Marketing business management system based on big data analysis

Abstract

The invention relates to a marketing business management system based on big data analysis. According to the system, historical sales data, customer behavior data, market environment data and the like of a store are collected, and after data cleaning and standardization processing, a time sequence analysis model is utilized to predict the product sales volume and the sales volume trend, so that a basis is provided for inventory planning and marketing layout; using an association rule mining model to discover potential association rules in customer purchase behaviors so as to optimize product display and promotion strategies; customer groups are subdivided according to customer consumption characteristics and behavioral habits by means of a clustering analysis model, and a personalized marketing scheme is formulated for each group. According to the system, scientific configuration and management efficiency improvement of store marketing business are realized, multiple data sources and an advanced data analysis model are creatively integrated, and accurate and efficient marketing business management is realized. Through deep mining of historical data, the system can insight customer demands and market trends, and provides scientific marketing decision support for stores.

Description

Technical Field

[0001] The present invention relates to the technical field of data management, and in particular to a marketing business management system based on big data analysis. Background Art

[0002] Traditional marketing business management mainly relies on manual experience and intuitive judgment, which makes it difficult to accurately grasp customer needs and market dynamics, often leading to marketing decision-making errors, such as inventory backlogs or out-of-stocks, and unreasonable allocation of marketing resources. With the intensification of market competition and the diversification of consumer demand, traditional methods can no longer meet the requirements of efficient and precise marketing. Although some stores have begun to use data to assist in decision-making, they are limited to a single data source and simple analysis, such as making rough predictions based on historical sales data, without integrating multi-source data such as customer behavior and market environment, and without using advanced AI models to deeply mine data value.

[0003] There is no mature and systematic integrated multi-source data collection, cleaning, analysis and AI model-based marketing business configuration management solution in the existing technology to solve these problems. Summary of the invention

[0004] In order to solve the above problems, the present invention provides a marketing business management system based on big data analysis, which solves the problems mentioned in the background technology, such as marketing decision-making errors caused by reliance on manual experience, and the inability of a single data source and simple analysis methods to accurately grasp market dynamics. It provides stores with a scientific marketing management method that does not rely on economic means such as marketing cost regulation, but is based on data-driven. Compared with traditional technologies, the innovation of the present invention lies in the integration of multi-source data collection, advanced data preprocessing technology and a variety of AI models to form a complete marketing business management solution. This enables stores to more accurately predict sales trends, explore customer needs, segment markets and formulate personalized marketing strategies, thereby occupying an advantageous position in the fierce market competition.

[0005] To achieve the above purpose, the technical solution adopted by the present invention is: a marketing business management system based on big data analysis, comprising: A data collection module is used to collect multi-source data, including historical sales data of stores, customer behavior data, and market environment data; The data preprocessing module is used to clean and standardize the collected data to remove duplicate, erroneous and incomplete data records, and to convert data from different sources and formats into a standard format that is easy to analyze; The data analysis module is used to construct multiple mathematical models to analyze the preprocessed data. The mathematical models include: A time series analysis model is used to analyze historical sales data and predict the sales volume and sales amount trends of various products in different time periods. An association rule mining model is used to mine potential association rules in customer purchase behavior data. A clustering analysis model is used to divide customers into groups with different labels based on their purchase frequency, purchase amount, and age. A marketing business configuration module is used to execute personalized marketing plans for each group according to the analysis results of the data analysis module.

[0006] Preferably, the historical sales data includes the sales quantity and amount of various products in different time periods.

[0007] Preferably, the customer behavior data includes the residence time in the store, the time of browsing goods, the specific path, and the purchase frequency of goods.

[0008] Preferably, the market environment data includes local population structure changes, market activities of competitors during specific periods, and discount data on specific activity days.

[0009] Preferably, the time series analysis model predicts the purchase growth rate by comprehensively considering the product type, sales quantity, and amount of historical sales data of goods. The specific equation is as follows:

[0010] Where, R i represents the purchase growth rate of goods ; k represents the influence coefficient of price on goods within a certain time; P i represents the current pricing of goods ; P avg represents the average price; Ni represents the purchase quantity of goods ; f(V i ) represents the influence function of time on the sales rate, where , is a constant used to adjust the influence of sales time on the sales rate; represents the mutual influence coefficient between sales time and sales rate; V i represents the sales rate of goods ; T i represents the sales time of product i, and predicts the sales volume and sales amount trends of various products in different time periods through the purchase growth rate R i .

