CN116629918A - User consumption prediction method and system based on cross-border electronic commerce - Google Patents
User consumption prediction method and system based on cross-border electronic commerce Download PDFInfo
- Publication number
- CN116629918A CN116629918A CN202310604745.8A CN202310604745A CN116629918A CN 116629918 A CN116629918 A CN 116629918A CN 202310604745 A CN202310604745 A CN 202310604745A CN 116629918 A CN116629918 A CN 116629918A
- Authority
- CN
- China
- Prior art keywords
- consumption
- user
- prediction
- probability
- amount
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 44
- 238000012549 training Methods 0.000 claims description 64
- 230000006399 behavior Effects 0.000 claims description 42
- 238000013145 classification model Methods 0.000 claims description 22
- 238000000605 extraction Methods 0.000 claims description 17
- 238000010801 machine learning Methods 0.000 claims description 15
- 230000001419 dependent effect Effects 0.000 claims description 10
- 238000005070 sampling Methods 0.000 claims description 7
- 238000004364 calculation method Methods 0.000 claims description 6
- 238000013528 artificial neural network Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000013507 mapping Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000003066 decision tree Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000012417 linear regression Methods 0.000 description 1
- 238000007477 logistic regression Methods 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 230000009897 systematic effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0201—Market modelling; Market analysis; Collecting market data
- G06Q30/0202—Market predictions or forecasting for commercial activities
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/27—Regression, e.g. linear or logistic regression
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/01—Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Life Sciences & Earth Sciences (AREA)
- Software Systems (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Business, Economics & Management (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Development Economics (AREA)
- Finance (AREA)
- Strategic Management (AREA)
- Mathematical Physics (AREA)
- Computing Systems (AREA)
- Accounting & Taxation (AREA)
- Computational Linguistics (AREA)
- Entrepreneurship & Innovation (AREA)
- General Business, Economics & Management (AREA)
- Medical Informatics (AREA)
- Marketing (AREA)
- Economics (AREA)
- Game Theory and Decision Science (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a user consumption prediction method based on cross-border electronic commerce, which comprises the following steps: s1: constructing a consumption probability prediction model; s2: constructing a consumption amount prediction model; s3: predicting consumption probability: predicting the consumption probability of the consumption prediction task by using the constructed consumption probability prediction model; s4: predicting a consumption amount: predicting the expense amount according to the expense prediction task by using the constructed expense amount prediction model; s5: generating a consumption prediction result: according to the consumption probability predicted value obtained in the step S3 and the consumption amount predicted value obtained in the step S4, a consumption predicted result is generated and output, and the user consumption prediction of the cross-border e-commerce platform is disassembled into two-step prediction of consumption probability and consumption amount; the invention has the characteristics of improving the prediction accuracy and stability.
Description
Technical Field
The invention relates to the technical field of cross-border e-commerce consumption prediction, in particular to a cross-border e-commerce based user consumption prediction method and system.
Background
The traditional cross-border e-commerce platform realizes more accurate and more efficient marketing by predicting the consumption behaviors of users on the platform, optimizes the user experience and the marketing budget utilization rate, adopts an RFM model to measure and analyze the historical consumption behaviors of the users from three dimensions of last consumption, consumption frequency and consumption amount so as to predict the future consumption behaviors of the users, but the RFM model depends on the historical consumption behavior data of the users, so that the consumption prediction is difficult to be carried out on new users which do not generate the consumption behaviors, and the cross-border e-commerce platform usually attracts a large number of new users by utilizing marketing activities so as to further need to carry out the consumption prediction on the new users, and the RFM model is not suitable for the application scene;
some platforms adopt a machine learning algorithm, firstly extracting a series of features from collected user related data, inputting the features into a machine learning regression model, taking the consumption amount of a user in a specific time period as a regression target, and learning the quantitative relation between the features and the regression target through fitting the model and the data; finally, predicting the future consumption amount of the user according to the trained regression model, wherein the existing method has two defects: firstly, the prediction of the consumption amount is biased, but the probability prediction of the possibility of the occurrence of the consumption behavior is difficult to provide, secondly, in the typical application scenario of the cross-border electronic commerce, the object of the user consumption prediction is that a large number of new users which do not generate consumption records yet are included as widely as possible on a platform, wherein the user proportion of the users with high-frequency historical consumption records is generally low, so that the data distribution is uneven, the training effect of a regression model is affected, and therefore, it is necessary to design a user consumption prediction method and system based on the cross-border electronic commerce for improving the prediction accuracy and stability.
