain the second non-target class terminal set;

[0074] Determine the third non-target class terminal set according to the second non-target class terminal set, the first proportion, and the preset removal threshold, and determine the non-target class terminals in the third non-target class terminal set as the non-target class terminals that meet the preset conditions.

[0075] Optionally, the processing unit is specifically configured to:

[0076] Determine a numerical set; the numerical set includes: a first numerical value, a second numerical value, and a third numerical value; the first numerical value is the number of non-target class terminals in the second non-target class terminal set; the second numerical value is the product of the number of non-target class terminals in the first non-target class terminal set and the first proportion; the third numerical value is the product of the number of non-target class terminals in the first non-target class terminal set and the preset removal threshold;

[0077] When the first numerical value is the smallest numerical value in the numerical set, determine the second non-target class terminal set as the third non-target class terminal set;

[0078] When the second value is the minimum value in the value set, the product of the number of non-target class terminals in the second non-target class terminal set and the reciprocal of the second value is determined as the first value to be removed, and from the second non-target class terminal set, non-target class terminals corresponding to the first value to be removed are selected in descending order of the heterogeneous target value to determine the third non-target class terminal set;

[0079] When the third value is the minimum value in the value set, the product of the number of non-target class terminals in the second non-target class terminal set and the reciprocal of the third value is determined as the second value to be removed, and from the second non-target class terminal set, non-target class terminals corresponding to the second value to be removed are selected in descending order of the heterogeneous target value to determine the third non-target class terminal set.

[0080] Optionally, the processing unit is specifically configured to:

[0081] Determine the third feature data of the terminal to be predicted; the third feature data includes the feature data of the first terminal information and the feature data of the first service information;

[0082] Select target feature data from the third feature data; the target feature data includes the continuous feature data and discrete feature data of the terminal to be predicted;

[0083] Input the target feature data into the terminal replacement prediction model to obtain an initial probability value;

[0084] When the initial probability value is greater than the preset probability value, determine whether the terminal to be predicted will be replaced during the target time period according to the initial probability value and the third feature data.

[0085] Optionally, the processing unit is specifically configured to:

[0086] Determine a target probability value according to the initial probability value and the third feature data;

[0087] The initial probability value, the third feature data, and the target probability value satisfy the following formula:

[0088]

[0089] Wherein, Y1 is the target probability value, Y2 is the initial probability value, a is the feature data of the number of days the terminal to be predicted has been used; b is the number of terminals the user corresponding to the terminal to be predicted has owned within n days; n is a positive integer;

[0090] When the target probability value is greater than or equal to the preset probability value, determine that the terminal to be predicted will be replaced during the target time period;

[0091] When the target probability value is less than the preset probability value, determine that the terminal to be predicted will not be replaced during the target time period.

[0092] In a third aspect, a terminal replacement prediction device is provided, including a memory and a processor. The memory is used to store computer-executable instructions, and the processor is connected to the memory through a bus. When the terminal replacement prediction device runs, the processor executes the computer-executable instructions stored in the memory, so that the terminal replacement prediction device executes the terminal replacement prediction method described in the first aspect.

[0093] The terminal replacement prediction device may be a network device or a part of the network device, such as a chip system in the network device. The chip system is used to support the network device to implement the functions involved in the first aspect and any possible implementation manner thereof. For example, it receives, determines, and shunts the data and / or information involved in the above terminal replacement prediction method. The chip system includes chips and may also include other discrete devices or circuit structures.

[0094] In a fourth aspect, a computer-readable storage medium is provided. The computer-readable storage medium includes computer-executable instructions. When the computer-executable instructions run on a computer, the computer executes the terminal replacement prediction method described in the first aspect.

[0095] In a fifth aspect, a computer program product is provided. When the computer program product runs on a computer, the computer executes the terminal replacement prediction method described in the first aspect and any possible design manner thereof.

[0096] It should be noted that the above computer instructions may be stored in whole or in part on the first computer storage medium. Among them, the first computer storage medium may be packaged together with the processor of the terminal replacement prediction device or separately packaged with the processor of the terminal replacement prediction device. The embodiments of the present application do not make any limitations in this regard.

[0097] The descriptions of the second aspect, the third aspect, the fourth aspect, and the fifth aspect in the present invention may refer to the detailed description of the first aspect; and the beneficial effects of the second aspect, the third aspect, the fourth aspect, and the fifth aspect may refer to the analysis of the beneficial effects of the first aspect, which will not be elaborated here.

[0098] In the embodiments of the present application, the name of the above terminal replacement prediction device does not limit the device or functional module itself. In actual implementation, these devices or functional modules may appear under other names. As long as the functions of each device or functional module are similar to those of the present invention and fall within the scope of the claims of the present invention and equivalent technologies.

[0099] These aspects or other aspects of the present invention will be more clearly understood in the following description.

[0100] The technical solutions provided by the present application at least bring the following beneficial effects:

[0101] In this application, after obtaining the first terminal information and the first service information of the terminal to be predicted, the terminal replacement prediction device can determine whether the terminal to be predicted will be replaced by using the first terminal information, the first service information, and a pre-trained terminal replacement prediction model. Since the pre-trained terminal replacement prediction model is obtained by training a model on training terminals including the terminal to be predicted according to unsupervised learning algorithms and supervised learning algorithms, the terminal replacement prediction device can analyze the matching degree between the overall requirements of the terminal to be predicted (such as service information, terminal information, etc.) and the terminal currently actually used by the terminal to be predicted, so as to predict the replacement demand of the terminal to be predicted, and thus quickly, accurately, and reasonably predict whether the terminal will be replaced, solving the technical problem of low accuracy in existing terminal replacement prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0102] Figure 1 It is a schematic hardware structure diagram of a terminal replacement prediction device provided by an embodiment of this application;

[0103] Figure 2 It is a schematic hardware structure diagram of another terminal replacement prediction device provided by an embodiment of this application;

[0104] Figure 3 It is a schematic flowchart of a process for training a terminal replacement prediction model provided by an embodiment of this application;

[0105] Figure 4 It is a schematic flowchart of a terminal replacement prediction method provided by an embodiment of this application;

[0106] Figure 5 It is a schematic diagram of the structure of a terminal replacement prediction device provided by an embodiment of this application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0107] Next, the technical solutions in the embodiments of this application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of this application. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0108] It should be noted that in the embodiments of this application, words such as "exemplary" or "for example" are used to represent examples, illustrations, or explanations. Any embodiment or design solution described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or having more advantages than other embodiments or design solutions. Exactly speaking, using words such as "exemplary" or "for example" is intended to present relevant concepts in a specific manner.

[0109] To facilitate a clear description of the technical solutions of the embodiments of the present application, in the embodiments of the present application, terms such as "first" and "second" are used to distinguish identical or similar items with basically the same functions and roles. Those skilled in the art can understand that terms such as "first" and "second" do not limit the quantity and execution order.

[0110] As described in the background art, existing terminal replacement prediction methods usually predict whether a user will replace a terminal at a certain future time based on the user's historical call behavior and traffic behavior data. However, according to the market research results, the relationship between the user's historical call behavior and traffic behavior and whether the user will replace a terminal at a certain future time is weakly correlated, resulting in low accuracy of terminal replacement prediction.

