A Marketing Visualization Analysis Method and System for Big Data

Abstract

This invention relates to the field of marketing analytics technology, specifically disclosing a marketing visualization analysis method and system for big data. The method includes: acquiring marketing promotion data and order quantities for a product; performing numerical analysis on the marketing promotion data and order quantities to determine the marketing analysis period; performing cluster analysis on the marketing performance values ​​and marketing push frequency of competing products corresponding to the marketing analysis period to identify competing products; subsequently determining whether the marketing effectiveness ratio of the competing products is higher than that of the product; if higher, constructing a predictive analysis model using different analytical model algorithms; optimizing the product's marketing promotion strategy; and selecting the predictive analysis model constructed by the best analytical model algorithm. The system includes a data acquisition module, a competitive analysis module, an effectiveness analysis module, a model construction module, and a model determination module, which can effectively optimize marketing decisions.

Description

Technical Field

[0001] This invention relates to the field of marketing analytics technology, and more specifically to a marketing visualization analytics method and system for big data. Background Technology

[0002] In today's wave of digital marketing, traditional marketing technologies are struggling to cope with massive amounts of data and complex behavioral analysis, failing to meet users' personalized needs and the requirements of precision marketing.

[0003] A Chinese patent application with announcement number CN118172112B discloses a marketing strategy intelligent analysis system and method based on big data, including: collecting user behavior data from authorized data sources, formulating personalized marketing strategies for users based on users' personalized temporary tags, formulating block marketing strategies based on block consumption trend tag data, and pushing overall block-based marketing data based on the analyzed block consumption trend tags and block consumption habits.

[0004] The aforementioned technologies, which adjust strategies based on marketing data dwell time and consumption data, lack a dynamic response mechanism to external factors such as sudden market changes and competitors' new strategies. However, by analyzing the effective marketing frequency of competing products and using multiple models to predict marketing optimization strategies, it is possible to respond more flexibly to market changes.

[0005] The aforementioned technologies do not fully utilize various advanced data analysis models. They only use recurrent neural network models to predict regional consumption trends, which may not be comprehensive enough in predicting the number and timing of marketing pushes. If autoregressive integral moving average model algorithms, multilayer perceptron model algorithms, and decision tree regression model algorithms are used for prediction, and the best model is selected through model evaluation methods, the accuracy and reliability of the prediction can be improved.

[0006] Therefore, the present invention provides a marketing visualization analysis method and system for big data. Summary of the Invention

[0007] The purpose of this invention is to provide a marketing visualization analysis method and system for big data, in order to solve the problems mentioned above.

[0008] The objective of this invention can be achieved through the following technical solutions:

[0009] A marketing visualization analysis method for big data includes the following steps:

[0010] Step 1: Obtain the product's marketing and promotion data and order quantity, and perform numerical analysis on the marketing and promotion data and order quantity to determine the marketing analysis period;

[0011] Step 2: Perform cluster analysis on the marketing performance values ​​and marketing push frequency of competing products corresponding to the marketing analysis period to determine the competing products;

[0012] Step 3: Statistically analyze the number of marketing pushes for the products analyzed by competitors, obtain the marketing effectiveness probability, and perform information entropy value analysis to determine whether the marketing effectiveness ratio of the products analyzed by competitors is higher than that of the products analyzed by competitors.

[0013] Step 4: If the result is higher, use different analytical model algorithms to build a predictive analysis model, output the predicted number of marketing pushes and push time, and optimize the product's marketing and promotion strategy.

[0014] Step 5: Based on different analytical model algorithms, perform numerical analysis on the optimized marketing promotion strategy, and select the best analytical model algorithm to construct the predictive analysis model.

[0015] As a further technical solution of the present invention: the method for determining the marketing analysis period is as follows:

[0016] Obtain the marketing performance value for each monitoring period, and then calculate the ratio between the marketing performance value for each monitoring period and the number of marketing pushes for that period to obtain the marketing conversion value.

[0017] The marketing conversion values ​​for each monitoring period are sorted, and the monitoring period with the highest marketing conversion value is selected as the marketing analysis period.

[0018] As a further technical solution of the present invention: the marketing performance value is obtained in the following way:

[0019] Obtain marketing and promotion data and order volume for different monitoring periods within the monitoring period;

[0020] Marketing and promotion data, including the number of times a product is accessed, favorited, and pushed to e-commerce platforms;

[0021] The visit conversion ratio is obtained by comparing the number of visits with the number of orders placed in each monitoring period.

[0022] The conversion rate of favorites is calculated by comparing the number of times a favorite is added to the list with the number of orders placed.