Disclosure of Invention
The invention aims to provide a cross-border electronic commerce-based user consumption prediction method and system, which are used for solving the problems in the background technology.
In order to solve the technical problems, the invention provides the following technical scheme: a user consumption prediction method based on cross-border electronic commerce comprises the following steps:
s1: constructing a consumption probability prediction model;
s2: constructing a consumption amount prediction model;
s3: predicting consumption probability: predicting the consumption probability of the consumption prediction task by using the constructed consumption probability prediction model;
s4: predicting a consumption amount: predicting the expense amount according to the expense prediction task by using the constructed expense amount prediction model;
s5: generating a consumption prediction result: and generating and outputting a consumption prediction result according to the consumption probability prediction value obtained in the step S3 and the consumption amount prediction value obtained in the step S4.
According to the above technical solution, the step of constructing the consumption probability prediction model includes:
based on user history data of a cross-border e-commerce platform, a machine learning classification model is constructed and used as a consumption probability prediction model, and the method for constructing a classification model training sample set comprises the following steps: a training sample data set of a classification model is formed by a certain number of training samples and is used for being input into a machine learning classification model so as to learn the relation between independent variables and dependent variables in the training samples, and aiming at a user consumption prediction problem, each training sample corresponds to a consumption prediction task and consists of three parts of information, namely prediction task parameters, user characteristics serving as the independent variables and classification marks serving as the dependent variables.
According to the above technical solution, the step of constructing the consumption amount prediction model includes:
constructing a training sample set based on the user history data, and training a regression model based on the training sample set to serve as a consumption amount prediction model;
the method for constructing the regression model training sample set comprises the following steps: the training sample set of the regression model consists of a certain number of training samples, wherein each training sample corresponds to one consumption prediction task and consists of three parts of information, namely a prediction task parameter, a user characteristic of an independent variable and a regression target of the dependent variable, and the prediction task parameter (u, T, T): the representation task is to predict the consumption behavior of a target user u in a target time period (T, t+T), the user characteristic x is a vector in form and is obtained by calculating historical data of the user u from a time point T through characteristic extraction, and the regression target z is a numerical value in form and is equal to the total consumption amount of the user u when the user u has the consumption behavior in the time period (T, t+T).
According to the above technical solution, the step of predicting the consumption probability includes:
for an input user consumption prediction task (u, T, T), predicting the probability of consumption behavior of the user u in a time period (T, t+T);
s3-1: acquiring historical data of a user u cut-off time point t through a data acquisition module;
s3-2: extracting user characteristics x from the historical data through a characteristic extraction module;
s3-3: inputting the user characteristic x into a constructed consumption probability prediction model to obtain a predicted value f (x);
s3-4: because the consumption probability prediction model is constructed based on the undersampled training sample set, the directly output predicted value f (x) needs to be calibrated according to the negative sample sampling rate w, the consumption probability predicted value p is obtained and output, and the calculation formula is as follows:
according to the above technical solution, the step of predicting the consumption amount includes:
for an input user consumption prediction task (u, T, T), predicting the consumption amount of the user u in a time period (T, t+T) by the following steps;
s4-1: acquiring historical data of a user u cut-off time point t through a data acquisition module;
s4-2: extracting user characteristics x from the historical data through a characteristic extraction module;
s4-3: and inputting the user characteristic x into the constructed expense amount prediction model to obtain and output an expense amount predicted value g.
According to the technical scheme, the user consumption prediction system based on the cross-border electronic commerce comprises:
the M1-data acquisition module is used for acquiring historical data generated by a specified user in a specified time period from a data source of a cross-border electronic commerce platform;
and the M2-feature extraction module is used for extracting the user features from the historical data of the user according to a predefined calculation mode.
Compared with the prior art, the invention has the following beneficial effects: according to the invention, the user consumption prediction of the cross-border e-commerce platform is disassembled into two-step prediction of consumption probability and consumption amount; predicting the consumption probability based on the machine learning classification model and predicting the consumption amount based on the machine learning regression model; finally, the consumption probability predicted value and the consumption amount predicted value are combined into a consumption predicted result, so that the consumption predicted result is suitable for the characteristics of user data of the cross-border electronic commerce, and the consumption probability and the consumption amount can be predicted at the same time.