[0111] To address the above problems, the embodiments of the present application provide a terminal replacement prediction method. After obtaining the first terminal information and the first service information of the terminal to be predicted, the terminal replacement prediction device can determine whether the terminal to be predicted will be replaced through the first terminal information, the first service information, and a pre-trained terminal replacement prediction model. Since the pre-trained terminal replacement prediction model is obtained by training a model for training terminals including the terminal to be predicted according to unsupervised learning algorithms and supervised learning algorithms, the terminal replacement prediction device can analyze the matching degree between the overall requirements (such as service information, terminal information, etc.) of the terminal to be predicted and the terminal currently actually used by the terminal to be predicted, thereby predicting the replacement demand of the terminal to be predicted, and thus quickly, accurately, and reasonably predicting whether the terminal will be replaced, solving the technical problem of low accuracy of existing terminal replacement prediction.

[0112] The above terminal replacement prediction device can be a device for predicting the performance of the device and line corresponding to the target port, or a chip in the device, or a system on chip in the device.

[0113] Optionally, the device can be a physical machine, such as: a desktop computer, also known as a desktop or desktop computer (desktop computer), a mobile phone, a tablet computer, a laptop computer, an ultra-mobile personal computer (ultra-mobile personal computer, UMPC), a netbook, a personal digital assistant (personal digital assistant, PDA), and other terminal devices.

[0114] Optionally, the above terminal replacement prediction device can also implement the functions to be achieved by the above terminal replacement prediction device through a virtual machine (virtual machine, VM) deployed on a physical machine.

[0115] For ease of understanding, the structure of the terminal replacement prediction device in the embodiments of the present application will be described below.

[0116] Figure 1 FIG. 1 shows a schematic diagram of a hardware structure of the terminal replacement prediction device provided in the embodiments of the present application. As Figure 1 shown, the terminal replacement prediction device includes a processor 11, a memory 12, a communication interface 13, and a bus 14. The processor 11, the memory 12, and the communication interface 13 can be connected through the bus 14.

[0117] The processor 11 is the control center of the terminal replacement prediction device, which can be a single processor or a collective term for multiple processing elements. For example, the processor 11 can be a general-purpose central processing unit (CPU), or other general-purpose processors. Among them, the general-purpose processor can be a microprocessor or any conventional processor, etc.

[0118] As an embodiment, the processor 11 can include one or more CPUs, such as Figure 1 the CPUs 0 and 1 shown in FIG.

[0119] The memory 12 can be a read-only memory (ROM) or other types of static storage devices that can store static information and instructions, a random access memory (RAM) or other types of dynamic storage devices that can store information and instructions, or an electrically erasable programmable read-only memory (EEPROM), a disk storage medium, or other magnetic storage devices, or any other medium that can be used to carry or store desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited thereto.

[0120] In a possible implementation, the memory 12 can exist independently of the processor 11. The memory 12 can be connected to the processor 11 through the bus 14 for storing instructions or program code. When the processor 11 calls and executes the instructions or program code stored in the memory 12, the terminal replacement prediction method provided in the embodiments of the present invention can be implemented.

[0121] In another possible implementation, the memory 12 can also be integrated with the processor 11.

[0122] A communication interface 13 is used to connect to other devices via a communication network. The communication network can be an Ethernet network, a radio access network, a wireless local area network (WLAN), etc. The communication interface 13 can include a receiving unit for receiving data and a transmitting unit for transmitting data.

[0123] A bus 14 can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, an Extended Industry Standard Architecture (EISA) bus, etc. This bus can be divided into an address bus, a data bus, a control bus, etc. For the sake of simplicity in representation, Figure 1 it is represented by only one thick line in the figure, but this does not mean there is only one bus or one type of bus.

[0124] It should be noted that Figure 1 the structure shown does not constitute a limitation on the terminal replacement prediction device. In addition to Figure 1 the components shown, the terminal replacement prediction device can include more or fewer components than those shown in the figure, or combine certain components, or have a different component layout.

[0125] Figure 2 Another hardware structure of the terminal replacement prediction device in an embodiment of the present application is shown. As Figure 2 shown, the communication device can include a processor 21 and a communication interface 22. The processor 21 is coupled to the communication interface 22.

[0126] The functions of the processor 21 can refer to the description of the above-mentioned processor 11. In addition, the processor 21 also has a storage function, which can refer to the function of the above-mentioned memory 12.

[0127] The communication interface 22 is used to provide data for the processor 21. This communication interface 22 can be an internal interface of the communication device or an external interface of the terminal replacement prediction device (equivalent to the communication interface 13).

[0128] It should be noted that Figure 1 (or Figure 2 ) the structure shown does not constitute a limitation on the terminal replacement prediction device. In addition to Figure 1 (or Figure 2 ) the components shown, the terminal replacement prediction device can include more or fewer components than those shown in the figure, or combine certain components, or have a different component layout.

[0129] The following provides a detailed introduction to the terminal replacement prediction method provided by the embodiments of the present application in conjunction with the accompanying drawings.

[0130] The terminal replacement prediction method provided by the embodiments of the present application includes: the terminal replacement prediction device trains a terminal replacement prediction model according to the feature data of the training terminal and a preset algorithm (simply referred to as the "terminal replacement prediction model training process"), and the process by which the terminal replacement prediction device determines whether the terminal to be predicted will be replaced (simply referred to as the "terminal replacement prediction process").

[0131] The following first describes the "terminal replacement prediction model training process".

[0132] As Figure 3 shown, the "terminal replacement prediction model training process" includes:

[0133] S301. The terminal replacement prediction device obtains the second terminal information and the second service information of the training terminal.

[0134] Among them, the training terminal includes a target type terminal and a non-target type terminal.

[0135] Specifically, when the terminal replacement prediction device trains a terminal replacement prediction model, a large amount of training data is required as the training set and the test set. Therefore, the terminal replacement prediction device can obtain the second terminal information and the second service information of the training terminal.

[0136] Optionally, the second terminal information of the training terminal may be the B-domain (business support system) operator data obtained by the terminal replacement prediction device from the operator database, and the second service information of the training terminal may be the O-domain (operation support system) operator data obtained by the terminal replacement prediction device from the operator database.

[0137] The O domain (operation domain), B domain (business domain), and M domain (management domain) specifically refer to the three major data domains in the field of big data in the telecommunications industry.

[0138] The B-domain operator data includes user data, such as user consumption habits, terminal information, ARPU grouping, service content, business audience groups, etc. It is mainly to build some business support systems to ensure that telecom operators can normally support their services.

[0139] The O-domain operator data includes service data, such as signaling, alarms, faults, network resources, etc. It is mainly for the construction of business support systems related to network-side signaling, activation instructions, management of network resource devices, resource usage, etc.

[0140] In the embodiments of the present application, the B-domain operator data is the billing-side data of users, and the data content mainly includes the natural attribute information of users (such as gender, age, etc.), billing-related information (such as package information, arrears information, etc.), and terminal information (such as the current terminal brand in use, historical terminal replacement records, etc.). The O-domain operator data is the data collected from the core network side of users, and the data content mainly includes the Internet usage behavior of users (such as the traffic used by APPs, the browsing duration of APPs, etc.) and call behavior (such as call duration, number of calls, etc.).

[0141] Optionally, when the terminal replacement prediction device obtains the second terminal information and the second service information of the training terminal, it can obtain them from the B-domain data platform and the O-domain data platform through the Secret File Transfer Protocol (SFTP).

[0142] It should be noted that the terminal replacement prediction device can also obtain the second terminal information and the second service information of the training terminal through other data transmission methods, and the embodiments of the present application do not limit this.

[0143] Optionally, after obtaining the second terminal information and the second service information of the training terminal, the terminal replacement prediction device can also enter the obtained data into a Hadoop cluster.

[0144] It should be noted that the terminal replacement prediction device can also store the second terminal information and the second service information of the training terminal through other data storage methods, and the embodiments of the present application do not limit this.

[0145] Optionally, the second terminal information and the second service information of the training terminal can be the terminal information and service information within a preset time period.

[0146] Exemplarily, the terminal replacement prediction device can obtain the second terminal information and the second service information within the time period starting from date t 0 , and ending at date t 1 .

[0147] In practical applications, t 1 can be the current date, and t 1 -t 0 ≥7.

[0148] S302. The terminal replacement prediction device determines the first feature data of the training terminal.

[0149] Specifically, after obtaining the second terminal information and the second service information of the training terminal, in order to facilitate subsequent rapid training to obtain a model, the terminal replacement prediction device can determine the first feature data of the training terminal.

[0150] Among them, the first feature data includes the feature data of the second terminal information and the feature data of the second service information; the first feature data includes the feature data of the target type terminal and the feature data of the non-target type terminal.

[0151] The target type terminal and the non-target type terminal are preset category terminals.

[0152] Optionally, the target type terminal may be a 5G terminal. In actual applications, the target type terminal may also define different target type terminals according to scenarios and requirements (such as high game performance terminals, high camera function terminals, etc.).

[0153] Optionally, the method for the terminal replacement prediction device to determine the first feature data of the training terminal specifically includes:

[0154] S3021. The terminal replacement prediction device removes abnormal data in the second terminal information and the second service information to obtain data to be processed.

[0155] Among them, the data to be processed includes labeled data to be processed, continuous data to be processed, and discrete data to be processed.

[0156] Optionally, the abnormal data includes: non-standard user data, user data with feature filling as null values, etc.

[0157] Optionally, the terminal replacement prediction device can remove continuous abnormal data in the second terminal information and the second service information through the three-sigma rule of thumb to obtain data to be processed.

[0158] S3022. The terminal replacement prediction device performs feature engineering processing on the data to be processed to obtain the feature data of the data to be processed.

[0159] Among them, the feature data of the data to be processed includes: labeled feature data, continuous feature data to be processed, and discrete feature data to be processed.

[0160] S3023. The terminal replacement prediction device performs normalization processing on the continuous feature data to be processed to obtain continuous feature data.

[0161] The continuous feature data to be processed and the continuous feature data satisfy the following formula:

[0162]

[0163] Among them, x i is the continuous feature data to be processed, x j is the continuous feature data, μ xi is the mathematical expectation of the continuous feature data to be processed, σxi is the standard deviation of the continuous feature data to be processed.

[0164] S3024. The terminal switching prediction device performs dummy variable conversion processing on the discrete feature data to be processed to obtain discrete feature data.

[0165] Exemplarily, for the discrete feature data X{a, b, c, d} to be processed, dummy variable conversion processing can be performed on it to obtain discrete feature data: X a ={1, 0}, X b ={1, 0}, X c ={1, 0}.

[0166] Exemplarily, the feature data and feature explanations of the second terminal information are shown in Table 1.

[0167] It should be noted that the users in the following tables are users who hold training terminals.

[0168] Table 1

[0169]

[0170]

[0171] Exemplarily, the feature data and feature explanations of the second service information are shown in Table 2.

[0172] Table 2

[0173] Second service information Feature explanation <![CDATA[X 14 > User's mobile phone number <![CDATA[X 15 > User's age <![CDATA[X 16 > User's gender <![CDATA[X 17 > Price of user's current package <![CDATA[X 18 > Network type of user's current package <![CDATA[X 19 > Total overdue fees of user <![CDATA[X 20 > Total number of terminals owned by user in the past n days <![CDATA[X 21 > Brand of user's current terminal <![CDATA[X 22 > Price range of user's current terminal <![CDATA[X 23 > Usage days of user's current terminal <![CDATA[X 24 > Network type of user's current terminal

[0174] In addition, in combination with the example in S301, the feature data in Table 2 can be constructed only based on the current date t1 without considering the start date t0.

[0175] Optionally, after determining the first feature data including the feature data of the second terminal information and the feature data of the second service information, the terminal switching prediction device can fuse the feature data of the second terminal information and the feature data of the second service information, and use X 0 、X 14 as the key value, and associate and merge the feature data of the second terminal information and the feature data of the second service information by means of an inner join. The key value finally retains X 0 .

[0176] Optionally, after determining the first feature data, the first feature data can be divided into three types of feature data, specifically including: labeled feature data, continuous feature data, and discrete feature data.

[0177] Exemplarily, the labeled feature data and feature explanations are shown in Table 3.

[0178] Table 3

[0179]

[0180]

[0181] The continuous feature data and feature explanations are shown in Table 4.

[0182] Table 4

[0183] Continuous feature data Feature explanation <![CDATA[X 1 > Total days of user's use of social APPs <![CDATA[X 2 > Total duration of user's use of social APPs <![CDATA[X 3 > Total traffic of user's use of social APPs <![CDATA[X 4 > Total days of user's use of game APPs <![CDATA[X 5 > Total duration of user's use of game APPs <![CDATA[X 6 > Total traffic of user's use of game APPs <![CDATA[X 7 > Total days of user's use of video APPs <![CDATA[X 8 > Total duration of user's use of video APPs <![CDATA[X 9 > Total traffic of user's use of video APPs <![CDATA[X 10 > Total days of user's use of shopping APPs <![CDATA[X 11 > Total duration of user's use of shopping APPs <![CDATA[X 12 > Total traffic of user's use of shopping APPs <![CDATA[X 13 > Total number of APPs used by user <![CDATA[X 15 > User's age <![CDATA[X 17 > Price of user's current package <![CDATA[X 19 > Total overdue fees of user <![CDATA[X 20 > Total number of terminals owned by user in the past n days

[0184] The discrete feature data and feature explanations are shown in Table 5.

[0185] Table 5

[0186]

[0187]

[0188] S303. The terminal replacement prediction device trains multiple unsupervised clustering learning models according to the unsupervised learning algorithm and the first feature data.

[0189] Specifically, after determining the first feature data of the training terminal, the terminal replacement prediction device can train multiple unsupervised clustering learning models according to the unsupervised learning algorithm and the first feature data.

[0190] Optionally, the unsupervised learning algorithm includes: K-Means algorithm, Hierarchical algorithm, Spectral algorithm, Agglomerative algorithm, DBSCAN algorithm, Fuzzy C-Means algorithm, Mean Shift algorithm, GMM algorithm, etc.

[0191] Exemplarily, taking the unsupervised learning algorithm as the K-Means algorithm as an example, the terminal replacement prediction device can use the continuous feature data and discrete feature data in the first feature data as the training set for training the unsupervised clustering learning model.

[0192] Specifically, the terminal replacement prediction device can set a range of k values (cluster number parameter).

[0193] The value range of k is generally 2 ≤ k ≤ the amount of training set data. In practical applications, the k value range is generally set to [3, 10]. The following continues to illustrate with k = [3, 10] as an example.

[0194] The terminal replacement prediction device can set the K-Means model parameters, specifically:

[0195] 1) Clusters: 3, 4, …, 10 (The number of clusters k is 3, 4, …, 10).

[0196] 2) Init: k-means++ (The initial cluster centroid algorithm uses K-Means++).

[0197] 3) n_init: 10 (The initial cluster centroid algorithm runs 10 times, and the best centroid group is selected according to inertia).

[0198] 4) max_iter: 2000 (The maximum number of iterations of the model is 2000).

[0199] In practical applications, the above parameters can be adjusted according to the scenario and requirements.

[0200] Subsequently, the terminal replacement prediction device trains 8 K-Means models with k = 3, 4, …, 10 respectively, and a total of 3 + 4 + … + 10 = 52 classification clusters are clustered, and are represented by C i , i = 1, 2, …, 52.

[0201] It should be noted that since 8 unsupervised clustering learning models are trained, each training terminal will appear in 8 different C i .

[0202] S304. The terminal replacement prediction device determines non-target class terminals that meet the preset conditions based on the first feature data and multiple unsupervised clustering learning models.

[0203] Specifically, after training multiple unsupervised clustering learning models, the terminal replacement prediction device can, according to the model output results of the multiple unsupervised clustering learning models, by determining the outlier target value, eliminate non-target class terminals (i.e., non-target class terminals that meet the preset conditions) that are repeatedly clustered into the high-density target class terminal classification clusters from the first feature data.

[0204] Optionally, the labeled feature data includes the network type feature data of the training terminal; the multiple unsupervised clustering learning models include multiple classification clusters; the method for the terminal replacement prediction device to determine non-target class terminals that meet the preset conditions based on the first feature data and multiple unsupervised clustering learning models specifically includes:

[0205] S3041. The terminal replacement prediction device determines the first proportion of target class terminals among the training terminals according to the network type feature data.

[0206] Exemplarily, taking the target class terminal as a 5G terminal and the non-target class terminal as a non-5G terminal as an example. Combining the above Table 3, through the labeled feature data X 24 (X24 = 1 indicates that the training terminal is a 5G terminal; X 24 = 0 indicates that the training terminal is a non-5G terminal), calculate the proportion of target-class terminals (5G terminals) in the first feature data to obtain P α = p(x 24 = 1).

[0207] S3042. The terminal replacement prediction device determines the second proportion of the target-class terminals in each classification cluster among multiple classification clusters.

[0208] Combined with the above example, the terminal replacement prediction device can calculate the proportion of target-class terminals (5G terminals) in each classification cluster C i to obtain

[0209] S3043. The terminal replacement prediction device selects, from the second proportions, the classification clusters corresponding to the second proportions that are greater than the first proportion and determines them as the first classification cluster set.

[0210] Combined with the above example, the terminal replacement prediction device can screen out higher than P α of the classification cluster C i to obtain the first classification cluster set

[0211] S3044. The terminal replacement prediction device determines the first non-target-class terminal set according to the network type feature data.

[0212] Combined with the above example, the terminal replacement prediction device can screen out non-target-class terminals (non-5G terminals) to obtain the first non-target-class terminal set X β = {x o |x 24 = 0}

[0213] S3045. The terminal replacement prediction device selects, from the first classification cluster set, the classification clusters to which each non-target-class terminal in the first non-target-class terminal set belongs to obtain the second classification cluster set.

[0214] Combined with the above example, the terminal replacement prediction device can, for each non-target-class terminal X β , in the first classification cluster set G, screen out the classification cluster C β to which X i belongs to obtain the second classification cluster set

[0215] S3046. The terminal replacement prediction device sums up the second proportions corresponding to each classification cluster in the second classification cluster set to obtain the heterogeneous target value of each non-target-class terminal.

[0216] Combined with the above example, the terminal replacement prediction device can target the second classification cluster set and sum up the classification clusters C i corresponding to to obtain the heterogeneous target value of each non-target class terminal

[0217] It should be noted that if there are x β that do not have a belonging classification cluster C in the first classification cluster set G i , then

[0218] S3047. The terminal replacement prediction device selects, from the heterogeneous target values of each non-target class terminal, the non-target class terminals corresponding to the heterogeneous target values greater than the preset heterogeneous target value to obtain the second non-target class terminal set.

[0219] Combined with the above example, the terminal replacement prediction device filters out the non-target class terminals with heterogeneous target values greater than 0, and arranges them in descending order to obtain the second non-target class terminal set

[0220] S3048. The terminal replacement prediction device determines the third non-target class terminal set according to the second non-target class terminal set, the first proportion, and the preset removal threshold, and determines the non-target class terminals in the third non-target class terminal set as the non-target class terminals that meet the preset conditions.

[0221] Optionally, the method by which the terminal replacement prediction device determines the third non-target class terminal set according to the second non-target class terminal set, the first proportion, and the preset removal threshold specifically includes:

[0222] S30481. The terminal replacement prediction device determines the numerical set.

[0223] Among them, the numerical set includes: the first numerical value, the second numerical value, and the third numerical value; the first numerical value is the number of non-target class terminals in the second non-target class terminal set; the second numerical value is the product of the number of non-target class terminals in the first non-target class terminal set and the first proportion; the third numerical value is the product of the number of non-target class terminals in the first non-target class terminal set and the preset removal threshold.

[0224] Combined with the above example, the terminal replacement prediction device defines |X β+ | = N1 (the number of non-target class terminals in the first non-target class terminal set X β ), |X β | × P α = N2, |X β | × P top = N3.

[0225] Among them, N1 is the first numerical value, N2 is the second numerical value, and N3 is the third numerical value. P top is the preset removal threshold for non-target type terminals (generally speaking, P top and P α are inversely proportional, and P top ≤0.3. In practical applications, the parameter P top ) can be adjusted according to the scenario and requirements.

[0226] S30482. When the first numerical value is the minimum numerical value in the numerical value set, the terminal switching prediction device determines the second non-target type terminal set as the third non-target type terminal set.

[0227] S30483. When the second numerical value is the minimum numerical value in the numerical value set, the terminal switching prediction device determines the product of the number of non-target type terminals in the second non-target type terminal set and the reciprocal of the second numerical value as the first value to be removed, and selects the non-target type terminals with the first value to be removed from the second non-target type terminal set in descending order of the heterogeneous target value to determine the third non-target type terminal set.

[0228] S30484. When the third numerical value is the minimum numerical value in the numerical value set, the terminal switching prediction device determines the product of the number of non-target type terminals in the second non-target type terminal set and the reciprocal of the third numerical value as the second value to be removed, and selects the non-target type terminals with the second value to be removed from the second non-target type terminal set in descending order of the heterogeneous target value to determine the third non-target type terminal set.

[0229] Combined with the above example, the third non-target type terminal set X cfd and the second non-target type terminal set X β+ , the first numerical value N1, the second numerical value N2, and the third numerical value N3 satisfy the following formula:

[0230]

[0231] S305. The terminal switching prediction device removes the feature data of non-target type terminals that meet the preset conditions from the first feature data to obtain the second feature data.

[0232] S306. The terminal switching prediction device trains a supervised classification learning model according to the supervised learning algorithm and the second feature data, and determines the supervised classification learning model as the terminal switching prediction model.

