E-commerce inventory monitoring and sales optimization method based on multi-category operation data

Abstract

This invention relates to a method for e-commerce inventory monitoring and sales optimization based on multi-category operational data, comprising: extracting consumer browsing, adding to cart, and purchase record behavioral data from the e-commerce platform database; obtaining a consumer preference quantification score based on the behavioral data; obtaining an inventory plan based on the consumer preference quantification score; simulating resource utilization under sales scenarios based on the inventory plan to determine final sales optimization parameters; and updating the e-commerce system configuration using the determined final sales optimization parameters to obtain a real-time inventory monitoring mechanism. This invention, through multi-dimensional data fusion and predictive models, accurately captures sales patterns and demand fluctuations, significantly improving inventory management efficiency and sales conversion rates, and providing intelligent decision support for e-commerce platforms.

Description

Technical Field

[0001] This invention relates to the field of e-commerce operation technology, and in particular to a method for e-commerce inventory monitoring and sales optimization based on multi-category operation data. Background Technology

[0002] In the e-commerce industry, inventory monitoring and sales optimization are crucial for ensuring operational efficiency and improving profitability. With the diversification of consumer demands and intensifying market competition, accurately grasping the sales dynamics of multiple product categories and rationally allocating inventory resources has become an unavoidable challenge for e-commerce platforms. Research in this area not only concerns cost control for businesses but also directly impacts user experience and market competitiveness, making it key to driving the industry's sustainable development.

[0003] However, many current methods often struggle to fully capture the complex behavioral characteristics of the sales process when dealing with multi-category operational data. Especially when faced with differences in consumption habits and market feedback across different product categories, existing solutions frequently overlook the dynamic changes and multi-dimensional relationships behind consumer behavior, limiting the accuracy of inventory allocation and sales forecasting. This limitation makes it difficult for companies to achieve efficient resource utilization and fully tap into sales potential in a rapidly changing market environment.

[0004] A deeper challenge lies in effectively integrating and quantifying various behavioral data throughout the sales process. In particular, behavioral data from browsing and adding items to the cart to the final purchase is highly fragmented and irregular. Without converting this data into analyzable numerical values, it's difficult to accurately reflect the sales performance of different product categories. For example, some product categories may have high pageview rates but low conversion rates, while others may have higher average order values ​​but fluctuating overall sales. This diversity makes it difficult to grasp the core issues using traditional statistical methods alone. Furthermore, this difficulty in data conversion affects the comprehensive assessment of sales activity and consumer purchasing power, often leading companies into a predicament of blindly adjusting inventory strategies.

[0005] Therefore, how to integrate scattered behavioral data into quantifiable features in multi-category operation scenarios, and how to accurately characterize the sales patterns and consumer preferences of different categories through these features, has become a key issue for e-commerce inventory monitoring and sales optimization. Summary of the Invention

[0006] The purpose of this invention is to propose an e-commerce inventory monitoring and sales optimization method based on multi-category operational data, in order to solve the problems existing in the prior art.

[0007] To achieve the above objectives, the present invention provides the following solution: E-commerce inventory monitoring and sales optimization methods based on multi-category operational data include: Extract behavioral data on consumer browsing, adding to cart, and purchase records from e-commerce platform databases; Based on the aforementioned behavioral data, a quantitative score of consumer preference is obtained; Based on the aforementioned consumer preference quantification scores, an inventory plan is obtained; Based on the aforementioned inventory plan, resource utilization is simulated under sales scenarios to determine the final sales optimization parameters. By updating the e-commerce system configuration with the determined final sales optimization parameters, a real-time inventory monitoring mechanism can be obtained.

[0008] Optionally, behavioral data on consumer browsing, adding to cart, and purchase records can be extracted from the e-commerce platform's database, including: By obtaining consumers' browsing history, shopping cart items, and purchase records from the e-commerce platform's database, an initial behavioral data set is formed, and preliminary data extraction is completed. Data cleaning techniques are used to process the initial behavioral data set, identify noisy data and missing value fields, and obtain a preliminarily cleaned data set. If outliers still exist in the dataset after initial cleaning, a second filtering is performed using a preset threshold range to remove data records that do not meet the standards, thus determining the behavioral data.