Drawings
The accompanying drawings are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate the invention and together with the embodiments of the invention, serve to explain the invention. In the drawings:
FIG. 1 is a flowchart of a user consumption prediction method based on a cross-border e-commerce according to an embodiment of the present invention;
fig. 2 is a schematic diagram of module composition of a user consumption prediction system based on a cross-border e-commerce according to a second embodiment of the present invention;
FIG. 3 is a flowchart for constructing a consumption probability prediction model according to an embodiment of the present invention;
fig. 4 is a flowchart of a method for constructing a cost amount prediction model according to an embodiment of the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Embodiment one: fig. 1 is a flowchart of a cross-border e-commerce based user consumption prediction method provided in an embodiment of the present invention, where the embodiment may apply a scenario of cross-border e-commerce platform user consumption behavior prediction, and the method may be performed by a cross-border e-commerce based user consumption predictor system provided in the embodiment, as shown in fig. 1, and the method specifically includes the following steps:
s1: constructing a consumption probability prediction model;
in the embodiment of the invention, a machine learning classification model is constructed based on user history data of a cross-border e-commerce platform and is used as a consumption probability prediction model, and the method for constructing a classification model training sample set comprises the following steps: a training sample data set of a classification model is formed by a certain number of training samples and is used for being input into a machine learning classification model so as to learn the relation between independent variables and dependent variables in the training samples, and aiming at a user consumption prediction problem, each training sample corresponds to a consumption prediction task and consists of three parts of information, namely prediction task parameters, user characteristics serving as the independent variables and classification marks serving as the dependent variables;
in the embodiment of the invention, the predicted task parameters of the training sample are marked as triples (u, T, T) and represent the consumption behavior of a predicted target user u in a target time period (T, t+T), wherein three parameters are respectively defined as a target user u, which is a user ID value of a unique identifier for distinguishing different users, a target time period starting point T, which is a time point in form and represents the consumption behavior of the user u to be predicted after the target user u, a target time period length T, which is a time length value and T form together to form a target time period (T, t+T), the user characteristics of the training sample are marked as x, a vector is formed by characteristic extraction, the historical data of the user u from the time point T is calculated, the classification mark of the training sample is marked as y, the target user u is a label with a discrete value of 0 or 1, the consumption behavior of the user u in the time period (T, t+T) is determined, wherein the value of 1 represents a positive sample, the consumption behavior of the user u in the time period is represented by the negative behavior of the user in the time period, and the consumption behavior of the user u in the time period is represented by the time period;
illustratively, the process of constructing the classification model training sample set is: according to a certain user consumption prediction scene, determining candidate values of task parameters: the method comprises the steps of defining a target time period starting point T as a reference time, defining a value range as a time period of a historical date, defining a value step length as 1 day, fixing the length T of the target time period to be 7 days, traversing the reference time T according to the value range and the step length determined in the last step, acquiring a user set of the cross-border e-commerce platform for each selected reference time, generating a prediction task for each user u in the set, wherein task parameters of the prediction task are expressed as triplets (u, T and T), generating 365 XN prediction tasks altogether through traversing time and users when the user quantity on the platform is always equal to N, extracting corresponding user characteristics for each prediction task, acquiring historical data of the user u from the time point T through a data acquisition module, and extracting user characteristics from the data through a characteristic extraction module; for each prediction task, generating a corresponding classification mark, firstly acquiring consumption behavior data of a user u in a time period (T, t+T) through a data acquisition module, and then generating the classification mark according to the data, wherein the value of at least one consumption behavior is 1, namely a corresponding positive sample, or the value of at least one consumption behavior is 0, and a corresponding negative sample, splicing each prediction task with the corresponding user characteristics and the classification mark, constructing a training sample, wherein the number of the training samples is equal to that of the prediction tasks, and constructing a classification model training sample set by all constructed training samples;