[0233] Optionally, the supervised learning algorithms include: XGBoost algorithm, Logistic Regression algorithm, Decision Trees algorithm, K-NN algorithm, SVM, Naive Bayes algorithm, Random Forest algorithm, AdaBoost algorithm, LightGBM algorithm, Neural Networks algorithm, etc.

[0234] Exemplarily, taking the XGBoost algorithm as an example of the supervised learning algorithm, the terminal replacement prediction device can use the continuous feature data and discrete feature data in the second feature data as the input feature data data2_x for training the unsupervised clustering learning model, and use the label feature data X in the second feature data 23 as the model output label data data2_y, and randomly allocate the data into a training set and a test set at a ratio of 7:3: data2_x_train, data2_y_train, data2_x_test, data2_y_test.

[0235] The terminal replacement prediction device can set the XGBoost model parameters, specifically:

[0236] 1) booster: gbtree (the model boosting method uses gradient boosting tree).

[0237] 2) objective: binary logistic (the objective function of the model is based on binary logistic regression):

[0238]

[0239]

[0240] 3) eval_metric: auc (the evaluation metric of the model uses AUC).

[0241] 4) n_estimators: 1000 (the number of model iterations is 1000).

[0242] 5) eta: 0.3 (the shrinkage step size in the model update process is 0.3).

[0243] 6) gamma: 0.5 (the minimum loss function decrease value required for node splitting in the model is 0.5).

[0244] 7) maxdepth: 6 (the maximum depth of the tree in the model is 6).

[0245] 8) min_child_weight: 1 (the sum of the sample weights of the smallest leaf nodes in the model is 1).

[0246] 9) subsample: 0.7 (The ratio of random sampling of data (rows) for each tree in the model is 0.7).

[0247] 10) colsample_bytree: 0.7 (The ratio of random sampling of features (columns) for each tree in the model is 0.7).

[0248] 11) lambda: 1 (The weight value of the L2 regularization term in the model is 1).

[0249] 12) alpha: 0 (The weight value of the L1 regularization term in the model is 0).

[0250] 13) scale_pos_weight: w (The ratio of the number of target class terminals to the number of non - target class terminals in the second feature data)

[0251]

[0252] In practical applications, the above parameters can be adjusted according to the scenario and requirements.

[0253] The terminal replacement prediction device can use the test sets data2_x_train and data2_y_train to train the XGBoost model, then evaluate the model performance through the test sets data2_x_test and data2_y_test, and obtain the supervised classification learning model when the model converges.

[0254] Next, the "terminal replacement prediction process" will be described.

[0255] After training the terminal replacement prediction model using the above method, it is possible to determine whether the terminal to be predicted will be replaced based on the first terminal information and the first service information of the terminal to be predicted, as well as the trained terminal replacement prediction model.

[0256] As Figure 4 shown, the method of the "terminal replacement prediction process" specifically includes:

[0257] S401. The terminal replacement prediction device obtains the first terminal information and the first service information of the terminal to be predicted.

[0258] The method by which the terminal replacement prediction device obtains the first terminal information and the first service information of the terminal to be predicted can specifically refer to the method in S301 by which the terminal replacement prediction device obtains the second terminal information and the second service information of the training terminal, which will not be elaborated here.

[0259] S402. The terminal replacement prediction device determines whether the terminal to be predicted will be replaced according to the first terminal information, the first service information, and the pre-trained terminal replacement prediction model.

[0260] Among them, the pre-trained terminal replacement prediction model is a model obtained by training a training terminal including the terminal to be predicted according to an unsupervised learning algorithm and a supervised learning algorithm.

[0261] Optionally, the method for the terminal replacement prediction device to determine whether the terminal to be predicted will be replaced according to the first terminal information, the first service information, and the pre-trained terminal replacement prediction model specifically includes:

[0262] S4021. The terminal replacement prediction device determines the third feature data of the terminal to be predicted.

[0263] Among them, the third feature data includes the feature data of the first terminal information and the feature data of the first service information.

[0264] For the method by which the terminal replacement prediction device determines the third feature data of the terminal to be predicted, reference can specifically be made to the method in S302 by which the terminal replacement prediction device determines the first feature data of the training terminal, which will not be elaborated here.

[0265] S4022. The terminal replacement prediction device selects target feature data from the third feature data.

[0266] Among them, the target feature data includes the continuous feature data and discrete feature data of the terminal to be predicted.

[0267] Exemplarily, the terminal replacement prediction device can screen out the continuous feature data and discrete feature data of non-target terminals (non-5G terminals) from the third feature data as the input feature data of the terminal replacement prediction model.

[0268] S4023. The terminal replacement prediction device inputs the target feature data into the terminal replacement prediction model to obtain an initial probability value.

[0269] Combined with the above example, the terminal replacement prediction device can input the continuous feature data and discrete feature data of non-target terminals (non-5G terminals) into the XGBoost model trained in S306 above to obtain the initial probability value Y of non-target terminals (non-5G terminals) being converted into target terminals (5G terminals). p .

[0270] S4024. When the initial probability value is greater than the preset probability value, the terminal replacement prediction device determines whether the terminal to be predicted will be replaced in the target time period according to the initial probability value and the third feature data.

[0271] Specifically, the terminal replacement prediction device can set a preset probability value y thr .

[0272] Exemplarily, when Y p ≥ y thr , a non-target class terminal (non-5G terminal) is defined as a potential target class terminal (potential 5G terminal), that is, the terminal to be predicted is about to replace the device.

[0273] Optionally, the terminal replacement prediction device determines whether the terminal to be predicted replaces the device in the target time period according to the initial probability value and the third feature data, including:

[0274] The terminal replacement prediction device determines the target probability value according to the initial probability value and the third feature data.

[0275] The initial probability value, the third feature data, and the target probability value satisfy the following formula:

[0276]

[0277] Where Y1 is the target probability value, Y2 is the initial probability value, a is the feature data of the number of days the terminal to be predicted has been used; b is the number of terminals the user corresponding to the terminal to be predicted has owned within n days; n is a positive integer.

[0278] It can be seen from the above formula that when the ratio of the number of days the user's current terminal has been used to the average number of days each terminal has been used in the past n days is smaller, the loss of the target probability value of the terminal to be predicted considering the time factor relative to the initial probability value is greater.

[0279] When the target probability value is greater than or equal to the preset probability value, the terminal replacement prediction device determines that the terminal to be predicted replaces the device in the target time period.

[0280] When the target probability value is less than the preset probability value, the terminal replacement prediction device determines that the terminal to be predicted does not replace the device in the target time period.

[0281] Exemplarily, combining the tag-type feature data X 23 (the number of days the user's current terminal has been used) and the continuous-type feature data X 20 (the cumulative number of terminals the user has owned in the past n days, generally n ≥ 730), according to the above formula, the target probability value Y of the terminal to be predicted considering the time factor can be obtained p_ .