[0009] Optionally, obtaining a consumer preference quantification score based on the behavioral data includes: Based on the behavioral data, a clustering algorithm is used to group the behavioral patterns of different product categories to determine the unique sales pattern characteristics of each category. Based on the aforementioned sales patterns, a quantitative score of consumer preference is obtained by integrating multi-dimensional behavioral indicators through weighted average calculation.

[0010] Optionally, identifying unique sales patterns for each product category includes: Extract user behavior records related to product categories from behavioral data, perform preliminary cleaning and formatting of user interaction data for each product category, and obtain basic data on structured behavioral patterns. Clustering algorithms are used to group the structured behavioral pattern data, and user behavior records for different product categories are divided into patterns to determine the behavioral pattern clusters for each category. Based on the results of behavioral pattern clustering, we analyze the frequency of user interaction and preference changes of each type of product in different time periods to obtain preliminary characteristics of the sales patterns of each product category. By further exploring the preliminary characteristics of sales patterns and combining them with the classification information of product categories, we can determine whether there are significant differences in patterns. If the characteristics of a certain product category are significantly different from those of other product categories, it will be marked as a unique feature category. For product categories marked with unique characteristics, extract the core data points from their behavioral pattern clusters, obtain the corresponding sales characteristic details, and determine their unique sales pattern manifestations. Based on the unique sales patterns identified, the sales characteristics of each product category are integrated to generate a highly targeted set of sales pattern features.

[0011] Optionally, based on the aforementioned sales pattern characteristics, multi-dimensional behavioral indicators are integrated by calculating a weighted average, including: By extracting pageview count and conversion rate from sales pattern characteristics, and analyzing the correlation between the two, a preliminary judgment on sales patterns can be obtained. If the number of views is high but the conversion rate is lower than the preset threshold, multi-dimensional data collection will be carried out on user behavior to obtain behavioral indicators involving page dwell time, click path and interaction frequency. Based on the collected behavioral indicators, a weighted average method is used to process the data of each dimension and calculate a comprehensive consumer preference quantitative score.

[0012] Optionally, obtaining an inventory plan based on the consumer preference quantification score includes: From the obtained quantitative scores of consumer preferences, time series data is extracted, and a time series forecasting model is used to predict future sales dynamics, determine potential inventory demand fluctuations, and obtain inventory plans. If the potential inventory demand fluctuation exceeds the preset threshold, the multi-category inventory allocation ratio is adjusted according to the forecast results to obtain an optimized inventory plan.

[0013] Optionally, a time series forecasting model can be used to predict future sales dynamics and determine potential inventory demand fluctuations, including: Time series data is cleaned and formatted to obtain structured time series records; For structured time series records, a time series forecasting model is used for analysis. An autoregressive integral moving average model is applied to process historical data to determine future sales trends and thus determine inventory plans.

[0014] Optionally, based on the inventory plan, the resource utilization rate under simulated sales scenarios is used to determine the final sales optimization parameters, including: The resource allocation information in the inventory plan is categorized and organized to determine the initial configuration status of each type of resource; Based on the initial configuration, a simulated environment for the sales scenario is built to obtain data on resource utilization efficiency under different scenarios; For the utilization efficiency data, analyze the degree of matching between resource allocation and sales scenarios. If the degree of matching is lower than the preset threshold, adjust the resource allocation ratio and regenerate the utilization efficiency data. By using the adjusted utilization efficiency data, key resource utilization bottlenecks are extracted, and the range of core parameters affecting sales optimization is determined. Based on the range of core parameters and the results of simulation analysis, the optimal parameter combination for the sales scenario is derived, and the final parameter determination result is obtained.