[0011] Preferably, the association rule mining model is based on the Apriori algorithm, which converts each transaction record into a list form of a set of commodities, generates candidate itemsets and frequent itemsets by determining support thresholds and confidence thresholds. For each frequent itemset, association rules are generated, and for each generated association rule, its confidence is calculated. According to the confidence threshold, the rules with confidence higher than the threshold are retained as valid association rules.

[0012] Preferably, the clustering analysis model is based on the K-means algorithm, which clusters customers according to their purchase frequency, purchase amount and age, analyzes the characteristics of each clustering category after dividing customers into k categories, and formulates personalized marketing plans for different groups.

[0013] Preferably, the marketing business configuration module determines the popular commodities and demand of each customer group according to the analysis results of the data analysis module, and intelligently allocates marketing material resources according to the characteristics of different customer groups and marketing channel preferences.

[0014] The beneficial effects of the present invention are as follows: (1) Improvement of marketing decision-making accuracy Traditional marketing business management relies on manual experience and intuitive judgment, making it difficult to accurately grasp customer needs and market dynamics, and easily leading to marketing decision-making mistakes. However, the present invention widely collects multi-source data, including historical sales data, customer behavior data, market environment data, etc., and deeply mines and analyzes the data using time series analysis models, association rule mining models and clustering analysis models, which can provide accurate sales forecasts, customer demand insights and market trend analyses for stores. This makes marketing decisions more scientific and accurate, avoids decision-making biases caused by subjective judgments, and significantly improves the success rate and effectiveness of marketing activities.

[0015] (2) Optimization of marketing resource allocation In the traditional marketing model, marketing resources are often unreasonably allocated, resulting in waste of some resources and poor marketing effects. The marketing business configuration module of the present invention realizes the optimized allocation of marketing resources according to the analysis results of the AI model. For example, through accurate inventory management, the replenishment plan is automatically adjusted according to sales forecasts and inventory status, avoiding inventory backlogs and out-of-stock phenomena, and improving inventory turnover; through personalized promotion strategies, exclusive promotion plans are formulated according to the characteristics of different customer groups, and marketing resources are accurately invested in target customers, improving the utilization efficiency of marketing resources; through marketing channel collaboration, online and offline channels are integrated to achieve the consistency and coherence of marketing information, avoiding duplicate investment and waste of marketing resources. This optimized allocation not only reduces marketing costs but also improves the return on marketing investment.

[0016] (3) Improvement of customer satisfaction and loyalty The present invention divides customers into different groups through a clustering analysis model and formulates personalized marketing plans for each group, which can better meet the diverse needs of different customer groups. Personalized product recommendations, promotional activities, and shopping experiences enable customers to feel the attention and emphasis of the store, enhancing customer satisfaction and loyalty. In addition, the potential association rules discovered by the association rule mining model in customers' purchase behaviors help optimize product displays and bundled sales, providing customers with a more convenient and efficient shopping experience and further enhancing customers' shopping pleasure and satisfaction.

[0017] (IV) Enhancement of Market Competitiveness In the fierce market competition, enterprises need to continuously innovate and optimize their marketing strategies to maintain a competitive edge. The present invention provides a scientific marketing management method based on data-driven approach rather than relying on economic means such as marketing expense regulation, enabling the store to more keenly capture market changes and customer needs, quickly adjust marketing strategies, and flexibly respond to market competition. Through precise marketing activities and high-quality services, enterprises can attract more customers, expand the market share, and thus stand out in the competition.

[0018] (V) Improvement of Marketing Efficiency The marketing business configuration module realizes the automated execution and real-time monitoring of marketing tasks, greatly improving marketing efficiency. For example, automatically sending marketing emails, text messages, push notifications, etc. to customers reduces the workload and time cost of manual operations; real-time monitoring of key indicators of marketing activities and timely adjustment of activity strategies ensure the smooth progress of marketing activities and timely feedback of effects. In addition, the data collection, processing, and analysis processes of the entire system are highly automated and intelligent, capable of quickly responding to market changes and business requirements, further enhancing the overall efficiency of marketing work.

[0019] In summary, the beneficial effects of the present invention are that it effectively solves problems such as marketing decision-making mistakes, unreasonable allocation of marketing resources, difficulty in meeting diverse customer needs, and insufficient market competitiveness in traditional marketing business management, providing the store with a precise, efficient, and scientific marketing business management method, which helps enterprises succeed in the fierce market competition and achieve sustainable development. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 is a block diagram of a marketing business management system based on big data analysis according to the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0021] Please refer to Figure 1 as shown, the present invention relates to a marketing business management system based on big data analysis, including: A data collection module for collecting multi-source data, which includes historical sales data of stores, customer behavior data, and market environment data. The data collection module is the foundation of the entire marketing business management system. It is responsible for widely collecting various types of data from multiple channels and sources. These data include, but are not limited to, historical sales data of stores, customer behavior data, and market environment data. Its role is to provide comprehensive, accurate, and detailed original information for subsequent data analysis and processing, ensuring that the entire system can make precise marketing decisions based on a sufficient data foundation.

[0022] The historical sales data includes the sales quantity and amount of various products in different time periods.

[0023] Historical sales data source: By connecting to the sales management systems of stores, such as POS (Point of Sale) systems, online mall databases, etc., extract the sales records of products, including information such as sales time, product name, sales quantity, and sales amount.

[0024] The customer behavior data includes the stay duration of customers in the store, the time spent browsing products and the specific paths, and the frequency of purchasing products.

[0025] Customer behavior data source: By connecting to the customer relationship management system (CRM) of the store, tracking codes of websites and mobile applications, in-store monitoring systems, etc., collect the behavior data of customers in the store, such as stay time, browsing paths, purchase frequency, purchase time, etc.

[0026] The market environment data includes local population structure changes, market activities of competitors during specific periods, and discount data on specific event days.

[0027] Market environment data source: Utilize external data providers, industry reports, government statistical data, and web crawler technology, etc., to collect data such as market trends, competitor activities, demographic information, economic indicators, etc.

[0028] Set up scheduled tasks according to business requirements to regularly extract data from each data source. For example, extract sales data from the POS system every hour and obtain customer behavior data from the website backend every day.

[0029] For some key data, such as online order data and real-time customer browsing behavior, collect them through real-time data interfaces to ensure the timeliness and accuracy of the data.

[0030] Finally, match and associate the data from different data sources and unify them into a common data framework. For example, associate the purchase records of customers with their behavior data to form a complete customer profile.

[0031] Convert data in different formats into a unified data format for the system, such as unifying the date format to YYYY - MM - DD and the numerical format to a standard digital format, etc.

[0032] A data preprocessing module for cleaning and standardizing the collected data to remove duplicate, incorrect, and incomplete data records, and uniformly convert data from different sources and in different formats into a standard format convenient for analysis; The data preprocessing module performs operations such as cleaning, converting, and integrating the collected raw data to improve the quality and usability of the data. Its role is to ensure that the data input into the data analysis model is accurate, complete, and consistent, thereby providing reliable data support for subsequent analysis and modeling.

[0033] Data cleaning: Removing duplicate data: Identify and delete duplicate data records through methods such as unique identifiers or hash values of data records. For example, in sales data, if there are two exactly identical sales records, only keep one.

[0034] Handling missing data: Adopt different methods to handle missing data according to the importance and degree of missing of the data. For cases where the key data has less missing, statistical methods such as mean, median, or mode can be used for filling; for cases where the non - key data has more missing, relevant records can be selected for deletion.

[0035] Correcting incorrect data: Use data validation rules and data quality management tools to identify and correct incorrect data records. For example, check whether the date format is correct, whether the numerical value is within a reasonable range, etc.

[0036] Data conversion: Standardization processing: Perform standardization conversion on the data to make it have the same dimension and scale. For example, normalize the customer's purchase amount and purchase frequency so that their values range between [0, 1] to eliminate the dimension differences between different features and improve the performance of the analysis model.

[0037] Data type conversion: Convert the data into a suitable data type according to the analysis requirements. For example, convert a date string into a date object, convert categorical data into a numerical one - hot encoding form, etc.

[0038] Data integration: Data merging: Merge different data sets that have been cleaned and converted to form a unified data set. For example, merge the customer's basic information data set, purchase behavior data set, and market environment data set into a complete data set for subsequent analysis and modeling.

[0039] Data aggregation: Aggregate data according to the analysis purpose. For example, summarize sales data monthly or quarterly, calculate indicators such as total monthly or quarterly sales amount, sales quantity, etc.; aggregate customer behavior data by customer groups to analyze the purchase behavior characteristics of different groups.

[0040] Data sampling: Determine the sampling method: According to the data volume and analysis requirements, select an appropriate sampling method, such as simple random sampling, stratified sampling, systematic sampling, etc. For example, for large-scale sales data, the stratified sampling method can be adopted, stratified according to product categories or sales regions, and then a certain proportion of samples are drawn from each layer for analysis.

[0041] Perform the sampling operation: According to the determined sampling method, draw a representative sample data set from the original data set to improve the efficiency of data processing and analysis without affecting the accuracy of the analysis results.

[0042] Through the above refinement of the data collection module and data preprocessing module, the roles and working processes of these two modules in the entire marketing business management system can be more clearly understood, thus providing strong support for the effective operation of the system and accurate marketing decisions.

[0043] The data analysis module is used to construct multiple mathematical models to analyze the preprocessed data. The mathematical models include: The time series analysis model is used to analyze historical sales data and predict the sales volume and sales amount trends of various products in different time periods; The time series analysis model predicts the purchase growth rate by comprehensively considering the product type, sales quantity, and amount of historical sales data of the commodity. The specific equation is as follows:

[0044] Where, R i represents the purchase growth rate of the commodity ; k represents the influence coefficient of price on the commodity within a certain time; P i represents the current pricing of the commodity ; P avg represents the average price; N i represents the purchase quantity of the commodity ; f(V i ) represents the influence function of time on the sales rate, where , is a constant used to adjust the influence of sales time on the sales rate; represents the mutual influence coefficient between sales time and sales rate; V i represents the commodity Sales rate, T i Indicates the sales time of product i, through the purchase growth rate R i Predict the sales volume and sales trend of various products in different time periods.

[0045] Operation process: Calculate the purchase growth rate R i : First, according to the product Current price P i And average price P avg Calculate the price difference |P i -P avg |. The greater the price difference, the smaller the purchase growth rate R due to the existence of the influence coefficient k i ; Consider the purchase quantity : Consider the positive effect of the purchase quantity on the purchase growth rate and multiply directly. The more the purchase quantity, the higher the purchase growth rate; The influence of the sales rate on the purchase growth rate: Through the sales rate function f(V i ), the higher the sales rate, the greater the purchase growth rate; The interaction of time on the sales rate: Through Item, consider the accelerating effect of time on the change of the sales rate. Especially when the sales rate is high, the sales month / season (time) has a stronger promoting effect on the purchase growth rate.

[0046] This comprehensive equation not only considers the basic purchase quantity and price, but also provides a more refined prediction of the purchase growth rate through the interaction term of time on the sales rate.

[0047] According to the predicted purchase growth rate, obtain the sales volume and sales trend of various products in different future time periods, and then provide a strong basis for the inventory planning, marketing focus layout and resource allocation of the store.

[0048] The time series analysis model provides data support and prediction basis for marketing decisions. Marketing personnel can formulate more scientific and reasonable marketing plans and decisions according to the prediction results of the model. It helps to evaluate the potential effects of different marketing plans, so as to select the optimal plan. This reduces the uncertainty and risk of decisions and improves the accuracy and success rate of decisions.

[0049] Association rule mining model, used to mine potential association rules in customer purchase behavior data; The association rule mining model is based on the Apriori algorithm. It converts each transaction record into a list form of a set of goods, generates candidate itemsets and frequent itemsets by determining the support threshold and confidence threshold. For each frequent itemset, it generates association rules. For each generated association rule, it calculates its confidence. According to the confidence threshold, it retains the rules with confidence higher than the threshold as valid association rules.

[0050] This model can analyze a large number of customer purchase records and find out the association combinations that frequently appear together between different goods. For example, in the supermarket shopping scenario, it can be found that the proportion of customers who buy milk and also buy bread is very high. The mining of such association combinations is achieved by analyzing the customer's shopping basket data. The model will calculate the occurrence frequency and support degree of different pairs of goods (i.e., the proportion of the number of transactions that buy both of these two goods in the total number of transactions).

[0051] In a specific embodiment of the present invention, in a store of a large chain supermarket, there are 10,000 shopping records. After analysis by the association rule mining model, it is found that the number of customers who buy shampoo and conditioner accounts for 30% of the total number of customers, and among the customers who buy shampoo, 60% also buy conditioner, which constitutes a very strong association combination of goods. Such association information is very useful for the store when planning the shelf display. It can place the associated goods in adjacent positions to facilitate customer purchase and thus improve the sales efficiency.

[0052] The association rule mining model can reveal the behavior patterns and habits of customers when purchasing goods. For example, the model can identify the rule that when customers buy a certain category of goods, they tend to buy other specific categories of goods. This helps the store understand the customer's consumption psychology and demand structure. For example, the model may find that customers who buy fitness equipment will also buy sports clothing and sports drinks.

[0053] In a specific embodiment of the present invention, in a sports goods store, by collecting and analyzing the sales data over a period of time, the association rule mining model finds that 40% of the customers who buy basketball will also buy sports knee pads, and 30% will also buy basketball shoes. This shows that when customers buy the sports equipment of basketball, they will consider the protection and professional equipment needs during the sports process. The store can, based on this discovery, display the related goods such as basketball, knee pads and basketball shoes in combination, and formulate targeted promotional strategies, such as launching a basketball equipment package, to meet the one-stop shopping needs of customers and enhance the customer's purchase experience and the store's sales volume.

[0054] The Apriori algorithm is based on the concept of frequent itemsets. It discovers frequent itemsets by generating candidate itemsets and calculating the support degree, and then generates association rules. The specific implementation steps are as follows: Set parameters: Determine the support threshold (min_support): Support represents the frequency of occurrence of a certain combination of items in all transactions. According to the business objectives and data characteristics of the supermarket, set an appropriate support threshold. For example, if you want to find combinations of items that appear in at least 10% of the transactions, you can set the support threshold to 0.1.

[0055] Determine the confidence threshold (min_confidence): Confidence represents the credibility of deriving the consequent from the antecedent of an association rule. It is also set according to business requirements. For example, setting it to 0.7 means that if the confidence of a rule is less than 70%, the rule is not considered valid.

[0056] Generate frequent item sets: Generate candidate 1-item sets (C1): Consider all individual items as candidate 1-item sets and count the number of times they appear in all transactions.

[0057] Generate frequent 1-item sets (L1): According to the support threshold, select the items from the candidate 1-item sets that meet the support requirement to form the frequent 1-item sets.

[0058] Generate candidate k-item sets (Ck, k≥2): Use the frequent (k - 1)-item sets to generate candidate k-item sets through self-joining and pruning operations. For example, to generate candidate 2-item sets from the frequent 1-item sets, pair up the items in the frequent 1-item sets and ensure that these combinations have all appeared in the previous frequent item sets (pruning operation).

[0059] Generate frequent k-item sets (Lk, k≥2): Calculate the support of the candidate k-item sets and retain the item sets that meet the support threshold to form the frequent k-item sets. Repeat this process until no new frequent item sets can be generated.

[0060] For each frequent item set, generate possible association rules. For example, for the frequent item set {milk, bread, butter}, rules such as {milk, bread} → {butter}, {milk} → {bread, butter}, etc. can be generated.

[0061] For each generated rule, calculate its confidence. For example, the confidence of the rule {milk, bread} → {butter} = (number of transactions where milk and bread are purchased) / (number of transactions where milk, bread, and butter are purchased).

[0062] The present invention takes into account the impact of time variation on confidence and improves the calculation method of confidence as follows:

[0063] Wherein, Represents the interaction coefficient between the sales time and the sales rate. Represents the confidence level between product i and product j (i→j), R ij Represents the average purchase growth rate of product i and product j at time T, T ij Represents the purchase time of product i and product j. In the numerator term is the number of transactions of product i and product j, and in the denominator term is the number of transactions of product i.

[0064] According to the confidence threshold, retain the rules with a confidence level higher than the threshold as valid association rules.

[0065] When the data volume is large, the support threshold can be appropriately increased to reduce the number of candidate itemsets and improve the algorithm efficiency. For example, for data containing 10,000 transaction records, the support threshold can be set to 0.01 (1%).

[0066] Data sparsity: If the data is sparse (i.e., the number of product types in each transaction is small), the support threshold can be appropriately reduced to avoid missing potential association rules. For example, for data where each transaction contains an average of 3 products, the support threshold can be set to 0.005 (0.5%).

[0067] According to the association rules, reasonably arrange the positions of products on the shelves. For example, place the products that are often purchased together in adjacent positions to improve the convenience of customers' shopping and increase sales opportunities.

[0068] Personalized recommendation: Recommend other products related to the customers' purchase history to them. For example, if customers often buy milk and bread, according to the association rules, recommend butter or other related products to them.

[0069] Promotion strategy formulation: When formulating promotion activities, consider the association relationship between products. For example, conduct bundled sales or offer discount promotions for product combinations that are often purchased together to increase sales volume and profit.

[0070] By analyzing the sales data after applying the association rules, evaluate the growth of sales volume, and compare the percentage increase in sales volume before and after applying the association rules (to improve sales performance).

[0071] Improve customer satisfaction: Through customer feedback or surveys, understand customers' satisfaction with the new shelf display and personalized recommendations. The results of the customer satisfaction survey show that customers' satisfaction with the shopping experience has increased from 70% to 80%.

[0072] Enhance market competitiveness: Evaluate the change in the supermarket or convenience store's competitiveness in the market, such as an increase in market share or an improvement in customer loyalty. Through market share analysis, it is found that the supermarket's market share has increased from 10% to 12%.

[0073] A clustering analysis model is used to divide customers into groups with different labels based on their purchase frequency, purchase amount, and age. The clustering analysis model is based on the K-means algorithm. It clusters the purchase frequency, purchase amount, and age of customers, divides customers into k categories, analyzes the characteristics of each clustering category, and formulates personalized marketing plans for different groups.

[0074] In this embodiment, K = 4 is set, and the purchase frequency, purchase amount, and age of customers are clustered. Determine the clustering category to which each customer belongs. The clustering result divides customers into 4 categories, namely C1, C2, C3, and C4.

[0075] Analyze the characteristics of each clustering category and find out the typical characteristics of each group.

[0076] It is found that customers in category C1 have a high purchase frequency, a medium purchase amount, and their ages are mainly between 20 and 30 years old; customers in category C2 have a low purchase frequency, a high purchase amount, and their ages are mainly between 40 and 50 years old; customers in category C3 have both a low purchase frequency and a low purchase amount, and their age distribution is relatively wide; customers in category C4 have a medium purchase frequency, a high purchase amount, and their ages are between 30 and 40 years old.

[0077] According to the characteristic analysis, name each clustering group to facilitate understanding and subsequent marketing strategy formulation. For example, name category C1 as "young and active customers", category C2 as "high-consumption middle-aged customers", category C3 as "occasional consumption customers", and category C4 as "mid-high-income young customers".

[0078] Formulate personalized marketing plans according to the classification Young and active customers (C1): Product recommendation: Recommend new products that meet the tastes and needs of young people, such as fashionable clothing, electronic products, fitness equipment, etc.

[0079] Promotion activities: Design high-frequency small-amount promotion activities, such as weekly discount days, new product trials, membership point doubling, etc.

[0080] Communication channels: Push marketing information through channels commonly used by young people, such as social media, mobile applications, etc.

[0081] High-consumption middle-aged customers (C2): Product recommendation: Provide high-end and quality products, such as high-end skin care products, imported foods, luxury household items, etc.

[0082] Promotion activities: Launch exclusive VIP services and customized discounts, such as special birthday discounts, high-end product experience activities, etc.

[0083] Communication channels: Communicate through e-mail, offline event invitations, etc., focusing on personalized and distinguished services.

[0084] Occasional customers (C3): Product recommendations: Basic recommendations on daily necessities and seasonal goods, such as daily necessities, seasonal clothing, etc.

[0085] Promotional activities: Provide simple and direct discount offers, such as discounts on purchases over a certain amount, package discounts, etc.

[0086] Communication channels: Deliver promotional information through concise methods such as text messages and posters to attract them to come to the store and buy.

[0087] Middle- to high-income young customers (C4): Product recommendations: Recommend technological products, fashion items, affordable luxury goods, and other products that are in line with their spending power and pursuit of quality life.

[0088] Promotions: Design attractive discount and offer packages, such as limited-time discounts, bundled purchase offers, etc.

[0089] Communication channels: Combine online and offline promotional methods, such as promoting in technology product forums, fashion magazines and other channels.

[0090] Program implementation and effect evaluation According to the formulated marketing strategy, personalized marketing activities are carried out for different customer groups. For example, the customer group can be identified through the membership system, and customized marketing emails or text messages can be sent to different groups.

[0091] Integrate marketing channels: Ensure that the marketing messages of online and offline channels are consistent to provide a seamless shopping experience. For example, online malls and offline stores simultaneously launch promotions targeting a certain group.

[0092] Collect sales data after marketing activities and evaluate the responses of different groups. For example, calculate the purchase conversion rate, average order value, repurchase rate and other indicators of each group and compare them with those before the marketing activities.

[0093] Customer feedback analysis: Collect customer feedback on personalized marketing activities to understand their satisfaction and suggestions for improvement. For example, obtain customer feedback through questionnaires, online reviews, etc.

[0094] Data analysis and adjustment: Based on the evaluation results, analyze the effectiveness of marketing activities and identify existing problems and deficiencies. For example, if the conversion rate of a certain group is low, it may be necessary to adjust the product recommendation strategy or promotion content.

[0095] Model Update: Update the clustering model regularly and reanalyze the characteristics of customer groups to adapt to changes in customer behavior. For example, re-conduct cluster analysis and update customer group segmentation based on new sales data every quarter.

[0096] Adjust feature selection and clustering algorithm parameters based on market changes and customer feedback. For example, if new consumer trends are discovered, add relevant features (such as demand for environmentally friendly products) for analysis.

[0097] In this embodiment, customers are clustered and divided into four categories according to purchase frequency, purchase amount and product type preference. For the first category of "fashion followers", they often buy fashion products, and the platform can inform them of new product listings in advance; for the second category of "cost-effective pursuers", who focus on price and practicality, the platform provides attractively priced products and coupons. Through this personalized marketing, the platform can enhance customer experience, enhance loyalty, and improve overall sales performance.

[0098] The marketing business configuration module is used to execute personalized marketing plans for each group based on the analysis results of the data analysis module.

[0099] The marketing business configuration module determines the popular products and demand of each customer group based on the analysis results of the data analysis module, and intelligently allocates marketing material resources according to the characteristics of different customer groups and marketing channel preferences. The specific methods are as follows: Generate personalized promotion plans based on clustering results and association rules. For example, for customers who occasionally consume but are price-sensitive, explore the related products of their frequently purchased products and launch "full amount discount" or "combination package" discounts to stimulate consumption frequency and amount.

[0100] Combined with time series forecasting, marketing activities can be planned reasonably. For example, if it is predicted that the sales volume of a certain product will decline at the beginning of a quarter, promotional activities can be arranged in advance before the decline to seize the market and stabilize sales.

[0101] Analyze the shopping channel preferences of each customer group to achieve online and offline marketing collaboration. For example, if a group prefers online browsing and offline experience, the module will generate strategies to push product information online, make appointments for offline experience discounts, and guide customers to scan the QR code and follow the official account offline to obtain exclusive online discounts, thus improving the omni-channel shopping experience.

[0102] Ensure the consistency of marketing information in all channels. Based on clustering and association rules, formulate unified cross-channel marketing words, visual styles, etc., strengthen brand image, and avoid customer loss due to information confusion.

[0103] With the help of real-time data analysis, marketing strategies can be adjusted in a timely manner. For example, during marketing activities, product recommendations can be dynamically optimized and promotion efforts can be adjusted based on customer feedback and sales data to ensure that strategies are in line with market changes and improve marketing effectiveness.

[0104] The above embodiments are only descriptions of the preferred embodiments of the present invention, and do not limit the scope of the present invention. Without departing from the design spirit of the present invention, various deformations and improvements made by those of ordinary skill in the art to the technical solutions of the present invention shall fall within the protection scope determined by the claims of the present invention.

Claims

1. A marketing business management system based on big data analysis, characterized in that, Including: A data collection module for collecting multi-source data, where the multi-source data includes historical sales data of stores, customer behavior data, and market environment data; A data preprocessing module for cleaning and standardizing the collected data to remove duplicate, incorrect, and incomplete data records, and uniformly converting data from different sources and different formats into a standard format convenient for analysis; A data analysis module for constructing multiple mathematical models to analyze the preprocessed data. The mathematical models include: A time series analysis model for analyzing historical sales data to predict the sales volume and sales amount trends of various products in different time periods; An association rule mining model for mining potential association rules in customer purchase behavior data; A clustering analysis model for dividing customers into groups with different labels based on the purchase frequency, purchase amount, and age of customers; A marketing business configuration module for executing personalized marketing plans for each group according to the analysis results of the data analysis module.

2. The marketing business management system based on big data analysis according to claim 1, characterized in that, The historical sales data includes the sales quantity and amount of various products in different time periods.

3. The marketing business management system based on big data analysis according to claim 1, wherein, The customer behavior data includes the residence time of customers in the store, the time spent browsing goods, the specific path, and the purchase frequency of goods.

4. A marketing business management system based on big data analysis according to claim 1, characterized in that, The market environment data includes local population structure changes, market activities of competitors during specific periods, and discount data on specific activity days.

5. A marketing business management system based on big data analysis according to claim 1, characterized in that, The time series analysis model predicts the purchase growth rate by comprehensively considering the product type, sales quantity, and amount of historical sales data of goods. The specific equation is as follows: ， Among them, R i represents the purchase growth rate of the product ; k represents the influence coefficient of price on the product within a certain period of time; P i represents the current price of the product ; P avg represents the average price; N i represents the purchase quantity of the product ; f(V i ) represents the influence function of time on the sales rate, where , is a constant used to adjust the influence of sales time on the sales rate; represents the interaction coefficient between sales time and sales rate; V i represents the sales rate of the product ; T i represents the sales time of product i, and predicts the sales volume and sales trend of various products in different time periods through the purchase growth rate R i .

6. The marketing business management system based on big data analysis according to claim 1, characterized in that The association rule mining model is based on the Apriori algorithm. Each transaction record is converted into a list form of a set of goods. Candidate itemsets and frequent itemsets are generated by determining the support threshold and confidence threshold. For each frequent itemset, association rules are generated. For each generated association rule, its confidence is calculated. According to the confidence threshold, rules with confidence higher than the threshold are retained as valid association rules.

7. A marketing business management system based on big data analysis according to claim 1, characterized in that, The clustering analysis model is based on the K-means algorithm. It clusters the purchase frequency, purchase amount, and age of customers, divides customers into k categories, analyzes the characteristics of each clustering category, and formulates personalized marketing plans for different groups.

8. A marketing business management system based on big data analysis according to claim 1, characterized in that, The marketing business configuration module determines the popular goods and demand of each customer group according to the analysis results of the data analysis module, and intelligently allocates marketing material resources based on the characteristics of different customer groups and marketing channel preferences.

Images