[0023] Based on the visit conversion rate, collection conversion rate and number of marketing pushes during the monitoring period, a marketing conversion matrix is ​​constructed and standardized to obtain a standard marketing conversion matrix;

[0024] Based on the standard marketing conversion matrix, the collection conversion ratio and visit conversion ratio for each monitoring period are weighted and summed to obtain the marketing characterization value.

[0025] As a further technical solution of the present invention: the method for determining the competitive analysis product is as follows:

[0026] Establish a marketing analysis coordinate system, using the marketing performance value and number of marketing pushes of competing products as coordinates, and using the four quadrants of the coordinate system as distinctions to divide competing products into four categories;

[0027] Based on the marketing analysis table and competing products categorized into four groups, a marketing classification model is constructed using a clustering algorithm to identify the competing products.

[0028] As a further technical solution of the present invention: the marketing analysis table is constructed as follows:

[0029] Obtain the marketing conversion values ​​of similar competing products during the marketing analysis period to obtain the competitor conversion value;

[0030] Obtain the competitor conversion value of all competing products, compare the competitor conversion value with the marketing conversion value, and select the competing products whose competitor conversion value is higher than the marketing conversion value;

[0031] Obtain the marketing performance metrics and number of marketing pushes for competing products whose conversion rates are higher than the marketing conversion rates, and construct a marketing analysis table.

[0032] As a further technical solution of the present invention: the method for determining whether the marketing effectiveness ratio of the competitive analysis product is higher than that of the product is as follows:

[0033] Calculate the information analysis entropy value of products based on the marketing effectiveness probability calculation;

[0034] If the information analysis entropy value is lower than the preset analysis entropy threshold, then the marketing effectiveness ratio of the competitive analysis product is higher than that of the product.

[0035] As a further technical solution of the present invention: the method for obtaining the marketing effectiveness ratio is as follows:

[0036] Obtain the number of visits and favorites for the products analyzed in the competitive analysis during the marketing analysis period, and divide the marketing analysis period into multiple time periods to obtain the analysis time periods;

[0037] The effective marketing count is obtained by summing and averaging the number of visits and favorites of the product at each analysis time during the marketing analysis period.

[0038] The effective marketing count is calculated by comparing the number of effective marketing campaigns with the number of marketing pushes.

[0039] The method for obtaining the marketing effectiveness probability is as follows:

[0040] Obtain the marketing effectiveness ratio of the competitive analysis product at each analysis time, and obtain the number of times the marketing effectiveness ratio of the competitive analysis product at each analysis time is higher than that of the product, thus obtaining the number of times the boundary is crossed at each time.

[0041] The probability of effective marketing is obtained by comparing the number of times the time boundary is crossed with the total number of times analyzed within the marketing analysis period.

[0042] As a further technical solution of the present invention: the optimal analysis model algorithm is determined as follows:

[0043] Based on the predicted values ​​of each analytical model algorithm, the negative ideal solution M is obtained. - distance d - k The distance d of the ideal solution + k Calculate the relative closeness C of the predicted values ​​for each analysis model algorithm. k Through the formula: Obtain the relative proximity C k ;

[0044] The relative closeness C of the predicted values ​​of each analysis model algorithm. k Sort the data and determine the relative proximity C. k The maximum value corresponds to the value of k, where k is the number of the different analysis model algorithm, and k=0 indicates that no analysis model algorithm was used;

[0045] If the relative proximity C k The maximum value corresponds to a non-zero k value, which determines the relative proximity C. k The analytical model algorithm corresponding to the largest numerical value is taken as the best analytical model algorithm, and a predictive analytical model is constructed according to the best analytical model algorithm.

[0046] As a further technical solution of the present invention: the predicted value of each analysis model algorithm is converted to the negative ideal solution M. - distance d - k The distance d of the ideal solution + k The method of obtaining it is:

[0047] Obtain the marketing effectiveness ratio and marketing conversion value after optimization by different analytical model algorithms, and label the marketing effectiveness ratio obtained by different analytical model algorithms as 'a'. k Marketing conversion value is marked as b k ;

[0048] Marketing effectiveness ratio a obtained based on different analytical model algorithms k Marketing conversion value b k The superior-inferior solution distance method is used to construct a model and determine the matrix M;

[0049] Obtain the positive ideal solution M of the normalized model determination matrix + and negative ideal solution M - ;

[0050] The predicted value of each analysis model algorithm is calculated to the ideal solution M using the distance formula. + distance d + k Negative ideal solution M - distance d - k .