in the embodiment of the invention, the method for training the classification model comprises the following steps: the constructed training sample set is undersampled firstly and then is input into a machine learning classification model for model training, and the undersampling principle of the training sample set is as follows: undersampling the negative samples in the original training sample set, namely: a portion of the negative samples is selected and combined with all positive samples to form a training sample subset, wherein the ratio of the number of undersampled negative samples to the number of total negative samples, denoted as the negative sample rate, w (0<w.ltoreq.1), for a determined negative sample rate, alternative undersampling methods include, but are not limited to: random sampling, hierarchical sampling, i.e., for example: the number of all positive samples is 10 ten thousand, the number of all negative samples is 1000 ten thousand, the sampling rate w of the negative samples is 0.05, the number of the negative samples after undersampling is 1000 ten thousand x 0.05=50 ten thousand, the proportion of the positive and negative samples is changed from 1:100 to 1:5 after undersampling, because of the characteristics of user data of a cross-border electronic commerce platform, users who have consumed behaviors are usually far fewer than users who have not consumed behaviors in a limited time period, the number of the positive samples is far fewer than the number of the negative samples in training samples of a classification model, and the unbalance of the number of the samples of different types, namely the unbalance of the positive and negative samples, influences the training of the classification model, leads to inaccurate prediction results, so the problem of the unbalance of the positive and negative samples is overcome by the method, and the unbalance degree is relieved;
illustratively, the training process of the classification model learns the mapping relationship from the user features to the classification labels by the machine learning algorithm based on the user features as arguments and the classification labels as arguments of each training sample inputted, i.e. learns how to correctly classify the user features inputted reflecting the user's history data into corresponding categories reflecting whether the user has consumption behavior in a future time period, in this embodiment, the optional machine learning classification model includes but is not limited to: logistic regression, gradient lifting decision tree and neural network, and the positive and negative sample proportion in the original training sample set reflects the actual distribution of classification marks in data, and in the undersampled training sample subset, the positive and negative sample proportion deviates from the actual distribution, so that the direct output predicted value of the classification model constructed based on the undersampled training sample subset has systematic deviation, and probability calibration is needed based on the negative sample sampling rate w.
S2: constructing a consumption amount prediction model;
in the embodiment of the invention, a training sample set is constructed based on user history data, and a regression model is trained based on the training sample set, and is used as a consumption amount prediction model, and the method for constructing the regression model training sample set comprises the following steps: the training sample set of the regression model consists of a certain number of training samples, wherein each training sample corresponds to one consumption prediction task and consists of three parts of information, namely a prediction task parameter, a user characteristic of an independent variable and a regression target of the dependent variable, and the prediction task parameter (u, T, T): the representation task is to predict the consumption behavior of a target user u in a target time period (T, t+T), wherein the user characteristic x is a vector in form and is obtained by calculating historical data of the user u from a time point T through characteristic extraction, and the regression target z is a numerical value in form and is equal to the total consumption amount of the user u when the consumption behavior exists in the time period (T, t+T);
illustratively, the process of constructing the regression model training sample set is: according to a certain user consumption prediction scene, determining candidate values of task parameters: setting the value range date of the starting point T of the target time period and the value step length as 1 day, fixing the length T of the target time period to be equal to 7 days, defining the starting point T of the target time period as reference time, traversing the reference time T according to the value range and the step length determined in the last step, acquiring a user set of the time point on a cross-border e-commerce platform for each selected reference time, generating a prediction task for each user u in the set, wherein task parameters are triplets (u, T and T), extracting corresponding user characteristics for each prediction task, firstly acquiring historical data of the time point T of the user u by a data acquisition module, extracting user characteristics from the data by a characteristic extraction module, generating corresponding regression targets for each prediction task by a data acquisition module, firstly acquiring consumption behavior data of the user u in the time period (T and t+T), generating regression targets according to the data, wherein the consumption behavior is equal to the total consumption amount of the regression targets, and the regression targets are not lost, screening the regression targets, constructing a model of the regression targets, and the regression targets are not influenced by the prediction tasks, and the model is not used as a negative training model, and the model is not influenced by the negative consumption targets;
illustratively, the training process of the regression model is: based on the constructed regression model training sample set, the mapping relation from the user characteristics as independent variables to the regression targets as dependent variables is learned by a machine learning algorithm, namely how to predict the consumption amount of the user in the future time period according to the historical data of the user, in the embodiment, the optional machine learning regression model comprises but is not limited to: linear regression, gradient lifting regression trees, neural networks.
