Real-time analysis method for mobile customer purchase intention grading

Abstract

The invention discloses a real-time analysis method for mobile customer purchase intention grading, which comprises the steps of collecting a customer in-store data set and an online data set, carrying out abnormal value processing on the data, establishing an association relationship between the data in a database by using a primary key and a foreign key through a unique identifier of a customer, associating the online data with the offline data, and carrying out real-time analysis on the online data and the offline data. A data matching technology is combined to identify and associate similar data records, correlation analysis is utilized, most representative features are selected from a data set, a cross validation technology is combined to construct and train a model, customer purchase intention data variables are continuously collected, an association relationship is established in a database, and the customer purchase intention data is acquired. According to the method, the mobile customer purchase intention data can be effectively and accurately analyzed in real time, and the customer purchase intention data can be fed back to the sales personnel in time.

Description

Technical Field

[0001] The invention relates to the field of information technology, and in particular to a real-time analysis method for grading mobile customer purchase intentions. Background Art

[0002] The invention entitled: User Purchase Intention Evaluation Method, System and Terminal Device is disclosed in Chinese Patent Publication No.: CN111311353B. In the core technical solution of the invention, based on the communication information between the customer service and the user, the communication information keywords at the first time point are determined as the benchmark phrase, the keywords of the communication information between the first time point and the second time point are determined as the intended phrase, and the keywords associated with the benchmark phrase are eliminated from the intended phrase, and the eliminated intended phrase is used as the target phrase, which is the user's purchase intention, solving the problem that it is difficult to accurately grasp the user's purchase intention.

[0003] In the existing technology, customer purchasing behavior is analyzed through a single-dimensional data analysis of communication information, ignoring the mutual influence and comprehensive effect of data of different dimensions. At the same time, the purchase intention level of offline liquidity customers is only analyzed through simple single-dimensional data, and the analysis accuracy can no longer meet the rapidly changing needs of the modern market.

[0004] The present invention proposes a solution to the above shortcomings: collect the customer's in-store data set and online data set, establish tables in the database after processing the collected data, use Jaccard similarity to identify the data records in the two tables, associate the two tables, combine the calculation of the Pearson correlation coefficient, evaluate the correlation between the features and purchase intention, and realize real-time and accurate analysis of customer purchase intention classification by real-time monitoring of customer behavior data. Summary of the invention

[0005] In order to solve the above-mentioned technical problems, the present invention provides a real-time analysis method for mobile customer purchase intention grading.

[0006] The technical solution of the present invention is implemented as follows: a real-time analysis method for mobile customer purchase intention classification, comprising:

[0007] S1. Collect customer store data sets and online data sets. The store data sets include customer service satisfaction data x1, and the online data sets include customer membership registration data x2;

[0008] S2, clean the data and deal with outliers;

[0009] S3. Data association uses primary keys and foreign keys in the database to establish associations between data, associate online and offline data, and use data matching technology and fuzzy matching to identify and associate similar data records;

[0010] S4. Extract features related to purchase intention from the dataset and use correlation analysis to select the most representative features;

[0011] S5. Model construction and training Use cross-validation technology to train and tune the model;

[0012] S6. Train and tune the model through step S5, continuously collect customer purchase intention data, and provide real-time feedback of the model prediction results to sales staff so that they can adjust their sales strategies in a timely manner.

[0013] Furthermore, in the step S1, customer service satisfaction data x1 and customer membership registration data x2 are collected, and customer data is retrieved by calling the storage information of the IoT device.

[0014] Furthermore, in the step S2, outlier processing is adopted to calculate the Z-score of the customer service satisfaction data x1 and the customer member registration data x2, where the Z-score is the difference between the data and the mean divided by the standard deviation;

[0015] The calculation formula is:

[0016]

[0017] Furthermore, in the step S3, primary key and foreign key association are used, and similar data records are identified through fuzzy matching in data matching technology;

[0018] Create a customer store data set table (Table1) and an online data set table (Table2):

[0019] Create a database:

[0020] In Table1, create a foreign key ID that references the primary key ID in Table2;

[0021] Data import:

[0022] Import the cleaned data into the table, making sure that the ID field in each table is unique;

[0023] Data Matching:

[0024] Use the fuzzy matching method in data matching technology to compare the ID fields in Table1 and Table2. The fuzzy matching uses Jaccard similarity to calculate the Jaccard similarity.