[0051] A marketing visualization and analysis system for big data includes the following modules:

[0052] Data acquisition module: Used to acquire marketing and promotion data and order quantity of products, and to perform numerical analysis on marketing and promotion data and order quantity to determine the marketing analysis period;

[0053] Competition Analysis Module: Used to perform cluster analysis on the marketing performance values ​​and marketing push frequency of competing products during the marketing analysis period to identify competing products;

[0054] Effectiveness Analysis Module: Used to statistically analyze the number of marketing pushes for products analyzed by competitors, obtain the marketing effectiveness probability, and perform information entropy value analysis to determine whether the marketing effectiveness ratio of the products analyzed by competitors is higher than that of the products analyzed by competitors.

[0055] Model building module: Used to build predictive analysis models using different analytical model algorithms, outputting the predicted number of marketing pushes and push time, and optimizing the marketing and promotion strategies of products;

[0056] Model determination module: Used to select the best predictive analysis model constructed by the analytical model algorithm.

[0057] The beneficial effects of this invention are:

[0058] (1) By constructing a marketing conversion matrix, the marketing analysis period is selected, and big data analysis and clustering algorithms are used to conduct numerical analysis on competing products to obtain the marketing representation value and push frequency of competing products, and to determine the competing products for analysis. This allows companies to understand the advantages and disadvantages of their competitors, which is conducive to formulating targeted competitive strategies.

[0059] (2) If the marketing effectiveness ratio of the competitive analysis product is higher than that of the product, a predictive analysis model is constructed using a variety of model algorithms such as autoregressive integral moving average, multilayer perceptron, and decision tree regression. Based on the time pattern of marketing push and the influence of complex nonlinear relationships on marketing, the number and time of marketing push of the product are predicted. Based on these prediction results, the enterprise adjusts its marketing promotion strategy, optimizes the push time node, and reasonably adjusts the push frequency, etc., and selects the best model algorithm, which is conducive to improving the utilization efficiency of marketing resources and optimizing marketing effect. Attached Figure Description

[0060] The invention will now be further described with reference to the accompanying drawings.

[0061] Figure 1 This is a flowchart of marketing data acquisition and analysis provided in Embodiment 1 of the present invention.

[0062] Figure 2 This is a flowchart of the model construction and analysis provided in Embodiment 2 of the present invention.

[0063] Figure 3 This is a module diagram of a marketing visualization analysis system for big data according to the present invention. Detailed Implementation

[0064] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0065] Example 1

[0066] Please see Figure 1 As shown, a marketing visualization analysis method for big data includes the following steps:

[0067] Step 1: Obtain the product's marketing and promotion data and order quantity, and perform numerical analysis on the marketing and promotion data and order quantity to determine the marketing analysis period;

[0068] In some embodiments, big data analytics is used to obtain marketing and promotion data and order volume of the same product from e-commerce platforms during different monitoring periods within the monitoring cycle.

[0069] Marketing and promotion data includes the number of times a product is accessed, added to favorites, and pushed to e-commerce platforms;

[0070] The number of visits to each monitoring period is compared with the number of orders placed to obtain the visit conversion ratio, which is denoted as q.i , where i is the monitoring period number, and the value of i ranges from [1, n];

[0071] The ratio of the number of times an item is added to favorites to the number of orders placed is used to obtain the favorites-to-conversion ratio, which is then denoted as b. i The number of marketing pushes is marked as c. i ;

[0072] It should be noted that the conversion rate of collections, conversion rate of visits, and number of marketing pushes are all calculated after removing the units from the total value.

[0073] A marketing conversion matrix Q is constructed based on the visit conversion rate, collection conversion rate, and the number of marketing pushes during the monitoring period. ix3 ;

[0074] For example, the marketing conversion matrix Q ix3 for q1 and b1 represent the visit conversion rate and collection conversion rate of the first e-commerce platform, respectively;

[0075] Marketing conversion matrix Q ix3 Standardization is performed to obtain a standard marketing conversion matrix. ;

[0076] Specifically, through the formula: For access conversion ratio q i Standardization processing, q max q min The conversion rate q is respectively i The maximum and minimum values;

[0077] Through the formula: The conversion rate of collections to b i Standardization processing, b max b min The conversion rate of collections is b. i The maximum and minimum values;

[0078] Through the formula: c of marketing push notifications i Standardization processing, c max c min c represents the number of marketing pushes. i The maximum and minimum values;

[0079] Collection conversion ratio after normalization , visit conversion rate Marketing push notification frequency c i Building a Standard Marketing Conversion Matrix ;

[0080] Based on standard marketing conversion matrix Collection conversion rate for each monitoring period , visit conversion rate The marketing performance value is obtained by performing a weighted summation.