[0282]

[0283] An embodiment of the present application provides a method for predicting terminal replacement. After obtaining the first terminal information and the first service information of the terminal to be predicted, the terminal replacement prediction device can determine whether the terminal to be predicted will be replaced through the first terminal information, the first service information, and a pre-trained terminal replacement prediction model. Since the pre-trained terminal replacement prediction model is obtained by training a model on training terminals including the terminal to be predicted according to unsupervised learning algorithms and supervised learning algorithms, the terminal replacement prediction device can predict the replacement demand of the terminal to be predicted by analyzing the matching degree between the overall demand of the terminal to be predicted (such as service information, terminal information, etc.) and the terminal currently actually used by the terminal to be predicted, so as to quickly, accurately, and reasonably predict whether the terminal will be replaced, and solve the technical problem of low accuracy of the existing terminal replacement prediction.

[0284] The above mainly introduces the solution provided by the embodiment of the present application from the perspective of the method. To implement the above functions, it includes the corresponding hardware structure and / or software module for executing each function. Those skilled in the art should easily realize that, in combination with the units and algorithm steps of each example described in the embodiments disclosed herein, the embodiments of the present application can be implemented in the form of hardware or a combination of hardware and computer software. Whether a certain function is executed in the form of hardware or computer software driving hardware depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present invention.

[0285] The embodiment of the present application can divide the function modules of the terminal replacement prediction device according to the above method examples. For example, each function module can be divided corresponding to each function, or two or more functions can be integrated into one processing module. The above integrated module can be implemented in the form of hardware or in the form of a software function module. Optionally, the division of modules in the embodiment of the present application is illustrative, only a logical function division, and there can be other division methods in actual implementation.

[0286] As Figure 5 shown, it is a schematic structural diagram of a terminal replacement prediction device provided by an embodiment of the present application. The terminal replacement prediction device includes: an acquisition unit 501 and a processing unit 502;

[0287] The acquisition unit 501 is configured to acquire the first terminal information and the first service information of the terminal to be predicted;

[0288] A processing unit 502, configured to determine whether a terminal to be predicted will change its device according to first terminal information, first service information, and a pre-trained terminal device change prediction model; the pre-trained terminal device change prediction model is a model obtained by training training terminals including the terminal to be predicted according to an unsupervised learning algorithm and a supervised learning algorithm.

[0289] Optionally, an acquisition unit 501 is further configured to acquire second terminal information and second service information of the training terminals; the training terminals include target type terminals and non-target type terminals;

[0290] The processing unit 502 is further configured to determine first feature data of the training terminals; the first feature data includes feature data of the second terminal information and feature data of the second service information; the first feature data includes feature data of target type terminals and feature data of non-target type terminals;

[0291] The processing unit 502 is further configured to train a plurality of unsupervised clustering learning models according to an unsupervised learning algorithm and the first feature data;

[0292] The processing unit 502 is further configured to determine non-target type terminals that meet a preset condition based on the first feature data and the plurality of unsupervised clustering learning models;

[0293] The processing unit 502 is further configured to remove the feature data of the non-target type terminals that meet the preset condition from the first feature data to obtain second feature data;

[0294] The processing unit 502 is further configured to train a supervised classification learning model according to a supervised learning algorithm and the second feature data, and determine the supervised classification learning model as the terminal device change prediction model.

[0295] Optionally, the first feature data includes: labeled feature data, continuous feature data, and discrete feature data;

[0296] The processing unit 502 is specifically configured to:

[0297] Remove abnormal data from the second terminal information and the second service information to obtain data to be processed; the data to be processed includes labeled data to be processed, continuous data to be processed, and discrete data to be processed;

[0298] Perform feature engineering processing on the data to be processed to obtain feature data of the data to be processed; the feature data of the data to be processed includes: labeled feature data, continuous feature data to be processed, and discrete feature data to be processed;

[0299] Perform normalization processing on the continuous feature data to be processed to obtain continuous feature data;

[0300] Perform dummy variable conversion processing on the discrete feature data to be processed to obtain discrete feature data.

[0301] Optionally, the labeled feature data includes network type feature data of the training terminal; the multiple unsupervised clustering learning models include multiple classification clusters;

[0302] The processing unit 502 is specifically configured to:

[0303] Determine the first proportion of the target class terminals in the training terminal according to the network type feature data;

[0304] Determine the second proportion of the target class terminals in each of the multiple classification clusters;

[0305] Select from the second proportions the second proportions corresponding to the classification clusters where the second proportion is greater than the first proportion, and determine the classification clusters as the first classification cluster set;

[0306] Determine the first non-target class terminal set according to the network type feature data;

[0307] Select from the first classification cluster set the classification clusters to which each non-target class terminal in the first non-target class terminal set belongs, to obtain the second classification cluster set;

[0308] Sum the second proportions corresponding to each classification cluster in the second classification cluster set to obtain the heterogeneous target values of each non-target class terminal;

[0309] Select from the heterogeneous target values of each non-target class terminal the non-target class terminals corresponding to the heterogeneous target values greater than the preset heterogeneous target value, to obtain the second non-target class terminal set;

[0310] Determine the third non-target class terminal set according to the second non-target class terminal set, the first proportion, and the preset removal threshold, and determine the non-target class terminals in the third non-target class terminal set as the non-target class terminals meeting the preset conditions.

[0311] Optionally, the processing unit 502 is specifically configured to:

[0312] Determine a numerical set; the numerical set includes: a first numerical value, a second numerical value, and a third numerical value; the first numerical value is the number of non-target class terminals in the second non-target class terminal set; the second numerical value is the product of the number of non-target class terminals in the first non-target class terminal set and the first proportion; the third numerical value is the product of the number of non-target class terminals in the first non-target class terminal set and the preset removal threshold;

[0313] When the first numerical value is the minimum numerical value in the numerical set, determine the second non-target class terminal set as the third non-target class terminal set;

[0314] When the second numerical value is the minimum numerical value in the numerical value set, the product of the number of non-target class terminals in the second non-target class terminal set and the reciprocal of the second numerical value is determined as the first value to be removed, and non-target class terminals corresponding to the first value to be removed are selected from the second non-target class terminal set in descending order of heterogeneous target values to determine the third non-target class terminal set;

[0315] When the third numerical value is the minimum numerical value in the numerical value set, the product of the number of non-target class terminals in the second non-target class terminal set and the reciprocal of the third numerical value is determined as the second value to be removed, and non-target class terminals corresponding to the second value to be removed are selected from the second non-target class terminal set in descending order of heterogeneous target values to determine the third non-target class terminal set.

[0316] Optionally, the processing unit 502 is specifically configured to:

[0317] Determine the third feature data of the terminal to be predicted; the third feature data includes the feature data of the first terminal information and the feature data of the first service information;

[0318] Select target feature data from the third feature data; the target feature data includes the continuous feature data and discrete feature data of the terminal to be predicted;

[0319] Input the target feature data into the terminal replacement prediction model to obtain an initial probability value;

[0320] When the initial probability value is greater than the preset probability value, determine whether the terminal to be predicted will be replaced during the target time period according to the initial probability value and the third feature data.

[0321] Optionally, the processing unit 502 is specifically configured to:

[0322] Determine a target probability value according to the initial probability value and the third feature data;

[0323] The initial probability value, the third feature data, and the target probability value satisfy the following formula:

[0324]

[0325] Wherein, Y1 is the target probability value, Y2 is the initial probability value, a is the feature data of the number of days the terminal to be predicted has been used; b is the number of terminals the user corresponding to the terminal to be predicted has owned within n days; n is a positive integer;

[0326] When the target probability value is greater than or equal to the preset probability value, determine that the terminal to be predicted will be replaced during the target time period;

[0327] When the target probability value is less than the preset probability value, determine that the terminal to be predicted will not be replaced during the target time period.