[0025] Select ID2 with the highest similarity:

[0026] For each ID1, find the ID2 with the highest Jaccard similarity and select the largest Jaccard similarity value;

[0027] Related data:

[0028] For successfully matched ID1 and ID2, use foreign key association to combine them into a complete data record. In the database, point the foreign key ID in Table1 to the matched ID2 in Table2.

[0029] Through the above steps, the primary key and foreign key are associated in the database, and similar data records are identified through Jaccard similarity in data matching technology.

[0030] Furthermore, in the S4 step, the correlation analysis is performed to select the most representative features:

[0031] The calculation formula of Pearson correlation coefficient in correlation analysis is:

[0032]

[0033] The formula for calculating covariance is:

[0034]

[0035] The formula for calculating the standard deviation is:

[0036]

[0037] The Pearson correlation coefficient is an indicator that measures the degree of linear correlation between two variables x and y. The correlation coefficient value is between -1 and 1. When the linear relationship between the two variables increases, the correlation coefficient tends to 1 or -1. When one variable increases, the other variable also increases, and the correlation coefficient is positive, otherwise it is negative. If the correlation coefficient is close to 0, it means that there is almost no linear correlation between the two variables. Through these simplified formulas, the Pearson correlation coefficient can be quickly calculated to evaluate the correlation between features and purchase intentions.

[0038] Furthermore, the setting of the cross-validation technique used in step S5 is:

[0039] Data preparation ensures that the dataset has been cleaned and divided into feature set (X) and label set (y);

[0040] Parameter settings determine the parameters of the logistic regression model, use the regularization strength (C), the maximum number of iterations, select the K value for K-fold cross validation, and set K=10;

[0041] K-fold cross validation divides the feature set (X) and label set (y) into K subsets of equal size, repeats K times, and uses one subset as the test set each time, and the remaining K-1 subsets as the training set. In each iteration, the training set is used to train the logistic regression model. The accuracy is: (the number of samples predicted correctly) / (the total number of samples). In cross validation, the training set is divided into K groups, and training and testing are performed in turn. The accuracy of each test is calculated.

[0042] Furthermore, in step S6, the model is trained and tuned through step S5, and customer behavior data is continuously collected: variable data of customer behavior is continuously collected to ensure that the data is cleaned, and then the data is input into the trained and tuned model, and the customer's real-time purchase intention results predicted by the model are fed back to the sales staff in real time.

[0043] Beneficial Effects

[0044] The present invention aims at the problem that the single-dimensional data in the prior art cannot meet the rapidly changing needs of the modern market in terms of analysis accuracy. By collecting the customer's store data set and online data set, the data is cleaned, and the data abnormality is judged by calculating the absolute value of the Z-score, and the data belonging to the abnormal value range is deleted or replaced to remove the abnormal values ​​in the collected data. After establishing a table in the database, the data is entered in the table, and the association relationship between the data is established by using the primary key and the foreign key, which solves the weakness of analyzing the customer's purchase intention behavior through single-dimensional data in the prior art. At the same time, similar data records are identified by calculating the Jaccard similarity of the data set, and the data with the highest field matching degree in the two tables are selected for association, and the features related to the purchase intention are extracted from the data set. Combined with the correlation analysis, the most representative features are selected by calculating the Pearson correlation coefficient, which solves the problem that the customer's purchase intention analysis accuracy in the prior art cannot meet the rapidly changing needs of the modern market. By continuously collecting customer behavior data, using the data to train and tune the model, the customer's purchase intention data is output in real time, and timely feedback is given to the sales staff, so that the sales staff can adjust the sales strategy in time according to the purchase intention data of different customers. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 It is a structural block diagram of a real-time analysis method for mobile customer purchase intention classification in an embodiment of the present invention;

[0046] Figure 2 A flowchart of the steps of a real-time analysis method for mobile customer purchase intention classification in an embodiment of the present invention;

[0047] Figure 3 The present invention is a flowchart of the steps of establishing a data set table for a real-time analysis method for mobile customer purchase intention grading in an embodiment of the present invention. DETAILED DESCRIPTION

[0048] In order to make the objects and advantages of the present invention more clearly understood, the present invention is further described below in conjunction with embodiments; it should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0049] The preferred implementation methods of the present invention are described below with reference to the accompanying drawings. It should be understood by those skilled in the art that these implementation methods are only used to explain the technical principles of the present invention and are not intended to limit the protection scope of the present invention.