[0081] Obtain the marketing performance value for each monitoring period, and then calculate the ratio between the marketing performance value for each monitoring period and the number of marketing pushes for that period to obtain the marketing conversion value.

[0082] The marketing conversion values ​​for each monitoring period are sorted in descending order, and the monitoring period with the highest marketing conversion value is selected as the marketing analysis period.

[0083] The benefits of this step are: to identify key marketing periods, to construct a marketing conversion matrix and calculate marketing performance values ​​by acquiring marketing promotion data and order volume, and to determine the period with the highest marketing conversion value as the marketing analysis period. This allows companies to focus on the period with the best marketing results, grasp key marketing nodes, and provide a time dimension basis for subsequent analysis and strategy formulation.

[0084] To improve data utilization efficiency, marketing data is systematically processed and analyzed to filter out the most valuable marketing time periods, avoiding wasting resources on invalid or low-value data, thus improving the efficiency of data processing and analysis.

[0085] Step 2: Perform cluster analysis on the marketing performance values ​​and marketing push frequency of competing products corresponding to the marketing analysis period to determine the competing products;

[0086] Based on the marketing analysis period, big data analysis technology is used to obtain the marketing conversion value of similar competing products during the marketing analysis period, thus obtaining the competitor conversion value;

[0087] Obtain the competitor conversion value of all competing products, compare the competitor conversion value with the marketing conversion value, and select the competing products whose competitor conversion value is higher than the marketing conversion value;

[0088] Obtain the marketing performance metrics and number of marketing pushes for competing products whose conversion values ​​are higher than their marketing conversion values, and construct a marketing analysis table. Each row of the marketing analysis table represents a competing product, with the first column recording the marketing performance metrics and the second column recording the number of marketing pushes.

[0089] A marketing analysis coordinate system is established using marketing performance value and marketing push frequency as two axes. The marketing performance value and marketing push frequency of competing products are used as coordinates, and the four quadrants of the coordinate system are used as distinctions to divide competing products into four categories.

[0090] Based on the marketing analysis table and competing products divided into four categories, a marketing classification model is constructed using a clustering algorithm to determine the competing products.

[0091] Specifically, the marketing analysis model is constructed as follows:

[0092] S1. Perform data preprocessing on the marketing analysis table, removing outliers and filling in missing values;

[0093] Specifically, check if there are missing or outlier values ​​in the marketing analysis table; for missing values, determine the mean of the column in the marketing analysis table corresponding to the missing value, and use the mean to fill in the missing values.

[0094] For outliers, obtain the standard deviation of the column in the marketing analysis table and remove data that deviates from the standard deviation by more than 3 times.

[0095] S2. Determine the numerical values ​​of the clusters and use the K-Means clustering algorithm to build a marketing analysis model;

[0096] SS1. Determine the cluster size as 4. From the marketing analysis table, randomly select 4 data points from the four regions of the initial classification of competing products as the initial cluster centers.

[0097] SS2. Using the remaining product marketing characteristics and ad push counts from the marketing analysis table as data points, the Euclidean distance formula is used to assign the data points to the initial cluster centers, resulting in 4 initial clusters.

[0098] SS3. For the initial cluster, calculate the average of the marketing representation value and the number of marketing pushes for all data points in each initial cluster and use them as the new cluster centers.

[0099] SS4. Repeat steps SS2 and SS3 to calculate the Euclidean distance between the new cluster centers and the cluster centers of the previous iteration, until the distance is lower than the distance termination threshold.

[0100] S3. Identify competitive analysis products based on limiting conditions;

[0101] Condition 1: Minimize the distance from each cluster center to the origin;

[0102] Condition 2: In the competitive analysis of cluster centers, the number of marketing pushes for a product is lower than the number of marketing pushes for another product.

[0103] The benefits of this step are: it clarifies competitors; through cluster analysis, combined with marketing analysis tables and the constructed marketing classification model, it identifies competitive products, enabling companies to clearly identify truly competitive products during key marketing periods and thus clarify competitors.

[0104] It helps in the formulation of competitive strategies. Understanding the marketing performance and number of times products are promoted in competitive analysis helps companies understand the strengths and weaknesses of their competitors, thus providing support for the formulation of competitive strategies.

[0105] The technical solution of this embodiment is as follows: obtain the marketing promotion data and order quantity of the product, perform numerical analysis on the marketing promotion data and order quantity to determine the marketing analysis period, perform cluster analysis on the marketing performance value and marketing push frequency of competing products during the marketing analysis period to determine the competing products, and analyze the marketing data of the competing products, which is conducive to providing data support for the formulation of marketing strategies.

[0106] Example 2

[0107] like Figure 2 As shown, a marketing visualization analysis method for big data also includes the following steps:

[0108] Step 3: Statistically analyze the number of marketing pushes for the products analyzed by competitors, obtain the marketing effectiveness probability, and perform information entropy value analysis to determine whether the marketing effectiveness ratio of the products analyzed by competitors is higher than that of the products analyzed by competitors.

[0109] Obtain the number of visits and favorites for products analyzed in the competitive analysis period;

[0110] The marketing analysis period is divided into multiple time periods to obtain the analysis time periods;

[0111] The effective marketing count is obtained by summing and averaging the number of visits and favorites of the product at each analysis time during the marketing analysis period.

[0112] The effective marketing count is calculated by comparing the number of effective marketing campaigns with the number of marketing pushes.

[0113] Obtain the marketing effectiveness ratio (CPR) of the product. If the CPR of the competitor product at each analysis time is higher than that of the product, obtain the marketing effectiveness probability p for each competitor product. j , where j is the number of the competitive analysis product, and the value of j ranges from [1, m];

[0114] Specifically, the probability of marketing effectiveness p j The method of obtaining it is:

[0115] Obtain the marketing effectiveness ratio of the competitive analysis product at each analysis time point, calculate the number of times the competitive analysis product's marketing effectiveness ratio is higher than the product's marketing effectiveness ratio, and obtain the number of times the boundary is crossed at each time point;

[0116] The ratio of the number of times the time boundary was crossed to the total number of analyzed times within the marketing analysis period yields the marketing effectiveness probability p. j ;

[0117] Through the formula: Obtain the entropy value H(p) of the competitive analysis product;

[0118] The information analysis entropy value H(p) is compared with the preset analysis entropy threshold to determine whether the marketing effectiveness ratio of the competing product is higher than that of the competing product.

[0119] If the information analysis entropy value H(p) is higher than the preset analysis entropy threshold, it indicates that the distribution of the marketing effectiveness probability of the competitive analysis product has a high degree of uncertainty, and it cannot be judged that the marketing effectiveness ratio of the competitive analysis product is higher than that of the product.

[0120] If the information analysis entropy value H(p) is lower than the preset analysis entropy threshold, it indicates that the distribution of the marketing effectiveness probability of the competitive analysis product is relatively concentrated, and it can be judged that the marketing effectiveness ratio of the competitive analysis product is higher than that of the product.

[0121] The beneficial effects of this step are: to evaluate the marketing effectiveness of competing products, to calculate the marketing effectiveness probability and information entropy value by statistically analyzing the number of marketing pushes of competing products, and to determine whether their marketing effectiveness ratio is higher than that of competing products, thus evaluating the marketing effectiveness of competing products;

[0122] Provide a basis for strategy adjustments: If competitors' marketing effectiveness ratio is higher, it indicates that their marketing strategy may be more effective, providing direction for subsequent strategy optimization.

[0123] Step 4: If the marketing effectiveness ratio of the competing product is higher than that of the competing product, use different analysis model algorithms to build a predictive analysis model, output the predicted number of marketing pushes and push time, and optimize the marketing promotion strategy of the product.

[0124] If the marketing effectiveness ratio of the product analyzed by the competitors is higher than that of the product analyzed by the competitors, obtain the number of marketing pushes and the push time of the product analyzed by the competitors during the marketing analysis period as the marketing prediction dataset;

[0125] Based on the marketing forecast dataset, a predictive analysis model is established using the autoregressive integral moving average model algorithm, the multilayer perceptron model algorithm, and the decision tree regression model algorithm to predict the number of marketing pushes and the push time of products.

[0126] It should be noted that the autoregressive integral moving average model algorithm is used to capture the patterns of marketing push timing and frequency over time. Marketing push data has a temporal sequence and a certain periodicity. The autoregressive integral moving average model can predict the timing and frequency of future marketing pushes by analyzing the autocorrelation, moving average, and other relationships in historical data.

[0127] The multilayer perceptron model algorithm can handle complex nonlinear relationships. The effectiveness of marketing push is affected by a variety of factors, and there may be complex nonlinear relationships between these factors. By constructing multiple hidden layers, the multilayer perceptron model can automatically learn complex patterns in the data and predict the timing and frequency of marketing pushes.

[0128] The decision tree regression model algorithm can classify and predict marketing pushes based on market demand and competitors' push strategies at different time periods. The structure of the decision tree can intuitively show the impact path of different factors on marketing pushes.

[0129] Specifically, the predictive analytics model is constructed as follows:

[0130] A1. Perform data preprocessing on the marketing dataset to remove invalid values ​​and fill in missing values;

[0131] A2. Divide the preprocessed marketing dataset into training and testing sets in a 7:3 ratio;

[0132] A3. Input the divided dataset into the autoregressive integral moving average model, the multilayer perceptron model, and the decision tree regression model to build a predictive analysis model and output the number of marketing pushes and the marketing push time.

[0133] Based on predictive analytics models, different analytics model algorithms output the number of marketing pushes and the duration of marketing pushes, and optimize the number of marketing pushes and the duration of marketing pushes for products in the next monitoring period.

[0134] The benefits of this step are: ensuring model reliability; calculating the distance between the predicted value of each analysis model algorithm and the positive and negative ideal solutions to obtain the relative closeness; selecting the best model algorithm based on the relative closeness; and quantitatively evaluating different model algorithms to ensure that the finally selected model algorithm has reliability and accuracy in predicting marketing effectiveness ratio and marketing conversion value.

[0135] Enhancing the scientific rigor of marketing strategies: Using optimal model algorithms to build predictive analytics models can provide a more scientific and accurate basis for enterprises to formulate marketing strategies, further improving the scientific nature and effectiveness of enterprise marketing decisions.

[0136] Step 5: Based on different analytical model algorithms, perform numerical analysis on the optimized marketing promotion strategy, and select the best analytical model algorithm to construct the predictive analysis model.

[0137] Obtain the marketing effectiveness ratio and marketing conversion value after optimization by different analytical model algorithms, and label the marketing effectiveness ratio obtained by different analytical model algorithms as 'a'. k Marketing conversion value is marked as b k, where k is the number of different analysis model algorithms, k=0, 1, 2, 3, where k=0 indicates that no analysis model algorithm was used;

[0138] Marketing effectiveness ratio a obtained based on different analytical model algorithms k Marketing conversion value b k The superior-inferior solution distance method is used to construct a model and determine the matrix M;

[0139] Specifically, the model determines the matrix. ;

[0140] The model determination matrix M is normalized. The specific normalization method is as follows:

[0141] Normalized , ;

[0142] The normalized model determination matrix is ​​obtained. ;

[0143] Obtain the positive ideal solution M of the normalized model determination matrix + =(a + k b + k ), through the formula: , , get a + k b + k ;

[0144] Obtain the negative ideal solution M of the normalized decision matrix - =(a - k b - k ), through the formula: , Get a - k b - k ;

[0145] Calculate the predicted value of each analysis model algorithm to the positive ideal solution M. + distance d + k Through the formula: Obtain the predicted value of each analysis model algorithm to the positive ideal solution M + distance d + k ;

[0146] Calculate the predicted value of each analysis model algorithm to the negative ideal solution M. -distance d - k Through the formula: Obtain the predicted value of each analysis model algorithm to the negative ideal solution M - distance d - k ;

[0147] Based on the predicted values ​​of each analytical model algorithm, the negative ideal solution M is obtained. - distance d - k The distance d of the ideal solution + k Calculate the relative closeness C of the predicted values ​​for each analysis model algorithm. k Through the formula: Obtain the relative proximity C k ;

[0148] The relative closeness C of the predicted values ​​of each analysis model algorithm. k Sort the data in descending order to determine the relative closeness C. k The k value corresponding to the maximum value;

[0149] If the relative proximity C k If the maximum value corresponds to k=0, it indicates that the marketing effectiveness ratio and marketing conversion value are good before the analysis model algorithm is optimized. However, if the predicted value of the analysis model algorithm does not meet expectations, it is necessary to continue to increase the number of marketing pushes and push time of competitive analysis products in different detection periods, reconstruct the marketing prediction dataset, and increase the number of training times of the analysis model algorithm.

[0150] If the relative proximity C k The maximum value corresponds to a non-zero k value, which determines the relative proximity C. k The analytical model algorithm corresponding to the largest numerical value is taken as the best analytical model algorithm, and a predictive analytical model is constructed according to the best analytical model algorithm.

[0151] It should be noted that in the formula for calculating relative closeness, the predicted value of each analytical model algorithm is closer to the negative ideal solution M. - distance d - k The larger the value, the further the model prediction is from the negative ideal solution, that is, away from the worst case, which in turn shows that the model prediction is closer to the ideal state to a certain extent.

[0152] The technical solution of this embodiment is as follows: Information entropy value analysis is performed on the number of marketing pushes for the competitive analysis product to determine whether the marketing effectiveness ratio of the competitive analysis product is higher than that of the product. Different analysis model algorithms are used to construct a predictive analysis model, which outputs the predicted number of marketing pushes and the push time. The marketing promotion strategy of the product is optimized, and the predictive analysis model constructed by the best analysis model algorithm is selected. Selecting the best analysis model algorithm is beneficial to the formulation of marketing strategies.

[0153] Example 3

[0154] A marketing visualization analysis method for big data further includes: Step 6, visually analyzing the product data of the product and the product data of competitors;

[0155] In some embodiments, the echarts framework is used to visualize product data and competitor analysis data using scatter plots, bubble charts, and tree diagrams.

[0156] The product data includes: marketing analysis period, marketing conversion ratio, competitive analysis products, number of marketing pushes, and marketing effectiveness ratio;

[0157] The benefits of this step are: Visually presenting data relationships: Using the ECharts framework, various data points related to the product and its competitors, such as marketing analysis periods and conversion rates, are visualized through scatter plots, bubble charts, and tree diagrams. This intuitive presentation allows businesses to more clearly observe the relationships and trends between the data, facilitating the discovery of potential market patterns and problems.

[0158] Supporting Decision Making: Visual analytics helps corporate management and marketing teams understand complex data information, providing intuitive and powerful support for developing marketing strategies and adjusting marketing directions, thereby improving the efficiency and accuracy of decision-making.

[0159] Example 4

[0160] like Figure 3 As shown, a marketing visualization and analysis system for big data includes the following modules:

[0161] Data acquisition module: Used to acquire marketing and promotion data and order quantity of products, and to perform numerical analysis on marketing and promotion data and order quantity to determine the marketing analysis period;

[0162] Competition Analysis Module: Used to perform cluster analysis on the marketing performance values ​​and marketing push frequency of competing products during the marketing analysis period to identify competing products;

[0163] Effectiveness Analysis Module: Used to statistically analyze the number of marketing pushes for products analyzed by competitors, obtain the marketing effectiveness probability, and perform information entropy value analysis to determine whether the marketing effectiveness ratio of the products analyzed by competitors is higher than that of the products analyzed by competitors.

[0164] Model building module: Used to build predictive analysis models using different analytical model algorithms, outputting the predicted number of marketing pushes and push time, and optimizing the marketing and promotion strategies of products;

[0165] Model determination module: Used to select the best predictive analysis model constructed by the analytical model algorithm.

[0166] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the present invention should still fall within the scope of the present invention.

Claims

1. A method for marketing visual analytics for big data, characterized by: Comprise the following steps: Step one: obtain the marketing promotion data of the commodity, the order quantity, and perform numerical analysis on the marketing promotion data and the order quantity to determine the marketing analysis period; Step two: perform cluster analysis on the marketing characteristic value and the marketing push times of the competitive commodity corresponding to the marketing analysis period to determine the competitive analysis commodity; Step three: perform statistical analysis on the marketing push times of the competitive analysis commodity to obtain the marketing effective probability and perform information entropy value analysis to determine whether the marketing effective ratio of the competitive analysis commodity is higher than that of the commodity; Step four: if it is higher, use different analysis model algorithms to construct a prediction analysis model, output the predicted marketing push times and push time, and optimize the marketing promotion strategy of the commodity; Step five: based on different analysis model algorithms, perform numerical analysis on the optimized marketing promotion strategy, and select the prediction analysis model constructed by the best analysis model algorithm; The determination method of the marketing analysis period is: Obtain the marketing characteristic value of each monitoring period, perform ratio processing on the marketing characteristic value of each monitoring period and the marketing push times corresponding to each monitoring period to obtain the marketing conversion value; Sort the marketing conversion value of each monitoring period, and select the monitoring period with the highest marketing conversion value as the marketing analysis period; The acquisition method of the marketing characteristic value is: Obtain the marketing promotion data and order quantity of different monitoring periods within the monitoring period; The marketing promotion data includes the access times, collection times, and marketing push times of the commodity on the e-commerce platform; Perform ratio processing on the access times and order quantity of each monitoring period to obtain the access conversion ratio; Perform ratio processing on the collection times and order quantity to obtain the collection conversion ratio; Based on the access conversion ratio, collection conversion ratio, and marketing push times of the monitoring period, construct a marketing conversion matrix and perform standardization processing to obtain a standard marketing conversion matrix; Based on the standard marketing conversion matrix, perform weighted sum processing on the collection conversion ratio and access conversion ratio of each monitoring period to obtain the marketing characteristic value; The determination method of whether the marketing effective ratio of the competitive analysis commodity is higher than that of the commodity is: Calculate the information analysis entropy value of the competitive analysis commodity based on the marketing effective probability; If the information analysis entropy value is lower than the preset analysis entropy threshold, the marketing effective ratio of the competitive analysis commodity is higher than that of the commodity; The acquisition method of the marketing effective ratio is: Obtain the access times and collection times of the competitive analysis commodity in the marketing analysis period, divide the marketing analysis period into multiple time points to obtain the analysis time point; Perform sum and mean value processing on the access times and collection times of the competitive analysis commodity in each analysis time point in the marketing analysis period to obtain the marketing effective times; Perform ratio processing on the marketing effective times and marketing push times to obtain the marketing effective ratio; The acquisition method of the marketing effective probability is: Obtain the marketing effective ratio of the competitive analysis commodity at each analysis time point, obtain the number of times that the marketing effective ratio of the competitive analysis commodity at each analysis time point is higher than that of the commodity, and obtain the time point over-limit times; Perform ratio processing on the time point over-limit times and the total number of analysis time points in the marketing analysis period to obtain the marketing effective probability; The best analysis model algorithm is determined in the following manner: based on the distance d of each analysis model algorithm's predicted value to the negative ideal solution M - - k , the distance d of each analysis model algorithm's predicted value to the positive ideal solution + k , the relative closeness C of each analysis model algorithm's predicted value is calculated k by the formula: The relative closeness C is obtained k ;​ the relative closeness C of the predicted value of each analysis model algorithm k ranking to determine the relative closeness C k the value of k for which the numerical maximum corresponds, where k is the number of different analysis model algorithms, k = 0 indicating that no analysis model algorithm is used; If the relative closeness C k If the k corresponding to the maximum value is not 0, the relative closeness C k The analysis model algorithm corresponding to the maximum value is taken as the best analysis model algorithm, and a prediction analysis model is constructed according to the best analysis model algorithm.

2. The method for marketing visual analytics for big data as claimed in claim 1 wherein: The competitive analysis commodity is determined in the following manner: A marketing analysis coordinate system is established, the marketing characteristic value and the marketing push times of the competitive commodity are taken as coordinates, four quadrants of the coordinate system are taken as a division, and the competitive commodity is divided into four categories; Based on the marketing analysis table and the competitive commodity divided into four categories, a marketing classification model is constructed using a clustering algorithm, and the competitive analysis commodity is determined.

3. The method for marketing visual analytics of big data according to claim 2, wherein: The marketing analysis table is constructed in the following manner: The marketing conversion value of the competitive commodity of the same type in the marketing analysis period is obtained, and the competitive conversion value is obtained; The competitive conversion value of all competitive commodities is obtained, the competitive conversion value is compared with the marketing conversion value, and the competitive commodity with the competitive conversion value higher than the marketing conversion value is selected; The marketing characteristic value and the marketing push times of the competitive commodity with the competitive conversion value higher than the marketing conversion value are obtained, and the marketing analysis table is constructed.

4. The method for marketing visual analytics of big data according to claim 1, wherein: the distance d of the prediction value of each analysis model algorithm to the negative ideal solution M - - k , the distance d of the prediction value of each analysis model algorithm to the positive ideal solution + k The acquisition method is that:​ Obtain the marketing effective ratio and marketing conversion value optimized by different analysis model algorithms, mark the marketing effective ratio obtained by different analysis model algorithms as a k , and mark the marketing conversion value as b k ; Based on the marketing effective ratio a obtained by different analysis model algorithms k , the marketing conversion value b k , and the model determination matrix M is constructed by using the ideal and poor solution distance method. obtaining a positive ideal solution M of the normalized model determination matrix + and a negative ideal solution M - ; The distance d of each analysis model algorithm's predicted value to the positive ideal solution M + + k , the negative ideal solution M - - k .​​ 5. A marketing visualized analysis system for big data, applied to the marketing visualized analysis method for big data in any one of claims 1-4, characterized in that: The following modules are included: A data acquisition module is configured to acquire marketing promotion data and order quantity of a commodity, and perform numerical analysis on the marketing promotion data and the order quantity to determine a marketing analysis period; A competitive analysis module is configured to perform clustering analysis on the marketing characteristic value and the marketing push times of the competitive commodity in the marketing analysis period to determine a competitive analysis commodity; An effective analysis module is configured to perform statistical analysis on the marketing push times of the competitive analysis commodity, obtain a marketing effective probability, and perform information entropy value analysis to determine whether a marketing effective ratio of the competitive analysis commodity is higher than a marketing effective ratio of the commodity; A model construction module is configured to construct a prediction analysis model using different analysis model algorithms, output predicted marketing push times and push times, and optimize a marketing promotion strategy of the commodity; A model determination module is configured to select a prediction analysis model constructed by the best analysis model algorithm.

Images