[0328] An embodiment of the present application further provides a computer-readable storage medium, which includes computer-executable instructions. When the computer-executable instructions run on a computer, the computer is caused to execute each step performed by the terminal replacement prediction device in the terminal replacement prediction method provided in the foregoing embodiment.

[0329] An embodiment of the present application further provides a computer program product, which can be directly loaded into a memory and contains software code. After being loaded and executed by a computer, the computer program product can implement each step performed by the terminal replacement prediction device in the terminal replacement prediction method provided in the foregoing embodiment.

[0330] In the foregoing embodiment, it can be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented using a software program, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer-executable instructions are loaded and executed on a computer, the processes or functions according to the embodiments of the present application are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from a website, a computer, a server, or a data center to another website, computer, server, or data center in a wired manner (such as coaxial cable, optical fiber, digital subscriber line (DSL)) or a wireless manner (such as infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or a data center that includes one or more integrated media. The available medium can be a magnetic medium (such as a floppy disk, a hard disk, a magnetic tape), an optical medium (such as a DVD), or a semiconductor medium (such as a solid state disk (SSD)), etc.

[0331] From the description of the above embodiments, those skilled in the art can clearly understand that for the convenience and simplicity of description, only the division of the above functional modules is used as an example. In actual applications, the above functions can be allocated to different functional modules according to needs, that is, the internal structure of the device is divided into different functional modules to complete all or part of the functions described above.

[0332] In several embodiments provided by the present invention, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the modules or units is only a logical function division, and there may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed. Another point is that the displayed or discussed coupling or direct coupling or communication connection between each other can be through some interfaces, and the indirect coupling or communication connection of the devices or units can be in electrical, mechanical or other forms. The units described as separate components may or may not be physically separated, and the components displayed as units can be one physical unit or multiple physical units, that is, they can be located in one place or distributed to multiple different places. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0333] In addition, each functional unit in various embodiments of the present invention can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit. The above-mentioned integrated units can be implemented in the form of hardware or in the form of software functional units. If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a readable storage medium. Based on such an understanding, the technical solution of the embodiments of this application, in essence, or the part that contributes to the prior art, or all or part of this technical solution, can be embodied in the form of a software product. This software product is stored in a storage medium and includes several instructions to enable a device (which can be a single-chip microcomputer, a chip, etc.) or a processor to execute all or part of the steps of the methods described in various embodiments of the present invention. The foregoing storage medium includes: USB flash drives, mobile hard disks, ROM, RAM, magnetic disks, or optical discs and other various media that can store program codes.

[0334] As described above, it is only the specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by those skilled in the art within the technical scope disclosed by the present invention should be covered by the protection scope of the present invention. Therefore, the protection scope of the present invention should be subject to the protection scope of the claims.

Claims

1. A method for predicting terminal replacement, characterized in that, it includes: Obtain the first terminal information and the first service information of the terminal to be predicted; Determine whether the terminal to be predicted will be replaced according to the first terminal information, the first service information, and a pre-trained terminal replacement prediction model; the pre-trained terminal replacement prediction model is a model obtained by training training terminals including the terminal to be predicted according to unsupervised learning algorithms and supervised learning algorithms; It further includes: Obtain the second terminal information and the second service information of the training terminals; the training terminals include target type terminals and non-target type terminals; Determine the first feature data of the training terminals; the first feature data includes the feature data of the second terminal information and the feature data of the second service information; the first feature data includes the feature data of the target type terminals and the feature data of the non-target type terminals; Train multiple unsupervised clustering learning models according to the unsupervised learning algorithm and the first feature data; Based on the first feature data and the multiple unsupervised clustering learning models, determine non-target type terminals that meet preset conditions; Remove the feature data of the non-target type terminals that meet the preset conditions from the first feature data to obtain second feature data; Train a supervised classification learning model according to the supervised learning algorithm and the second feature data, and determine the supervised classification learning model as the terminal replacement prediction model; Wherein, the first feature data includes: label type feature data, continuous type feature data, and discrete type feature data; The determining the first feature data of the training terminals includes: Remove abnormal data from the second terminal information and the second service information to obtain data to be processed; the data to be processed includes label type data to be processed, continuous type data to be processed, and discrete type data to be processed; Perform feature engineering processing on the data to be processed to obtain the feature data of the data to be processed; the feature data of the data to be processed includes: the label type feature data, the continuous type feature data to be processed, and the discrete type feature data to be processed; Perform normalization processing on the continuous type feature data to be processed to obtain the continuous type feature data; Perform dummy variable conversion processing on the discrete type feature data to be processed to obtain the discrete type feature data; Wherein, the label type feature data includes the network type feature data of the training terminals; the multiple unsupervised clustering learning models include multiple classification clusters; The determining, based on the first feature data and the multiple unsupervised clustering learning models, non-target type terminals that meet preset conditions includes: Determine the first proportion of the target type terminals in the training terminals according to the network type feature data; Determine the second proportion of the target type terminals in each of the multiple classification clusters; Select, from the second proportions, the classification clusters corresponding to the second proportions that are greater than the first proportion and determine them as the first classification cluster set; Determine the first non-target type terminal set according to the network type feature data; Select the classification clusters to which each non-target class terminal in the first non-target class terminal set belongs from the first classification cluster set to obtain a second classification cluster set; Sum the second proportions corresponding to each classification cluster in the second classification cluster set to obtain the heterogeneous target value of each non-target class terminal; Select the non-target class terminals corresponding to the heterogeneous target values greater than the preset heterogeneous target value from the heterogeneous target values of each non-target class terminal to obtain a second non-target class terminal set; Determine a third non-target class terminal set according to the second non-target class terminal set, the first proportion, and the preset removal threshold, and determine the non-target class terminals in the third non-target class terminal set as the non-target class terminals that meet the preset conditions.

2. The terminal replacement prediction method according to claim 1, wherein, The determining the third non-target class terminal set according to the second non-target class terminal set, the first proportion, and the preset removal threshold includes: Determine a numerical set; the numerical set includes: a first numerical value, a second numerical value, and a third numerical value; the first numerical value is the number of non-target class terminals in the second non-target class terminal set; the second numerical value is the product of the number of non-target class terminals in the first non-target class terminal set and the first proportion; the third numerical value is the product of the number of non-target class terminals in the first non-target class terminal set and the preset removal threshold; When the first numerical value is the smallest numerical value in the numerical set, determine the second non-target class terminal set as the third non-target class terminal set; When the second numerical value is the smallest numerical value in the numerical set, determine the product of the number of non-target class terminals in the second non-target class terminal set and the reciprocal of the second numerical value as the first value to be removed, and select the non-target class terminals of the first value to be removed from the second non-target class terminal set in descending order of the heterogeneous target value to determine the third non-target class terminal set; When the third numerical value is the smallest numerical value in the numerical set, determine the product of the number of non-target class terminals in the second non-target class terminal set and the reciprocal of the third numerical value as the second value to be removed, and select the non-target class terminals of the second value to be removed from the second non-target class terminal set in descending order of the heterogeneous target value to determine the third non-target class terminal set.

3. The terminal replacement prediction method according to claim 1 or 2, wherein, The determining whether the terminal to be predicted will be replaced according to the first terminal information, the first service information, and the pre-trained terminal replacement prediction model includes: Determine the third feature data of the terminal to be predicted; the third feature data includes the feature data of the first terminal information and the feature data of the first service information; Select target feature data from the third feature data; the target feature data includes the continuous feature data and discrete feature data of the terminal to be predicted; Input the target feature data into the terminal replacement prediction model to obtain an initial probability value; When the initial probability value is greater than a preset probability value, determine whether the terminal to be predicted will be replaced during the target time period according to the initial probability value and the third feature data.

4. The terminal replacement prediction method according to claim 3, characterized in that the determining whether the terminal to be predicted will be replaced during the target time period according to the initial probability value and the third feature data includes: determining a target probability value according to the initial probability value and the third feature data; The initial probability value, the third feature data, and the target probability value satisfy the following formula: where Y1 is the target probability value, Y2 is the initial probability value, a is the feature data of the usage days of the terminal to be predicted; b is the number of terminals owned by the user corresponding to the terminal to be predicted within n days; n is a positive integer; when the target probability value is greater than or equal to the preset probability value, determine that the terminal to be predicted will be replaced during the target time period; when the target probability value is less than the preset probability value, determine that the terminal to be predicted will not be replaced during the target time period.

5. A terminal replacement prediction device, characterized in that it includes: an acquisition unit and a processing unit; The acquisition unit is used to acquire the first terminal information and the first service information of the terminal to be predicted; The processing unit is used to determine whether the terminal to be predicted will be replaced according to the first terminal information, the first service information, and a pre-trained terminal replacement prediction model; the pre-trained terminal replacement prediction model is a model obtained by training training terminals including the terminal to be predicted according to unsupervised learning algorithms and supervised learning algorithms; The acquisition unit is further used to acquire the second terminal information and the second service information of the training terminal; the training terminal includes target type terminals and non-target type terminals; The processing unit is further used to determine the first feature data of the training terminal; the first feature data includes the feature data of the second terminal information and the feature data of the second service information; the first feature data includes the feature data of the target type terminals and the feature data of the non-target type terminals; The processing unit is further used to train multiple unsupervised clustering learning models according to the unsupervised learning algorithm and the first feature data; The processing unit is further used to determine non-target type terminals that meet preset conditions based on the first feature data and the multiple unsupervised clustering learning models; The processing unit is further used to remove the feature data of the non-target type terminals that meet the preset conditions from the first feature data to obtain second feature data; The processing unit is further used to train a supervised classification learning model according to the supervised learning algorithm and the second feature data, and determine the supervised classification learning model as the terminal replacement prediction model; wherein, the first feature data includes: label type feature data, continuous type feature data, and discrete type feature data; The processing unit is specifically used for: Remove the abnormal data in the second terminal information and the second service information to obtain the data to be processed; the data to be processed includes label-type data to be processed, continuous-type data to be processed, and discrete-type data to be processed; Perform feature engineering processing on the data to be processed to obtain the feature data of the data to be processed; the feature data of the data to be processed includes: the label-type feature data, the continuous-type feature data to be processed, and the discrete-type feature data to be processed; Perform normalization processing on the continuous-type feature data to be processed to obtain the continuous-type feature data; Perform dummy variable transformation processing on the discrete-type feature data to be processed to obtain the discrete-type feature data; Wherein, the label-type feature data includes the network type feature data of the training terminal; the multiple unsupervised clustering learning models include multiple classification clusters; The processing unit is specifically configured to: Determine the first proportion of the target type of terminal in the training terminal according to the network type feature data; Determine the second proportion of the target type of terminal in each classification cluster of the multiple classification clusters; Select, from the second proportions, the classification clusters corresponding to the second proportions that are greater than the first proportion to determine a first set of classification clusters; Determine a first set of non-target type terminals according to the network type feature data; Select, from the first set of classification clusters, the classification clusters to which each non-target type terminal in the first set of non-target type terminals belongs to obtain a second set of classification clusters; Sum the second proportions corresponding to each classification cluster in the second set of classification clusters to obtain the heterogeneous target values of each non-target type terminal; Select, from the heterogeneous target values of each non-target type terminal, the non-target type terminals corresponding to the heterogeneous target values that are greater than a preset heterogeneous target value to obtain a second set of non-target type terminals; Determine a third set of non-target type terminals according to the second set of non-target type terminals, the first proportion, and a preset removal threshold, and determine the non-target type terminals in the third set of non-target type terminals as the non-target type terminals that meet the preset conditions.

6. The terminal switching prediction device according to claim 5, wherein, The processing unit is specifically configured to: Determine a set of values; the set of values includes: a first value, a second value, and a third value; the first value is the number of non-target type terminals in the second set of non-target type terminals; the second value is the product of the number of non-target type terminals in the first set of non-target type terminals and the first proportion; the third value is the product of the number of non-target type terminals in the first set of non-target type terminals and the preset removal threshold; When the first value is the minimum value in the set of values, determine the second set of non-target type terminals as the third set of non-target type terminals; When the second value is the minimum value in the value set, the product of the number of non-target class terminals in the second non-target class terminal set and the reciprocal of the second value is determined as the first value to be removed, and non-target class terminals corresponding to the first value to be removed are selected from the second non-target class terminal set in descending order of the heterogeneous target value to determine the third non-target class terminal set; When the third value is the minimum value in the value set, the product of the number of non-target class terminals in the second non-target class terminal set and the reciprocal of the third value is determined as the second value to be removed, and non-target class terminals corresponding to the second value to be removed are selected from the second non-target class terminal set in descending order of the heterogeneous target value to determine the third non-target class terminal set.

7. The terminal replacement prediction device according to claim 5 or 6, characterized in that the processing unit is specifically configured to: determine third feature data of the terminal to be predicted; the third feature data includes feature data of the first terminal information and feature data of the first service information; select target feature data from the third feature data; the target feature data includes continuous feature data and discrete feature data of the terminal to be predicted; input the target feature data into the terminal replacement prediction model to obtain an initial probability value; when the initial probability value is greater than a preset probability value, determine whether the terminal to be predicted will be replaced during the target time period according to the initial probability value and the third feature data.

8. The terminal replacement prediction device according to claim 7, characterized in that the processing unit is specifically configured to: determine a target probability value according to the initial probability value and the third feature data; the initial probability value, the third feature data and the target probability value satisfy the following formula: where Y1 is the target probability value, Y2 is the initial probability value, a is the feature data of the number of days the terminal to be predicted has been used; b is the number of terminals the user corresponding to the terminal to be predicted has owned within n days; n is a positive integer; when the target probability value is greater than or equal to the preset probability value, determine that the terminal to be predicted will be replaced during the target time period; when the target probability value is less than the preset probability value, determine that the terminal to be predicted will not be replaced during the target time period.

9. A terminal replacement prediction device, characterized in that it includes a memory and a processor; the memory is used to store computer execution instructions, and the processor is connected to the memory through a bus; when the terminal replacement prediction device runs, the processor executes the computer execution instructions stored in the memory, so that the terminal replacement prediction device executes the terminal replacement prediction method according to any one of claims 1-4.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium includes computer-executable instructions that, when run on a computer, cause the computer to execute the terminal switching prediction method according to any one of claims 1-4.

Images