[0050] It can be understood that the terms "first", "second", etc. used in the present application can be used in this article to describe various elements, but these elements are not limited by these terms. These terms are only used to distinguish the first element from another element.

[0051] When used herein, the singular forms "a", "an", and "said / the" may also include plural forms, unless the context clearly indicates otherwise. It should also be understood that the terms "include / comprise" or "have" and the like specify the presence of stated features, wholes, steps, operations, components, parts, or combinations thereof, but do not exclude the possibility of the presence or addition of one or more other features, wholes, steps, operations, components, parts, or combinations thereof. At the same time, the term "and / or" used in this specification includes any and all combinations of the relevant listed items.

[0052] See also Figure 1-Figure 3 As shown, a real-time analysis method for mobile customer purchase intention classification includes:

[0053] S1. Collect customer store data sets and online data sets. The store data sets include customer service satisfaction data x1, and the online data sets include customer membership registration data x2;

[0054] S2, clean the data and deal with outliers;

[0055] S3. Data association uses primary keys and foreign keys in the database to establish associations between data, associate online and offline data, and use data matching technology and fuzzy matching to identify and associate similar data records;

[0056] S4. Extract features related to purchase intention from the dataset and use correlation analysis to select the most representative features;

[0057] S5. Model construction and training Use cross-validation technology to train and tune the model;

[0058] S6. Train and tune the model through step S5, continuously collect customer behavior data, and provide real-time feedback of the model prediction results to sales staff so that they can adjust their sales strategies in a timely manner.

[0059] like Figure 1-Figure 2 As shown, in the step S1, a customer store entry data set and an online data set are collected, the store entry data set includes customer service satisfaction data x1, and the online data set includes customer member registration data x2;

[0060] Specifically, in step S1, customer service satisfaction data x1 is obtained, including data on satisfaction questionnaires filled out by customers and customer satisfaction feedback in real time, and customer membership registration data x2 is obtained, including customer membership registration information, points accumulation and redemption status.

[0061] like Figure 1-Figure 2 As shown, in the step S2, the collected data is cleaned and outliers are processed;

[0062] Specifically, in step S2, outlier processing is adopted. For the customer service satisfaction data x1 and the customer membership registration data x2, the Z-scores of the customer service satisfaction data and the customer membership registration data are calculated. The Z-score is the difference between the data and the mean divided by the standard deviation;

[0063] The calculation formula is:

[0064]

[0065] Among them, x1 is the original value of the customer service satisfaction data, μ1 is the average value of the customer service satisfaction data, and σ1 is the standard deviation of the customer service satisfaction data;

[0066]

[0067] Among them, x2 is the original value of customer member registration data, μ2 is the mean value of customer member registration data, and σ2 is the standard deviation of customer member registration data;

[0068] Define outliers as values ​​with a Z-score greater than 3 or less than -3, and then remove or replace the outliers;

[0069] For outliers, use the absolute value of the Z-score to determine the extent of the outlier:

[0070] abs_z=|z|

[0071] Then, delete or replace the values ​​where abs_z>3.

[0072] like Figure 1-Figure 3As shown, in the step S3, the data association uses primary keys and foreign keys in the database to establish the association relationship between the data, associate the online and offline data, and use data matching technology and fuzzy matching to identify and associate similar data records;

[0073] Specifically, in the S3 step, primary key and foreign key associations are used, and similar data records are identified through fuzzy matching in data matching technology;

[0074] Create a customer store data set table (Table1) and an online data set table (Table2). The specific steps are as follows:

[0075] S301, establish database:

[0076] In Table1, create a unique primary key ID for each record;

[0077] In Table2, create a unique primary key ID for each record;

[0078] In Table1, create a foreign key ID that references the primary key ID in Table2;

[0079] S302, data import:

[0080] Import the cleaned data into Table1 and Table2, and ensure that the ID fields in Table1 and Table2 are unique;

[0081] S303, data matching:

[0082] Use the fuzzy matching method in data matching technology to compare the ID fields in Table1 and Table2. The fuzzy matching uses Jaccard similarity to calculate the Jaccard similarity:

[0083] For each ID in Table1 (referred to as ID1), calculate the Jaccard similarity between it and all IDs in Table2 (referred to as ID2);

[0084] Assume that there are n IDs in Table1 and m IDs in Table2;

[0085] For each ID1, find the most similar ID2 in Table2, that is, the ID2 with the highest Jaccard similarity;

[0086] The calculation formula is:

[0087]

[0088] Where A is the set corresponding to ID1, B is the set corresponding to ID2, |A∩B| is the size of the intersection of sets A and B, and |A∪B| is the size of the union of sets A and B;

[0089] The value of Jaccard similarity is between 0 and 1. When the two sets are exactly the same, the Jaccard similarity is 1. When the two sets have no intersection, the Jaccard similarity is 0. The closer the Jaccard is to 1, the more similar the two sets are.

[0090] S304, select ID2 with the highest similarity:

[0091] For each ID1, find the ID2 with the highest Jaccard similarity and select the largest Jaccard similarity value;

[0092] S305. Related data:

[0093] For successfully matched ID1 and ID2, use foreign key association to combine them into a complete data record. In the database, point the foreign key ID in Table1 to the matched ID2 in Table2.

[0094] Through the above steps, the primary key and foreign key are associated in the database, and similar data records are identified through Jaccard similarity in data matching technology.

[0095] like Figure 1-Figure 3 As shown, in step S4, features related to purchase intention are extracted from the data set, and the most representative features are selected using correlation analysis;

[0096] Specifically, in step S4, correlation analysis is performed to select the most representative features:

[0097] The calculation formula of Pearson correlation coefficient in correlation analysis is:

[0098]

[0099] in:

[0100] r: Pearson correlation coefficient;

[0101] Covariance (x, y): the covariance between feature x and purchase intention y;

[0102] StdDev(x): standard deviation of feature x;

[0103] Standard deviation (y): standard deviation of purchase intention y;

[0104] The formula for calculating covariance is:

[0105]

[0106] in:

[0107] xi: the i-th observation value of feature x;

[0108] yi: the i-th observation of purchase intention;

[0109] The mean value of feature x;

[0110] Mean value of purchase intention;

[0111] n: the number of observations;

[0112] The formula for calculating the standard deviation is:

[0113]

[0114] in:

[0115] xi: the i-th observation value of feature x;

[0116] The mean value of feature x;

[0117] n: the number of observations;

[0118] The specific calculation process is:

[0119] The following data is preset:

[0120] Observed values ​​of feature x: 81, 78, 84, 86, 82, 79, 84, 85, 87, 83;

[0121] Observed values ​​of purchase intention y: 82, 81, 84, 89, 81, 83, 85, 87, 91, 84;

[0122] Covariance(x,y) calculation:

[0123]

[0124]

[0125] Standard deviation calculation:

[0126]

[0127]

[0128] Pearson correlation coefficient calculation:

[0129]

[0130] The closer the absolute value of the Pearson correlation coefficient is to 1, the stronger the correlation is. The calculated result is 0.77, which means that feature x is positively correlated with purchase intention y. As the value of feature x increases, the value of purchase intention y increases, indicating that there is a strong positive linear correlation between feature x and purchase intention y.

[0131] like Figure 1-Figure 3 As shown, in the step S5, the model construction and training use cross-validation technology to train and tune the model;

[0132] Specifically, the settings for the cross-validation technique used in step S5 are:

[0133] Data preparation ensures that the dataset has been cleaned and divided into feature set (X) and label set (y);

[0134] Parameter settings determine the parameters of the logistic regression model, use the regularization strength (C), the maximum number of iterations, select the K value for K-fold cross validation, and set K=10;

[0135] K-fold cross validation divides the feature set (X) and label set (y) into K subsets of equal size, repeats K times, and uses one subset as the test set each time, and the remaining K-1 subsets as the training set. In each iteration, the training set is used to train the logistic regression model. The accuracy is: (the number of samples predicted correctly) / (the total number of samples). In cross validation, the training set is divided into K groups, and training and testing are performed in turn. The accuracy of each test is calculated.

[0136] like Figure 1-Figure 3 As shown, in the step S6, the model is trained and tuned through the step S5, customer behavior data is continuously collected, and the model prediction results are fed back to the sales staff in real time so that they can adjust the sales strategy in time;

[0137] Specifically, in step S6, the model is trained and tuned through step S5, and customer behavior data is continuously collected: variable data of customer behavior is continuously collected, the association relationship between data is established in the database through primary keys and foreign keys, and Pearson correlation coefficient analysis is used to select the most representative features, and the model is trained and tuned to output the customer's real-time purchase intention data, and the model prediction results are displayed to sales staff through dashboards, push notifications, and instant messaging tools to ensure that sales staff can promptly understand the customer's real-time purchase intention classification and adjust sales strategies.

[0138] like Figure 1-Figure 3 As shown, the present invention solves the shortcomings of the prior art that the single-dimensional data cannot meet the rapidly changing needs of the modern market in terms of analysis accuracy, and works as follows:

[0139] Collect the dataset:

[0140] Collect customer service satisfaction data x1, customer membership registration data x2, and retrieve customer data by calling the storage information of the tablet device;

[0141] Data processing:

[0142] The collected data is cleaned and outlier processing is performed to calculate the Z-scores of customer service satisfaction data and customer membership registration data. The Z-score calculation formula is:

[0143]

[0144] Where x is the original value, μ is the mean, and σ is the standard deviation;

[0145] Define outliers as values ​​with a Z-score greater than 3 or less than -3, and then remove or replace the outliers;

[0146] Data association and matching:

[0147] Create a customer store data set table (Table1) and an online data set table (Table2), and create a unique primary key ID for each record in each table. Calculate the Jaccard similarity using the following formula:

[0148]

[0149] Among them, A is the set corresponding to ID1, B is the set corresponding to ID2, |A∩B| is the size of the intersection of sets A and B, and |A∪B| is the size of the union of sets A and B;

[0150] Use primary key and foreign key association to associate data: For each ID1, find the most similar ID2 in Table2 by calculating the Jaccard similarity, select the ID2 with the largest Jaccard similarity value, and use the foreign key association through ID1 and ID2 to combine them into a complete data record;

[0151] Correlation analysis:

[0152] Extract features related to purchase intention, use correlation analysis to select the most representative features, and use the Pearson correlation coefficient calculation formula in correlation analysis:

[0153]

[0154] in:

[0155] r: Pearson correlation coefficient;

[0156] Covariance (x, y): the covariance between feature x and purchase intention y;

[0157] StdDev(x): standard deviation of feature x;

[0158] Standard deviation (y): standard deviation of purchase intention y;

[0159] Covariance calculation result: (x, y) = 7.62;

[0160] The standard deviation of feature x is calculated as follows: (x) = 3.11;

[0161] The standard deviation of purchase intention y is calculated as follows: (y) = 3.19;

[0162] Pearson correlation coefficient calculation results:

[0163] The absolute value of the Pearson correlation coefficient is 0.77, indicating that there is a strong positive linear correlation between feature x and purchase intention y;

[0164] Model building and training:

[0165] Use cross-validation technology, select the K value of K-fold cross-validation, set K=10, and train and tune the model;

[0166] Output feedback:

[0167] Continuously collect variable data on customer behavior, ensure that the data is cleaned, input the data into a trained and tuned model, and display the customer's real-time purchase intention results predicted by the model to sales staff through dashboards, push notifications, and instant messaging tools to ensure that sales staff can promptly understand the customer's real-time purchase intention classification and adjust sales strategies.

[0168] So far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it is easy for those skilled in the art to understand that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will fall within the protection scope of the present invention.

[0169] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and rules of the present invention shall be included in the protection scope of the present invention.

Claims

1. A real-time analysis method for mobile customer purchase intention classification, characterized by: S1. Collect customer store data sets and online data sets. The store data sets include customer service satisfaction data x1, and the online data sets include customer membership registration data x2; S2, clean the data and deal with outliers; S3. Data association uses primary keys and foreign keys in the database to establish associations between data, associate online and offline data, and use data matching technology and fuzzy matching to identify and associate similar data records; S4. Extract features related to purchase intention from the dataset and use correlation analysis to select the most representative features; S5. Model construction and training Use cross-validation technology to train and tune the model; S6. Train and tune the model through step S5, continuously collect customer behavior data, and provide real-time feedback of the model prediction results to sales staff so that they can adjust their sales strategies in a timely manner.

2. A real-time analysis method for mobile customer purchase intention classification according to claim 1, characterized in that: The customer service satisfaction data x1 includes data on satisfaction questionnaires filled out by customers and customer satisfaction feedback in real time; the customer member registration data x2 includes customer member registration information, points accumulation and redemption status.

3. The method for real-time analysis of mobile customer purchase intention classification according to claim 1, characterized in that: The outliers are processed by calculating the Z-score of the customer service satisfaction data x1 and the customer membership registration data x2, where the Z-score is the difference between the data and the mean divided by the standard deviation; The calculation formula is: Where x is the original value, μ is the mean, and σ is the standard deviation; Define outliers as values ​​with a Z-score greater than 3 or less than -3, and then remove or replace the outliers.

4. The method for real-time analysis of mobile customer purchase intention classification according to claim 1, characterized in that: The primary key and foreign key are associated and similar data records are identified through fuzzy matching in data matching technology; Create a customer store data set table Table1 and an online data set table Table2: Create a database: In Table1, create a unique primary key ID for each record; In Table2, create a unique primary key ID for each record; In Table1, create a foreign key ID that references the primary key ID in Table2; Data import: Import the cleaned data into Table1 and Table2, and ensure that the ID fields in Table1 and Table2 are unique; Data Matching: Use the fuzzy matching method in data matching technology to compare the ID fields in Table1 and Table2. The fuzzy matching uses Jaccard similarity to calculate the Jaccard similarity: For each ID in Table1, denoted as ID1, calculate the Jaccard similarity between it and all IDs in Table2, denoted as ID2; Assume that there are n IDs in Table1 and m IDs in Table2; For each ID1, find the ID2 with the highest Jaccard similarity in Table2; The calculation formula is: Where A is the set corresponding to ID1, B is the set corresponding to ID2, |A∩B| is the size of the intersection of sets A and B, and |A∪B| is the size of the union of sets A and B; The value of Jaccard similarity is between 0 and 1. The closer Jaccard is to 1, the more similar the two sets are. Select ID2 with the highest similarity: For each ID1, find the ID2 with the highest Jaccard similarity and select the largest Jaccard similarity value; Related data: For successfully matched ID1 and ID2, use foreign key association to combine into data records. In the database, point the foreign key ID in Table1 to the matched ID2 in Table2. The database uses primary key and foreign key associations, and uses Jaccard similarity in data matching technology to identify similar data records.

5. The method for real-time analysis of mobile customer purchase intention classification according to claim 1, characterized in that: The correlation analysis selects the most representative features: The calculation formula of Pearson correlation coefficient in correlation analysis is: in: r: Pearson correlation coefficient; Covariance (x, y): the covariance between feature x and purchase intention y; StdDev(x): standard deviation of feature x; Standard deviation (y): standard deviation of purchase intention y; The formula for calculating covariance is: The formula for calculating covariance is: in: xi: the i-th observation value of feature x; yi: the i-th observation of purchase intention; The mean value of feature x; Mean value of purchase intention; n: the number of observations; The formula for calculating the standard deviation is: in: xi: the i-th observation value of feature x; The mean value of feature x; n: The number of observations.

6. The method for real-time analysis of mobile customer purchase intention classification according to claim 1, characterized in that: The settings for the cross validation technique are: Data preparation ensures that the dataset has been cleaned and divided into feature set X and label set y; Parameter settings determine the parameters of the logistic regression model, use the regularization strength C, the maximum number of iterations, select the K value for K-fold cross validation, and set K=10; K-fold cross validation divides the feature set X and label set y into K equal-sized subsets, repeating K times, each time using one subset as the test set and the remaining K-1 subsets as the training set. In each iteration, the training set is used to train the logistic regression model.

7. The method for real-time analysis of mobile customer purchase intention classification according to claim 1, characterized in that: The model is trained and tuned through step S5, and customer behavior data is continuously collected: variable data of customer behavior is continuously collected, the data is input into the trained and tuned model, and the customer's real-time purchase intention results predicted by the model are fed back to the salesperson in real time.

Images