S3: predicting consumption probability: predicting the consumption probability of the consumption prediction task by using the constructed consumption probability prediction model;
in the embodiment of the invention, for an input user consumption prediction task (u, T, T), the probability that the user u has consumption behaviors in a time period (T, t+T) is predicted by the following steps;
s3-1: acquiring historical data of a user u cut-off time point t through a data acquisition module;
s3-2: extracting user characteristics x from the historical data through a characteristic extraction module;
s3-3: inputting the user characteristic x into a constructed consumption probability prediction model to obtain a predicted value f (x);
s3-4: because the consumption probability prediction model is constructed based on the undersampled training sample set, the directly output predicted value f (x) needs to be calibrated according to the negative sample sampling rate w, the consumption probability predicted value p is obtained and output, and the calculation formula is as follows:
s4: predicting a consumption amount: predicting the expense amount according to the expense prediction task by using the constructed expense amount prediction model;
in the embodiment of the invention, for an input user consumption prediction task (u, T, T), the consumption amount of the user u in a time period (T, t+T) is predicted by the following steps;
s4-1: acquiring historical data of a user u cut-off time point t through a data acquisition module;
s4-2: extracting user characteristics x from the historical data through a characteristic extraction module;
s4-3: and inputting the user characteristic x into the constructed expense amount prediction model to obtain and output an expense amount predicted value g.
S5: and (3) generating a consumption prediction result, and generating and outputting the consumption prediction result according to the consumption probability prediction value obtained in the step (S3) and the consumption amount prediction value obtained in the step (S4).
In the embodiment of the invention, the consumption probability predicted value p obtained in the third step and the consumption amount predicted value g obtained in the fourth step are combined into a consumption predicted result and output, and prediction of consumption behaviors in two dimensions of probability and amount is provided, and in an exemplary consumption predicted result, p=0.6 and g=100, which indicate that a target user predicts that the consumption behaviors exist at a probability of 0.6 and the consumption amount is 100 and that the consumption behaviors exist at a probability of 1-0.6 in a target time period, further, the product of p and g is expressed as a mathematical expectation of the consumption amount, namely: desired amount of consumption=0.6×100+ (1-0.6) ×0=60=pg.
Embodiment two:
the second embodiment of the invention provides a method and a system for predicting user consumption based on a cross-border e-commerce, and fig. 2 is a schematic diagram of module composition of a system for predicting user consumption based on a cross-border e-commerce, as shown in fig. 2, the system comprises:
the M1-data acquisition module is used for acquiring historical data generated by a specified user in a specified time period from a data source of a cross-border electronic commerce platform;
and the M2-feature extraction module is used for extracting the user features from the historical data of the user according to a predefined calculation mode.
In some embodiments of the present invention, the M1-data acquisition module comprises:
a consumption predicted target user for defining a consumption predicted target user;
a consumption prediction target time period starting point, which is used for setting a consumption prediction target time period starting point;
the consumption prediction target time period length is used for setting the consumption prediction target time period length;
the user characteristics are used for acquiring multiple types of characteristics of the user;
the classification mark is used for classifying whether the consumption behavior exists or not;
a regression target for defining a regression target for the amount of the expense;
the consumption probability predicted value is used for calculating the consumption probability predicted value;
a predicted value of the amount of consumption for calculating the predicted value of the amount of consumption,
in some embodiments of the invention, the M2-feature extraction module comprises:
the user behavior characteristics are used for defining various user behavior characteristics according to various user behavior data;
and the user attribute features are used for extracting the user attribute features by using the personal information of the user.
It is noted that relational terms such as first and second, and the like are used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Moreover, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus.
Finally, it should be noted that: the foregoing description is only a preferred embodiment of the present invention, and the present invention is not limited thereto, but it is to be understood that modifications and equivalents of some of the technical features described in the foregoing embodiments may be made by those skilled in the art, although the present invention has been described in detail with reference to the foregoing embodiments. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (9)
1. A user consumption prediction method based on cross-border electronic commerce is characterized by comprising the following steps of: the method comprises the following steps:
s1: constructing a consumption probability prediction model;
s2: constructing a consumption amount prediction model;
s3: predicting consumption probability: predicting the consumption probability of the consumption prediction task by using the constructed consumption probability prediction model;
s4: predicting a consumption amount: predicting the expense amount according to the expense prediction task by using the constructed expense amount prediction model;
s5: generating a consumption prediction result: and generating and outputting a consumption prediction result according to the consumption probability prediction value obtained in the step S3 and the consumption amount prediction value obtained in the step S4.