[0015] The beneficial effects of this invention are as follows: This invention discloses an e-commerce inventory monitoring and sales optimization method based on multi-category operational data. Addressing the challenges of diverse sales patterns across multiple product categories, uneven inventory allocation leading to resource waste and low sales conversion rates in e-commerce platforms, the method cleanses data to remove noise and missing values, constructing a clean behavioral dataset. Clustering algorithms are then used to analyze the behavioral patterns of each product category, extracting key sales characteristics. When high pageviews are observed but low conversion rates, this invention calculates a consumer preference score using a weighted average, combines this with a time-series forecasting model to predict future sales dynamics, and assesses inventory demand fluctuations. If fluctuations exceed a threshold, the inventory allocation ratio is dynamically adjusted to optimize the inventory plan. Simulated sales scenarios are used to evaluate resource utilization, determine final sales optimization parameters, update e-commerce configurations, and achieve real-time inventory monitoring. The core innovation of this invention lies in its ability to accurately capture sales patterns and demand fluctuations through multi-dimensional data fusion and predictive models, significantly improving inventory management efficiency and sales conversion rates, and providing intelligent decision support for e-commerce platforms. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a schematic diagram of the e-commerce inventory monitoring and sales optimization method based on multi-category operation data according to an embodiment of the present invention. Detailed Implementation

[0018] 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.

[0019] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0020] like Figure 1 As shown, this embodiment proposes an e-commerce inventory monitoring and sales optimization method based on multi-category operational data, including: Extract behavioral data on consumer browsing, adding to cart, and purchase records from e-commerce platform databases; Based on the aforementioned behavioral data, a quantitative score of consumer preference is obtained; Based on the aforementioned consumer preference quantification scores, an inventory plan is obtained; Based on the aforementioned inventory plan, resource utilization is simulated under sales scenarios to determine the final sales optimization parameters. By updating the e-commerce system configuration with the determined final sales optimization parameters, a real-time inventory monitoring mechanism can be obtained.

[0021] Furthermore, behavioral data on consumer browsing, adding to cart, and purchase records extracted from e-commerce platform databases includes: By obtaining consumers' browsing history, shopping cart items, and purchase records from the e-commerce platform's database, an initial behavioral data set is formed, and preliminary data extraction is completed. Data cleaning techniques are used to process the initial behavioral data set, identify noisy data and missing value fields, and obtain a preliminarily cleaned data set. If outliers still exist in the dataset after initial cleaning, a second filtering is performed using a preset threshold range to remove data records that do not meet the standards, thus determining the behavioral data.

[0022] Specifically, in this embodiment, when extracting consumer behavior data from the e-commerce platform database, the user's browsing, shopping cart additions, and purchase records from the user behavior log table over the past 30 days can be filtered using Structured Query Language (SQL). Assuming the extracted data contains 1 million records, with browsing records accounting for 60%, shopping cart additions accounting for 25%, and purchase records accounting for 15%, the implementation involves using a database management system (such as MySQL) to write query statements, sorting by user ID and timestamp to ensure data integrity, while setting time windows to filter out expired data. Next, data cleaning is performed to remove noise and missing values. For example, if 5% of the records have empty user IDs or abnormal timestamps, an algorithm is used to infer and complete these missing values ​​using the user's IP address and session ID. If completion is not possible, the missing records are discarded, ultimately retaining approximately 950,000 valid records. For noisy data, such as duplicate click records (approximately 2%), a deduplication algorithm is used to merge duplicates by user ID, product ID, and timestamp to reduce data redundancy.

[0023] Furthermore, based on the aforementioned behavioral data, obtaining a quantitative score for consumer preferences includes: Based on the behavioral data, a clustering algorithm is used to group the behavioral patterns of different product categories to determine the unique sales pattern characteristics of each category. Based on the aforementioned sales patterns, a quantitative score of consumer preference is obtained by integrating multi-dimensional behavioral indicators through weighted average calculation.

[0024] Furthermore, the unique sales patterns and characteristics of each product category were identified, including: Extract user behavior records related to product categories from behavioral data, perform preliminary cleaning and formatting of user interaction data for each product category, and obtain basic data on structured behavioral patterns. Clustering algorithms are used to group the structured behavioral pattern data, and user behavior records for different product categories are divided into patterns to determine the behavioral pattern clusters for each category. Based on the results of behavioral pattern clustering, we analyze the frequency of user interaction and preference changes of each type of product in different time periods to obtain preliminary characteristics of the sales patterns of each product category. By further exploring the preliminary characteristics of sales patterns and combining them with the classification information of product categories, we can determine whether there are significant differences in patterns. If the characteristics of a certain product category are significantly different from those of other product categories, it will be marked as a unique feature category. For product categories marked with unique characteristics, extract the core data points from their behavioral pattern clusters, obtain the corresponding sales characteristic details, and determine their unique sales pattern manifestations. Based on the unique sales patterns identified, the sales characteristics of each product category are integrated to generate a highly targeted set of sales pattern features.

[0025] Specifically, in this embodiment, based on the cleaned behavioral dataset, cluster analysis can be performed on the behavioral patterns of different product categories to reveal the unique sales patterns of each category. Assume there is a dataset containing 1 million user behavior records, covering three major categories: clothing, electronics, and home furnishings. Each record includes fields such as user ID, product ID, browsing duration (in minutes), number of purchases, and access timestamp. In the clustering phase, the K-means algorithm is selected and implemented using Python's scikit-learn library. The number of clusters, K=3 (corresponding to the three major categories), is set, and the reasonableness of the K value is verified using the Elbow Method.

[0026] Furthermore, based on the aforementioned sales pattern characteristics, multi-dimensional behavioral indicators are integrated by calculating a weighted average, including: By extracting pageview count and conversion rate from sales pattern characteristics, and analyzing the correlation between the two, a preliminary judgment on sales patterns can be obtained. If the number of views is high but the conversion rate is lower than the preset threshold, multi-dimensional data collection will be carried out on user behavior to obtain behavioral indicators involving page dwell time, click path and interaction frequency. Based on the collected behavioral indicators, a weighted average method is used to process the data of each dimension and calculate a comprehensive consumer preference quantitative score.

[0027] Specifically, in this embodiment, in response to the sales pattern of high page views but low conversion rates, data analysis is first used to automatically extract data from a certain e-commerce platform over the past 30 days. It is found that a certain product has 50,000 page views, but the purchase conversion rate is only 1.2%, which is far lower than the industry average of 3.5%. Therefore, it is determined that there is a problem of high consumer interest but low purchase intention. Next, multi-dimensional behavioral indicators are integrated, including browsing time, page dwell depth, number of favorites, and comment interaction rate. Assuming that the average browsing time for a certain product is 3.5 minutes (the industry average is 2.8 minutes), the page dwell depth is 4.2 pages (the average is 3.5 pages), the percentage of favorites is 8% (the average is 5%), and the comment interaction rate is 2.5% (the average is 1.8%), a consumer preference quantitative score is calculated using a weighted average algorithm. The weights are set as follows: browsing time 0.3, dwell depth 0.3, favorites percentage 0.2, and interaction rate 0.2. The calculation formula is: score = 3.5 × 0.3 + 4.2 × 0.3 + 8 × 0.2 + 2.5 × 0.2 = 1.05 + 1.26 + 1.6 + 0.5 = 4.41 (out of 5). Where, S: the final calculated quantitative score of consumer preference, Xi W represents the actual value of the i-th user behavior metric (such as page dwell time, click path depth, percentage of favorites, interaction rate, etc.). i : The weight coefficient corresponding to the i-th behavioral indicator (the sum of all weights is 1), i=1...n, n: the total number of behavioral indicators involved in the calculation.

[0028] Furthermore, based on the aforementioned consumer preference quantification score, the inventory plan is obtained as follows: From the obtained quantitative scores of consumer preferences, time series data is extracted, and a time series forecasting model is used to predict future sales dynamics, determine potential inventory demand fluctuations, and obtain inventory plans. If the potential inventory demand fluctuation exceeds the preset threshold, the multi-category inventory allocation ratio is adjusted according to the forecast results to obtain an optimized inventory plan.

[0029] Furthermore, time series forecasting models are used to predict future sales dynamics and assess potential inventory demand fluctuations, including: Time series data is cleaned and formatted to obtain structured time series records; For structured time series records, a time series forecasting model is used for analysis. An autoregressive integral moving average model is applied to process historical data to determine future sales trends and thus determine inventory plans.

[0030] Specifically, in this embodiment, the inventory plan is optimized by judging potential inventory demand fluctuations and adjusting the multi-category inventory allocation ratio: the inventory demand fluctuations for the next 30 days are predicted by using historical sales data and machine learning models. Assuming that the predicted demand for a certain category A is 1,000 units with a standard deviation of 150 units, and the preset fluctuation threshold is a standard deviation of 100 units, the calculated fluctuation value of 150 is greater than the threshold of 100, and the adjustment mechanism is automatically triggered. The analysis shows that category A has a high demand fluctuation risk. Next, based on the forecast results, the multi-category inventory allocation ratio is adjusted using a linear programming algorithm. Assuming the platform has three categories of goods, A, B, and C, with a total inventory capacity of 5,000 units, the initial allocation ratio is 40% for A, 30% for B, and 30% for C, which translates to 2,000 units, 1,500 units, and 1,500 units respectively. The forecast shows that the demand for A increases by 20% to 1,200 units, the demand for B decreases by 10% to 1,350 units, and the demand for C remains unchanged. Using the objective function (minimizing inventory costs and stockout risk), the new ratio is calculated to be 45% for A (2,250 units), 27% for B (1,350 units), and 28% for C (1,400 units), ensuring a total of 5,000 units.

[0031] Furthermore, based on the aforementioned inventory plan, resource utilization is simulated under sales scenarios to determine the final sales optimization parameters, including: The resource allocation information in the inventory plan is categorized and organized to determine the initial configuration status of each type of resource; Based on the initial configuration, a simulation environment for the sales scenario is built, and a preset sales behavior model is used to simulate the scenario to obtain data on the utilization efficiency of resources in different scenarios. For the utilization efficiency data, analyze the degree of matching between resource allocation and sales scenarios. If the degree of matching is lower than the preset threshold, adjust the resource allocation ratio and regenerate the utilization efficiency data. By using the adjusted utilization efficiency data, key resource utilization bottlenecks are extracted, and the range of core parameters affecting sales optimization is determined. Based on the range of core parameters and the results of simulation analysis, the optimal parameter combination for the sales scenario is derived, and the final parameter determination result is obtained.

[0032] Specifically, in this embodiment, for the optimized inventory plan, resource utilization is evaluated and final sales optimization parameters are determined by simulating sales scenarios. The specific implementation method is as follows: First, assume that the optimized inventory plan is for an e-commerce platform to set an inventory of 1000 units for a popular electronic product, distributed across 5 regional warehouses, with an initial inventory of 200 units in each warehouse. A simulated sales scenario is constructed using historical sales data. Assume that the total daily demand follows a normal distribution with a mean of 50 units and a standard deviation of 10 units. A Monte Carlo simulation algorithm is used to generate 30 days of random demand data. The daily sales volume and remaining inventory of each warehouse are calculated, resulting in a total sales volume of 1450 units. Three warehouses have insufficient inventory on day 20, with a cumulative stockout of 100 units. The resource utilization rate is calculated as the actual sales volume divided by the total inventory, i.e., 1450 / 1000 = 145.0%, indicating that the overall inventory is tight, but some regions face the risk of stockouts. Next, the inventory turnover rate of each warehouse was analyzed. It was found that warehouse A had a turnover rate of 1.8, while warehouse B had a rate of 1.2, a significant difference. Using a regional demand forecasting model (based on linear regression, with input variables being historical sales volume and holiday factors, and a forecasting error controlled within 5%), the inventory allocation ratio was adjusted. Warehouse A's inventory was reduced by 10% to 180 units, while warehouse B's inventory was increased to 220 units. After resimulation, resource utilization increased to 148.0%, and stockouts decreased to 60 units. Finally, sales optimization parameters were determined, and a sales discount strategy was optimized using a genetic algorithm. An initial discount rate of 5% was set. Iterative calculations showed that when the discount rate was 8%, total sales increased to 1500 units within 30 days, resource utilization reached 150.0%, and stockouts further decreased to 40 units.

[0033] In this embodiment, the preset sales behavior model is a deterministic simulation model of e-commerce sales scenarios built based on statistical distribution fitting and numerical simulation algorithms. It belongs to mathematical simulation models. This model is a standardized simulation mathematical model pre-constructed based on historical operational data of e-commerce platforms, sales pattern characteristics of multiple product categories, and quantitative results of consumer behavior preferences. It is used to accurately simulate user purchasing behavior, market demand fluctuations, commodity inventory consumption rate, and warehousing resource allocation and utilization patterns in real e-commerce sales scenarios. It provides quantitative basis and simulation support for the simulation verification of optimized inventory plans and the iterative determination of sales optimization parameters. The model incorporates four core technical rules: Demand Simulation Rules: Based on time-series sales forecasts and consumer preference quantification scores, it fits the distribution patterns of market demand for different time periods and product categories, enabling automated generation of demand data across multiple scenarios; Behavior Conversion Rules: Based on the browsing-add-to-purchase conversion paths and efficiency of each product category obtained through cluster analysis, it recreates the entire conversion logic from user interaction to purchase completion; Inventory Consumption Rules: Combining multi-category inventory allocation ratios and inventory safety thresholds, it simulates the real processes of product inventory consumption, stockout triggers, and warehouse scheduling; Scenario Adaptation Rules: Covering typical e-commerce sales scenarios such as daily sales, demand fluctuations, and inventory shortages, it can output core quantitative evaluation indicators such as resource utilization rate, inventory turnover rate, and stockout risk rate. In this technical solution, the model is used to construct a sales scenario simulation environment. The optimized inventory plan is used as input parameters to perform simulation calculations. By analyzing the resource utilization efficiency data output from the simulation, resource allocation bottlenecks are identified, and sales parameters are iteratively optimized to ultimately determine the optimal sales configuration suitable for multiple scenarios.

[0034] Furthermore, by updating the e-commerce system configuration using the determined final sales optimization parameters, a real-time inventory monitoring mechanism is obtained, including: By analyzing the correlation between sales data and inventory status, a dynamic adjustment model is constructed. A logistic regression algorithm is used to process the mapping between historical sales data and inventory changes, obtaining an optimization benchmark for sales parameters. Based on this benchmark, sales parameters are adjusted and updated to the e-commerce system's configuration module, completing the system settings synchronization process and determining the latest system configuration status. If an abnormal inventory status is detected after the system configuration update, the real-time monitoring module is triggered to obtain the deviation between inventory management data and sales data, determining whether the inventory is below a preset threshold. If the inventory is below the preset threshold, a replenishment signal is sent to relevant modules through a data synchronization mechanism to obtain the execution status of the replenishment plan and determine the initiation conditions for the replenishment process. The real-time monitoring mechanism continuously tracks changes in inventory status, analyzes the matching degree between sales data and inventory management data, and obtains real-time updates of inventory status. Based on these real-time updates, the frequency parameters of the monitoring mechanism are adjusted and synchronized to the e-commerce system's operating module, determining the optimized configuration of the monitoring mechanism. Through this optimized configuration, the execution efficiency of data synchronization is analyzed, and the response rules in the system settings are updated to achieve a stable operating state for the e-commerce system.

[0035] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Various modifications and improvements made to the technical solutions of the present invention by those skilled in the art without departing from the spirit of the present invention should fall within the protection scope defined by the claims of the present invention.

Claims

1. A method for e-commerce inventory monitoring and sales optimization based on multi-category operational data, characterized in that, include: Extract behavioral data on consumer browsing, adding to cart, and purchase records from e-commerce platform databases; Based on the aforementioned behavioral data, a quantitative score of consumer preference is obtained; Based on the aforementioned consumer preference quantification scores, an inventory plan is obtained; Based on the aforementioned inventory plan, resource utilization is simulated under sales scenarios to determine the final sales optimization parameters. By updating the e-commerce system configuration with the determined final sales optimization parameters, a real-time inventory monitoring mechanism can be obtained.

2. The e-commerce inventory monitoring and sales optimization method based on multi-category operational data according to claim 1, characterized in that, The behavioral data extracted from e-commerce platform databases, including consumer browsing, adding to cart, and purchase records, includes: By obtaining consumers' browsing history, shopping cart items, and purchase records from the e-commerce platform's database, an initial behavioral data set is formed, and preliminary data extraction is completed. Data cleaning techniques are used to process the initial behavioral data set, identify noisy data and missing value fields, and obtain a preliminarily cleaned data set. If outliers still exist in the dataset after initial cleaning, a second filtering is performed using a preset threshold range to remove data records that do not meet the standards, thus determining the behavioral data.

3. The e-commerce inventory monitoring and sales optimization method based on multi-category operational data according to claim 1, characterized in that, Based on the aforementioned behavioral data, obtaining a quantitative score for consumer preferences includes: Based on the behavioral data, a clustering algorithm is used to group the behavioral patterns of different product categories to determine the unique sales pattern characteristics of each category. Based on the aforementioned sales patterns, a quantitative score of consumer preference is obtained by integrating multi-dimensional behavioral indicators through weighted average calculation.

4. The e-commerce inventory monitoring and sales optimization method based on multi-category operational data according to claim 3, characterized in that, Identifying the unique sales patterns and characteristics of each product category includes: Extract user behavior records related to product categories from behavioral data, perform preliminary cleaning and formatting of user interaction data for each product category, and obtain basic data on structured behavioral patterns. Clustering algorithms are used to group the structured behavioral pattern data, and user behavior records for different product categories are divided into patterns to determine the behavioral pattern clusters for each category. Based on the results of behavioral pattern clustering, we analyze the frequency of user interaction and preference changes of each type of product in different time periods to obtain preliminary characteristics of the sales patterns of each product category. By further exploring the preliminary characteristics of sales patterns and combining them with the classification information of product categories, we can determine whether there are significant differences in patterns. If the characteristics of a certain product category are significantly different from those of other product categories, it will be marked as a unique feature category. For product categories marked with unique characteristics, extract the core data points from their behavioral pattern clusters, obtain the corresponding sales characteristic details, and determine their unique sales pattern manifestations. Based on the unique sales patterns identified, the sales characteristics of each product category are integrated to generate a highly targeted set of sales pattern features.

5. The e-commerce inventory monitoring and sales optimization method based on multi-category operational data according to claim 3, characterized in that, Based on the aforementioned sales patterns, multi-dimensional behavioral indicators are integrated by calculating a weighted average, including: By extracting pageview count and conversion rate from sales pattern characteristics, and analyzing the correlation between the two, a preliminary judgment on sales patterns can be obtained. If the number of views is high but the conversion rate is lower than the preset threshold, multi-dimensional data collection will be carried out on user behavior to obtain behavioral indicators involving page dwell time, click path and interaction frequency. Based on the collected behavioral indicators, a weighted average method is used to process the data of each dimension and calculate a comprehensive consumer preference quantitative score.

6. The e-commerce inventory monitoring and sales optimization method based on multi-category operational data according to claim 1, characterized in that, Based on the aforementioned consumer preference quantification score, the inventory management plan includes: From the obtained quantitative scores of consumer preferences, time series data is extracted, and a time series forecasting model is used to predict future sales dynamics, determine potential inventory demand fluctuations, and obtain inventory plans. If the potential inventory demand fluctuation exceeds the preset threshold, the multi-category inventory allocation ratio is adjusted according to the forecast results to obtain an optimized inventory plan.

7. The e-commerce inventory monitoring and sales optimization method based on multi-category operational data according to claim 6, characterized in that, Using time series forecasting models to predict future sales dynamics and assess potential inventory demand fluctuations includes: Time series data is cleaned and formatted to obtain structured time series records; For structured time series records, a time series forecasting model is used for analysis. An autoregressive integral moving average model is applied to process historical data to determine future sales trends and thus determine inventory plans.

8. The e-commerce inventory monitoring and sales optimization method based on multi-category operational data according to claim 1, characterized in that, Based on the aforementioned inventory plan, resource utilization under simulated sales scenarios is used to determine the final sales optimization parameters, including: The resource allocation information in the inventory plan is categorized and organized to determine the initial configuration status of each type of resource; Based on the initial configuration, a simulated environment for the sales scenario is built to obtain data on resource utilization efficiency under different scenarios; For the utilization efficiency data, analyze the degree of matching between resource allocation and sales scenarios. If the degree of matching is lower than the preset threshold, adjust the resource allocation ratio and regenerate the utilization efficiency data. By using the adjusted utilization efficiency data, key resource utilization bottlenecks are extracted, and the range of core parameters affecting sales optimization is determined. Based on the range of core parameters and the results of simulation analysis, the optimal parameter combination for the sales scenario is derived, and the final parameter determination result is obtained.

Images