2. The cross-border e-commerce based user consumption prediction method as claimed in claim 1, wherein the method comprises the steps of: the step of constructing the consumption probability prediction model comprises the following steps:
based on user history data of a cross-border e-commerce platform, a machine learning classification model is constructed and used as a consumption probability prediction model, and the method for constructing a classification model training sample set comprises the following steps: a training sample data set of a classification model is formed by a certain number of training samples and is used for being input into a machine learning classification model so as to learn the relation between independent variables and dependent variables in the training samples, and aiming at a user consumption prediction problem, each training sample corresponds to a consumption prediction task and consists of three parts of information, namely prediction task parameters, user characteristics serving as the independent variables and classification marks serving as the dependent variables.
3. The cross-border e-commerce based user consumption prediction method as claimed in claim 1, wherein the method comprises the steps of: the step of constructing the consumption amount prediction model comprises the following steps:
constructing a training sample set based on the user history data, and training a regression model based on the training sample set to serve as a consumption amount prediction model;
the method for constructing the regression model training sample set comprises the following steps: the training sample set of the regression model consists of a certain number of training samples, wherein each training sample corresponds to one consumption prediction task and consists of three parts of information, namely a prediction task parameter, a user characteristic of an independent variable and a regression target of the dependent variable, and the prediction task parameter (u, T, T): the representation task is to predict the consumption behavior of a target user u in a target time period (T, t+T), the user characteristic x is a vector in form and is obtained by calculating historical data of the user u from a time point T through characteristic extraction, and the regression target z is a numerical value in form and is equal to the total consumption amount of the user u when the user u has the consumption behavior in the time period (T, t+T).
4. The cross-border e-commerce based user consumption prediction method as claimed in claim 1, wherein the method comprises the steps of: the step of predicting the consumption probability comprises the following steps:
for an input user consumption prediction task (u, T, T), predicting the probability of consumption behavior of the user u in a time period (T, t+T);
s3-1: acquiring historical data of a user u cut-off time point t through a data acquisition module;
s3-2: extracting user characteristics x from the historical data through a characteristic extraction module;
s3-3: inputting the user characteristic x into a constructed consumption probability prediction model to obtain a predicted value f (x);
s3-4: because the consumption probability prediction model is constructed based on the undersampled training sample set, the directly output predicted value f (x) needs to be calibrated according to the negative sample sampling rate w, the consumption probability predicted value p is obtained and output, and the calculation formula is as follows:
5. the cross-border e-commerce based user consumption prediction method as claimed in claim 1, wherein the method comprises the steps of: the step of predicting the amount of consumption includes:
for an input user consumption prediction task (u, T, T), predicting the consumption amount of the user u in a time period (T, t+T) by the following steps;
s4-1: acquiring historical data of a user u cut-off time point t through a data acquisition module;
s4-2: extracting user characteristics x from the historical data through a characteristic extraction module;
s4-3: and inputting the user characteristic x into the constructed expense amount prediction model to obtain and output an expense amount predicted value g.
6. The cross-border e-commerce based user consumption prediction method as claimed in claim 1, wherein the method comprises the steps of: the step of generating a consumption prediction result includes:
and combining the consumption probability predicted value obtained in the third step and the consumption amount predicted value obtained in the fourth step into a consumption predicted result and outputting the consumption predicted result, thereby providing the prediction of the consumption behavior in two dimensions of probability and amount.
7. A user consumption prediction system based on cross-border electronic commerce is characterized in that: the system comprises:
the M1-data acquisition module is used for acquiring historical data generated by a specified user in a specified time period from a data source of a cross-border electronic commerce platform;
and the M2-feature extraction module is used for extracting the user features from the historical data of the user according to a predefined calculation mode.
8. The cross-border e-commerce based user consumption prediction system of claim 7, wherein: the M1-data acquisition module comprises:
a consumption predicted target user for defining a consumption predicted target user;
a consumption prediction target time period starting point, which is used for setting a consumption prediction target time period starting point;
the consumption prediction target time period length is used for setting the consumption prediction target time period length;
the user characteristics are used for acquiring multiple types of characteristics of the user;
the classification mark is used for classifying whether the consumption behavior exists or not;
a regression target for defining a regression target for the amount of the expense;
the consumption probability predicted value is used for calculating the consumption probability predicted value;
and the consumption amount predicted value is used for calculating the consumption amount predicted value.
9. The cross-border e-commerce based user consumption prediction system of claim 7, wherein: the M2-feature extraction module comprises:
the user behavior characteristics are used for defining various user behavior characteristics according to various user behavior data;
and the user attribute features are used for extracting the user attribute features by using the personal information of the user.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310604745.8A CN116629918A (en) | 2023-05-26 | 2023-05-26 | User consumption prediction method and system based on cross-border electronic commerce |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310604745.8A CN116629918A (en) | 2023-05-26 | 2023-05-26 | User consumption prediction method and system based on cross-border electronic commerce |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116629918A true CN116629918A (en) | 2023-08-22 |
Family
ID=87609437
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310604745.8A Pending CN116629918A (en) | 2023-05-26 | 2023-05-26 | User consumption prediction method and system based on cross-border electronic commerce |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116629918A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117708764A (en) * | 2024-02-06 | 2024-03-15 | 青岛天高智慧科技有限公司 | Intelligent analysis method for student consumption data based on campus card |
-
2023
- 2023-05-26 CN CN202310604745.8A patent/CN116629918A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117708764A (en) * | 2024-02-06 | 2024-03-15 | 青岛天高智慧科技有限公司 | Intelligent analysis method for student consumption data based on campus card |
| CN117708764B (en) * | 2024-02-06 | 2024-05-03 | 青岛天高智慧科技有限公司 | Intelligent analysis method for student consumption data based on campus card |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109615226B (en) | Operation index abnormity monitoring method | |
| Saito et al. | Open bandit dataset and pipeline: Towards realistic and reproducible off-policy evaluation | |
| CN111460167A (en) | Method for positioning pollution discharge object based on knowledge graph and related equipment | |
| CN107644297B (en) | Energy-saving calculation and verification method for motor system | |
| CN112149352B (en) | Prediction method for marketing activity clicking by combining GBDT automatic characteristic engineering | |
| CN116629918A (en) | User consumption prediction method and system based on cross-border electronic commerce | |
| CN110310012B (en) | Data analysis method, device, equipment and computer readable storage medium | |
| CN115660725A (en) | Method for depicting multi-dimensional energy user portrait | |
| CN117291655B (en) | Consumer life cycle operation analysis method based on entity and network collaborative mapping | |
| CN114548494A (en) | Visual cost data prediction intelligent analysis system | |
| CN113111256A (en) | Production work order recommendation method based on depth knowledge map | |
| CN116777281A (en) | ARIMA model-based power equipment quality trend prediction method and device | |
| CN115952914A (en) | Big data-based electric power metering operation and maintenance work judgment planning method | |
| CN113987261A (en) | Video recommendation method and system based on dynamic trust perception | |
| CN114169763A (en) | Measuring instrument demand prediction method, system, computing device and storage medium | |
| CN107122496B (en) | Near infrared spectrum substance content indexing method, device and medium based on category analysis method | |
| Liu et al. | Short-term Load Forecasting Approach with SVM and Similar Days Based on United Data Mining Technology | |
| CN117422314B (en) | Enterprise data evaluation method and equipment based on big data analysis | |
| CN115808504B (en) | Online drift compensation method for gas sensor for concentration prediction | |
| CN112163714B (en) | XGboost-based campus client group load release algorithm | |
| CN107357805B (en) | Near infrared spectrum substance content indexing method and device based on similarity discrimination method | |
| Simidjievski et al. | Learning ensembles of process-based models by bagging of random library samples | |
| Yan et al. | SELECTING AND COMBINING ROBUST FORECASTING MODELS USING RULES BASED ON VALIDITY AND RELIABILITY. | |
| CN117669860A (en) | Electrical equipment energy efficiency evaluation method, device, equipment and medium | |
| CN117349270A (en) | Multi-mode mutual-inspection water intake metering data anomaly detection online processing method and device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |