Furniture sales analysis system based on big data

Abstract

The invention discloses a furniture sales analysis system based on big data, and relates to the technical field of furniture sales, and the system collects and integrates e-commerce platform browsing data, ordering records, offline sales data and logistics system information in real time, and standardizes and stores the data in a unified database in combination with an ETL technology; a seasonal rule of furniture demands is analyzed and mined by using a time sequence, regional consumption preferences are identified in combination with a clustering algorithm, dynamic changes and regional differences of market demands are captured, the problem that a traditional method lags behind a rapid change market is solved, and personalized demand increase can be reflected; in the aspect of inventory management, regional inventory configuration is dynamically optimized by adopting a linear programming algorithm based on a demand prediction result, inventory is reasonably allocated by balancing a regional supply-demand relationship on the basis of preferentially meeting a demand-prosperous region, inventory backlog and resource waste caused by regional difference are reduced, and in addition, regional lowest and highest inventory limitations are introduced, so that the demand-prosperous region is optimized. And the problem of differentiated processing of regional requirements is effectively solved.

Description

Technical Field

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

[0002] In the field of furniture sales, there is a widespread problem of unsaleable inventory backlog. To address this problem, the current mainstream optimization analysis method is to forecast demand by reviewing past sales records. However, this optimization analysis method lags behind in the context of a rapidly changing market environment and is unable to meet the needs of dynamic adjustments.

[0003] In addition, there are many types of furniture products, and consumer preferences are affected by many factors. With the increasing demand for personalized customized furniture, the market share of many standardized furniture has dropped significantly. Traditional analysis methods cannot capture the impact of multiple factors in a timely manner, making it difficult for companies to adjust production and inventory in a timely manner. Some products are unsalable for a long time because they do not meet market demand, resulting in inventory backlogs and increased storage costs.

[0004] Moreover, some traditional analysis methods are also insufficient in dealing with regional demand differences. Furniture consumption demand varies significantly due to geographical factors. Some styles of furniture may sell well in urban areas but be unsalable in rural areas. Traditional analysis methods often formulate inventory allocation strategies based on empirical judgment or sales data grouping, which can easily lead to overstocking of some styles in certain areas. Therefore, a more dynamic and accurate furniture sales analysis system based on big data is urgently needed to solve such problems. Summary of the invention

[0005] In view of the above existing problems, the present invention is proposed.

[0006] The present invention provides a furniture sales analysis system based on big data to solve the problems of traditional furniture sales analysis methods in responding to market changes and identifying regional demand differences, and the long-term existence of unsalable inventory, which limits the resource utilization of enterprises.

[0007] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0008] The embodiment of the present invention provides a furniture sales analysis system based on big data, which includes:

[0009] The data management module is responsible for collecting, integrating and storing furniture sales data. The data management module uses extraction, transformation and loading (ETL) technology to integrate sales data. Sales data includes browsing data and order records obtained from the merchant platform, data from the offline sales system, and inventory and delivery information from the logistics system.

[0010] Demand analysis and forecasting module, based on historical and real-time sales data, to explore market demand patterns and forecast sales trends;

[0011] In the demand analysis and forecasting module, the integrated sales data is used to explore the seasonal changes in furniture demand through time series analysis, and the clustering algorithm is combined to identify the consumption preferences of products in different regions. The forecast results are output, including the demand popularity of each product, the possible peak demand time points in the future, and the consumption areas with strong demand;

[0012] Inventory optimization module, used to dynamically adjust regional inventory configuration and balance regional demand;

[0013] Slow-selling product management module, used to mark slow-selling products;

[0014] In the slow-selling product management module, the allocation adjustment data of the inventory optimization module and the real-time collected online and offline sales records are combined to identify and mark the slow-selling products; based on the marking results, the staff will formulate a promotion plan.

[0015] As a preferred solution of the furniture sales analysis system based on big data described in the present invention, the inventory optimization module takes the demand forecasting result of the demand analysis and forecasting module as input, and allocates and adjusts the inventory through a linear programming algorithm.

[0016] As a preferred solution of the furniture sales analysis system based on big data described in the present invention, the analysis method of the furniture sales analysis system includes:

[0017] Step S1, collect furniture sales data, including browsing and ordering data of e-commerce platforms, sales records of offline systems, and inventory and delivery information of logistics systems; integrate sales data from different sources through extraction, transformation, and loading ETL technology, and store them in the data management module;

[0018] Step S2, based on the integrated sales data, using time series analysis and clustering algorithms, to explore the demand trends and regional distribution patterns of different furniture products, and to predict sales conditions. The prediction results include product demand popularity, potential demand peak time points, and consumption areas;

[0019] Step S3, adjusting the regional inventory configuration according to the demand forecast result; the method of adjusting the regional inventory configuration is: using a linear programming algorithm to reallocate the existing inventory, and balancing the supply and demand of the regions with equal demand on the premise of meeting the demand of the regions with strong demand;

[0020] Step S4, using the inventory reallocation data in step S3, combined with the real-time collected browsing and ordering data of the e-commerce platform and the sales records of the offline system, mark the unsaleable products online and offline, and the staff will design a promotion plan based on the marking results.

[0021] As a preferred solution of the furniture sales analysis system based on big data described in the present invention, the steps of integrating sales data from different sources by extracting, converting and loading ETL technology and storing them in the data management module are as follows:

[0022] Data is extracted from multiple data sources, including e-commerce platforms, offline sales systems, and logistics systems. Data from each source is extracted separately to form an initial data set. The extraction formula is:

[0023] D e1 =E(T1, F1), D e2 =E(T2, F2), D e3 =E(T3,F3),

[0024] Among them, D e1 , D e2 , D e3 They represent the data sets extracted from the e-commerce platform, offline system, and logistics system, respectively. T1, T2, and T3 represent the original data tables, respectively. F1, F2, and F3 are the extraction conditions of the three different platforms, including the screening time interval and order status, respectively. E(·) represents the extraction operation;

[0025] The extracted data is formatted and cleaned, and the format standardization formula is:

[0026] D t1 =T(D e1 , S1), D t2 =T(D e2 , S2), D t3 =T(D e3 , S3),

[0027] Among them, D t1 , D t2 , D t3 They represent the transformed data sets of different platforms, S1, S2, and S3 are the transformation rules of the corresponding platforms, and T(·) represents the standardization operation.

[0028] Perform data cleaning, the cleaning formula is:

[0029] D c1 =C(D t1 , R1), D c2 =C(D t2 , R2), D c3 =C(Dt3 , R3),

[0030] Among them, D c1 , D c2 , D c3 are the cleaned data sets, R1, R2, and R3 are the cleaning rules of different platforms, and C(·) represents the cleaning operation;

[0031] The converted and cleaned data is stored in the data management module for data loading. The data loading formula is:

[0032] DB=L(D c1 , P1, I1)∪L(D c2 , P2, I2)∪L(D c3 , P3, I3),

[0033] Where DB represents the target database, P1, P2, and P3 represent the partitioning rules of each data set, I1, I2, and I3 represent the indexing strategies of each data set, and L(·) represents the loading operation.

[0034] As a preferred solution of the furniture sales analysis system based on big data described in the present invention, the steps of mining the demand trends and regional distribution patterns of different furniture products based on the integrated sales data and using time series analysis and clustering algorithms to predict the sales situation are as follows:

[0035] To perform time series analysis:

[0036] Arrange the historical sales data in chronological order to form a time series. The time series definition formula is:

[0037] S p ={x p，1 , x p，2 , ..., x p，n},

[0038] Among them, S p represents the time series of product p, x p，t represents the sales volume of product p at time t, t∈1,2,...,n, n is the total length of the time series,

[0039] The additive model is used to decompose the time series into trend, seasonal and random components. The decomposition formula is:

[0040] x p，t =T p，t +S p，t +R p，t ,

[0041] Among them, T p，t is the trend component, Sp，t is the seasonal component, R p，t is a random component,

[0042] Using linear regression model to predict trend component T p，t , the trend prediction formula is:

[0043] T p，t =α p +β p t,

[0044] Among them, α p , β p is the regression coefficient, and t is the time point.

[0045] As a preferred solution of the furniture sales analysis system based on big data described in the present invention, the step of mining the demand trends and regional distribution patterns of different furniture products and predicting the sales situation also includes:

[0046] Perform cluster analysis:

[0047] Extract the feature vector and construct the feature vector with the sales data of each region as the dimension. The feature vector is defined as:

[0048] v p，r =[v p，r，1 , v p，r，2 , …, v p，r，m ],

[0049] Among them, v p，r is the characteristic vector of product p in region r, v p，r，i represents the sales volume of product p at the i-th time point in region r, i∈1,2,...m, m represents the total time points,

[0050] Clustering is performed by minimizing the intra-cluster variance, and the clustering goal is:

[0051]

[0052] Where k is the number of clusters, C i represents the i-th cluster, v represents the specific cluster, μ i is the cluster center.

[0053] As a preferred solution of the furniture sales analysis system based on big data described in the present invention, in the step of predicting the sales situation, the final prediction result is output as H p，r :

[0054] H p，r =f(T p，t , S p，t ), T peak =max(Sp，t ), R main = arg max(H p，r ),

[0055] Among them, H p，r represents the demand heat of product p in region r, f(·) represents the heat demand function, T peak represents the peak demand time point, R main Indicates areas of high demand.

[0056] As a preferred solution of the furniture sales analysis system based on big data described in the present invention, the step of adjusting the regional inventory configuration according to the demand forecast result is:

[0057] To define inventory requirements:

[0058] The inventory demand of each region r is calculated based on the demand forecast results. The inventory demand calculation formula is:

[0059] Q r =∑ p (H p，r ·w p ),

[0060] Among them, Q r represents the total inventory demand in region r, H p，r is the demand heat of product p in region r, w p is the weight factor of product p, which is used to indicate the priority of different products.

[0061] The goal is to meet the needs of areas with high demand while balancing the inventory allocation in other areas. The optimization goal is defined and the optimization objective function is minZ:

[0062] min Z=∑ r |Q r -S r |,

[0063] Among them, Z represents the total deviation between all regional demands and actual inventory, S r is the existing inventory in region r.

[0064] As a preferred solution of the furniture sales analysis system based on big data described in the present invention, the step of adjusting the regional inventory configuration according to the demand forecast result also includes:

[0065] Introduce multiple constraints for inventory adjustments, including:

[0066] Total inventory unchanged constraint:

[0067] ∑ r S′ r =∑r S r ,

[0068] Among them, S′ r is the adjusted inventory level,

[0069] Regional minimum inventory constraints:

[0070] S′ r ≥L r ,

[0071] Among them, L r is the minimum inventory requirement for region r;

[0072] Using the linear programming method to solve the adjustment plan, the overall inventory adjustment model is:

[0073] minZ=∑ r |Q r -S′ r |,

[0074] ∑ r S′ r =∑ r S r , S′ r ≥L r , S′ r ≤U r ,

[0075] Among them, U r is the maximum inventory limit for region r;

[0076] Calculate the adjusted inventory S′ for each region r , output allocation plan, adjust inventory calculation formula:

[0077] S′ r =S r +Δ r ,

[0078] in,

[0079] Δ r is the inventory adjustment in region r, solved by linear programming.

[0080] As a preferred solution of the furniture sales analysis system based on big data described in the present invention, the step of marking the unsalable products online and offline by using the data of inventory reallocation in step S3, combined with the browsing and ordering data of the e-commerce platform collected in real time and the sales records of the offline system, is as follows:

[0081] By comparing inventory and sales data, we can identify unsalable products. The unsalable coefficient is calculated as follows:

[0082]

[0083] Among them, K p，r represents the unsalable coefficient of product p in region r, S′ p，r is the inventory of product p in region r, M p，r is the sales volume of product p in region r,

[0084] Determine the unsalable threshold and set the unsalable threshold θ. If K p，r >θ, then Flag(p, r) = 1, otherwise, Flag(p, r) = 0,

[0085] Among them, Flag(p, r) indicates whether product p is a slow-selling product in region r, 1 means slow-selling, 0 means not slow-selling,

[0086] Combined with the real-time collected sales records, the unsalable status is dynamically updated. The dynamic unsalable judgment formula is:

[0087]

[0088] Among them, ΔM p，r For the real-time increase in sales data,

[0089] According to the marking results, the promotion budget is allocated for the staff to implement the promotion plan.

[0090] The beneficial effects of the present invention are as follows: the present invention collects and integrates browsing data, order records, offline sales data and logistics system information of the e-commerce platform in real time, and stores them in a unified database in a standardized manner in combination with the ETL technology; then, the seasonal law of furniture demand is mined by time series analysis, and regional consumption preferences are identified in combination with clustering algorithms, so as to capture the dynamic changes of market demand and regional differences, overcome the problem that traditional methods lag behind the rapidly changing market, and can also reflect the increase in personalized demand; in terms of inventory management, based on the demand forecast results, a linear programming algorithm is used to dynamically optimize regional inventory configuration, and by balancing the regional supply and demand relationship, inventory is reasonably allocated on the basis of giving priority to satisfying areas with strong demand, thereby reducing inventory backlogs and resource waste caused by regional differences. In addition, regional minimum and maximum inventory limits are introduced to effectively solve the problem of differentiated processing of regional demand.

[0091] In the present invention, the slow-selling product management module dynamically updates the slow-selling status by calculating the slow-selling coefficient and determining the threshold value in combination with real-time sales data, and marks the slow-selling products. The staff can design promotion plans based on the marking results, accurately promote the conversion and sales of slow-selling products, improve inventory turnover, and effectively reduce storage costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0092] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0093] Figure 1 It is a schematic diagram of the framework of the furniture sales analysis system based on big data of the present invention. DETAILED DESCRIPTION

[0094] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0095] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0096] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0097] Example 1, reference Figure 1 , this embodiment provides a furniture sales analysis system based on big data, including:

[0098] The data management module is responsible for collecting, integrating and storing furniture sales data. The data management module uses extraction, transformation and loading (ETL) technology to integrate sales data. Sales data includes browsing data and order records obtained from the merchant platform, data from the offline sales system, and inventory and delivery information from the logistics system.

[0099] Demand analysis and forecasting module, based on historical and real-time sales data, to explore market demand patterns and forecast sales trends;

[0100] In the demand analysis and forecasting module, the integrated sales data is used to explore the seasonal changes in furniture demand through time series analysis, and the clustering algorithm is combined to identify the consumption preferences of products in different regions. The forecast results are output, including the demand popularity of each product, the possible peak demand time points in the future, and the consumption areas with strong demand;

[0101] Inventory optimization module, used to dynamically adjust regional inventory configuration and balance regional demand;

[0102] The inventory optimization module uses the demand forecast results of the demand analysis and forecasting module as input and adjusts the inventory allocation through a linear programming algorithm;

[0103] Slow-selling product management module, used to mark slow-selling products;

[0104] In the slow-moving product management module, the allocation adjustment data of the inventory optimization module and the real-time online and offline sales records are combined to identify and mark slow-moving products; based on the marking results, the staff will formulate a promotion plan;

[0105] In this embodiment, an analysis method of the furniture sales analysis system is also provided, including:

[0106] Step S1, collect furniture sales data, including browsing and ordering data of e-commerce platforms, sales records of offline systems, and inventory and delivery information of logistics systems; integrate sales data from different sources through extraction, transformation, and loading ETL technology, and store them in the data management module;

[0107] The steps to integrate sales data from different sources through extraction, transformation and loading (ETL) technology and store it in the data management module are as follows:

[0108] Data is extracted from multiple data sources, including e-commerce platforms, offline sales systems, and logistics systems. Data from each source is extracted separately to form an initial data set. The extraction formula is:

[0109] D e1 =E(T1, F1), D e2 =E(T2, F2), D e3 =E(T3,F3),

[0110] Among them, D e1 , D e2 , D e3 They represent the data sets extracted from the e-commerce platform, offline system, and logistics system, respectively. T1, T2, and T3 represent the original data tables, respectively. F1, F2, and F3 are the extraction conditions of the three different platforms, including the screening time interval and order status, respectively. E(·) represents the extraction operation;

[0111] The extracted data is formatted and cleaned, and the format standardization formula is:

[0112] D t1 =T(D e1 , S1), D t2 =T(D e2 , S2), D t3 =T(D e3 , S3),

[0113] Among them, D t1 , D t2 , D t3 They represent the transformed data sets of different platforms, S1, S2, and S3 are the transformation rules of the corresponding platforms, and T(·) represents the standardization operation.

[0114] Perform data cleaning, the cleaning formula is:

[0115] D c1 =c(D t1 , R1), D c2 =c(D t2 , R2), D c3 =c(D t3 , R3),

[0116] Among them, D c1 , D c2 , D c3 are the cleaned data sets, R1, R2, and R3 are the cleaning rules of different platforms, and c(·) represents the cleaning operation;

[0117] The converted and cleaned data is stored in the data management module for data loading. The data loading formula is:

[0118] DB=L(D c1 , P1, I1)∪L(D c2 , P2, I2)∪L(D c3 , P3, I3),

[0119] Where DB represents the target database, P1, P2, and P3 represent the partitioning rules of each data set, I1, I2, and I3 represent the indexing strategies of each data set, and L(·) represents the loading operation;

[0120] Specifically, the ETL process is used to normalize and integrate data from multiple sources to improve data quality, and the data is stored in a unified format through loading operations to facilitate subsequent analysis.

[0121] Step S2, based on the integrated sales data, using time series analysis and clustering algorithms, to explore the demand trends and regional distribution patterns of different furniture products, and to predict sales conditions. The prediction results include product demand popularity, potential demand peak time points, and consumption areas;

[0122] Based on the integrated sales data, time series analysis and clustering algorithms are used to explore the demand trends and regional distribution patterns of different furniture products. The steps for predicting sales are as follows:

[0123] To perform time series analysis:

[0124] Arrange the historical sales data in chronological order to form a time series. The time series definition formula is:

[0125] S p ={x p，1 , x p，2 , ..., x p，n},

[0126] Among them, S p represents the time series of product p, x p，t represents the sales volume of product p at time t, t∈1,2,...,n, n is the total length of the time series,

[0127] The additive model is used to decompose the time series into trend, seasonal and random components. The decomposition formula is:

[0128] x p，t =T p，t +S p，t +R p，t ,

[0129] Among them, T p，t is the trend component, S p，t is the seasonal component, R p，t is a random component,

[0130] Using linear regression model to predict trend component T p，t , the trend prediction formula is:

[0131] T p，t =α p +β p t,

[0132] Among them, α p , β p is the regression coefficient, t is the time point;

[0133] The steps of exploring the demand trends and regional distribution patterns of different furniture products and forecasting sales also include:

[0134] Perform cluster analysis:

[0135] Extract the feature vector and construct the feature vector with the sales data of each region as the dimension. The feature vector is defined as:

[0136] v p，r =[v p，r，1 , v p，r，2 , ..., v p，r，m ],

[0137] Among them, v p，r is the characteristic vector of product p in region r, v p，r，irepresents the sales volume of product p at the i-th time point in region r, i∈1,2,...m, m represents the total time points,

[0138] Clustering is performed by minimizing the intra-cluster variance, and the clustering goal is:

[0139]

[0140] Where k is the number of clusters, C i represents the i-th cluster, v represents the specific cluster, μ i is the cluster center,

[0141] In the step of predicting sales, the final prediction result is output as H p，r :

[0142] H p，r =f(T p，t , S p，t ), T peak =max(S p，t ), R main = arg max(H p，r ),

[0143] Among them, H p，r represents the demand heat of product p in region r, f(·) represents the heat demand function, T peak represents the peak demand time point, R main Indicates areas of high demand;

[0144] Specifically, time series analysis is combined with clustering algorithms to forecast demand from multiple dimensions such as trends, seasonality, and regional preferences. Decomposition and regression modeling are used to clarify the patterns of sales changes, and clustering algorithms are used to further explore regional consumption characteristics and determine demand heat and peak time points.

[0145] Step S3, adjusting the regional inventory configuration according to the demand forecast result; the method of adjusting the regional inventory configuration is: using a linear programming algorithm to reallocate the existing inventory, and balancing the supply and demand of the regions with equal demand on the premise of meeting the demand of the regions with strong demand;

[0146] According to the demand forecast results, the steps to adjust the regional inventory configuration are:

[0147] To define inventory requirements:

[0148] The inventory demand of each region r is calculated based on the demand forecast results. The inventory demand calculation formula is:

[0149] Q r =∑ p (H p，r ·w p ),

[0150] Among them, Q r represents the total inventory demand in region r, H p，r is the demand heat of product p in region r, w p is the weight factor of product p, which is used to indicate the priority of different products.

[0151] The goal is to meet the needs of areas with high demand while balancing the inventory allocation in other areas. The optimization goal is defined and the optimization objective function is minZ:

[0152] min Z=∑ r |Q r -S r |,

[0153] Among them, Z represents the total deviation between all regional demands and actual inventory, S r is the existing inventory in region r;

[0154] Based on the demand forecast results, the steps to adjust regional inventory configuration also include:

[0155] Introduce multiple constraints for inventory adjustments, including:

[0156] Total inventory unchanged constraint:

[0157] ∑ r S′ r =∑ r S r ,

[0158] Among them, S′ r is the adjusted inventory level,

[0159] Regional minimum inventory constraints:

[0160] S′ r ≥L r ,

[0161] Among them, L r is the minimum inventory requirement for region r;

[0162] Using the linear programming method to solve the adjustment plan, the overall inventory adjustment model is:

[0163] minZ=∑ r |Q r -S′ r |,

[0164] ∑ r S′ r =∑ r S r , S′ r ≥L r, S′ r ≤U r ,

[0165] Among them, U r is the maximum inventory limit for region r;

[0166] Calculate the adjusted inventory S′ for each region r , output allocation plan, adjust inventory calculation formula:

[0167] S′ r =S r +Δ r ,

[0168] in,

[0169] Δ r is the inventory adjustment amount of region r, solved by linear programming,

[0170] Specifically, by optimizing the model and setting constraints, dynamic matching of demand and inventory is achieved, and linear programming is used to prioritize regional demand while balancing the overall inventory configuration to maximize resource utilization.

[0171] Step S4, using the inventory reallocation data in step S3, combined with the real-time collected browsing and ordering data of the e-commerce platform and the sales records of the offline system, marking the unsalable products online and offline, and the staff will design a promotion plan based on the marking results;

[0172] Using the data of inventory reallocation in step S3, combined with the real-time collected browsing and ordering data of the e-commerce platform and the sales records of the offline system, the steps of marking the unsalable products online and offline are as follows:

[0173] By comparing inventory and sales data, we can identify unsalable products. The unsalable coefficient is calculated as follows:

[0174]

[0175] Among them, K p，r represents the unsalable coefficient of product p in region r, S′ p，r is the inventory of product p in region r, M p，r is the sales volume of product p in region r,

[0176] Determine the unsalable threshold and set the unsalable threshold θ. If K p，r >θ, then Flag(p, r) = 1, otherwise, Flag(p, r) = 0,

[0177] Among them, Flag(p, r) indicates whether product p is a slow-selling product in region r, 1 means slow-selling, 0 means not slow-selling,

[0178] Combined with the real-time collected sales records, the unsalable status is dynamically updated. The dynamic unsalable judgment formula is:

[0179]

[0180] Among them, ΔM p，r For the real-time increase in sales data,

[0181] Allocate promotion budget based on the marking results for staff to implement the promotion plan;

[0182] Specifically, the slow-moving product marking process is based on dynamic inventory and real-time sales data, and is accurately marked through the determination of the slow-moving coefficient and threshold. It is further combined with the promotion intensity allocation model to optimize resource utilization, promote the conversion and sales of slow-moving products, and ultimately improve the overall inventory turnover rate.

[0183] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A furniture sales analysis system based on big data, characterized by: include, The data management module is responsible for collecting, integrating and storing furniture sales data. The data management module uses extraction, transformation and loading (ETL) technology to integrate sales data. Sales data includes browsing data and order records obtained from the merchant platform, data from the offline sales system, and inventory and delivery information from the logistics system. Demand analysis and forecasting module, based on historical and real-time sales data, to explore market demand patterns and forecast sales trends; In the demand analysis and forecasting module, the integrated sales data is used to explore the seasonal changes in furniture demand through time series analysis, and the clustering algorithm is combined to identify the consumption preferences of products in different regions. The forecast results are output, including the demand popularity of each product, the possible peak demand time points in the future, and the consumption areas with strong demand; Inventory optimization module, used to dynamically adjust regional inventory configuration and balance regional demand; Slow-selling product management module, used to mark slow-selling products; In the slow-selling product management module, the allocation adjustment data of the inventory optimization module and the real-time collected online and offline sales records are combined to identify and mark the slow-selling products; based on the marking results, the staff will formulate a promotion plan.

2. The furniture sales analysis system based on big data as claimed in claim 1, characterized in that: The inventory optimization module uses the demand forecasting results of the demand analysis and forecasting module as input and adjusts the inventory allocation through a linear programming algorithm.

3. The furniture sales analysis system based on big data as claimed in claim 2, characterized in that: The analysis method of the furniture sales analysis system includes: Step S1, collecting furniture sales data, including browsing and ordering data on e-commerce platforms, sales records on offline systems, and inventory and delivery information on logistics systems; Integrate sales data from different sources through extraction, transformation, and loading (ETL) technology and store it in the data management module; Step S2, based on the integrated sales data, using time series analysis and clustering algorithms, to explore the demand trends and regional distribution patterns of different furniture products, and to predict sales conditions. The prediction results include product demand popularity, potential demand peak time points, and consumption areas; Step S3, adjusting the regional inventory configuration according to the demand forecast result; the method of adjusting the regional inventory configuration is: using a linear programming algorithm to reallocate the existing inventory, and balancing the supply and demand of the regions with equal demand on the premise of meeting the demand of the regions with strong demand; Step S4, using the inventory reallocation data in step S3, combined with the real-time collected browsing and ordering data of the e-commerce platform and the sales records of the offline system, mark the unsaleable products online and offline, and the staff will design a promotion plan based on the marking results.

4. The furniture sales analysis system based on big data as claimed in claim 3, characterized in that: The steps of integrating sales data from different sources by extracting, transforming and loading ETL technology and storing it in the data management module are: Data is extracted from multiple data sources, including e-commerce platforms, offline sales systems, and logistics systems. Data from each source is extracted separately to form an initial data set. The extraction formula is: <h2 style=";text-align:left;direction:ltr">D<h2 style=";text-align:left;direction:ltr"> e1 <h2 style=";text-align:left;direction:ltr"> (E(T1, F1), D)<h2 style=";text-align:left;direction:ltr"> e2 <h2 style=";text-align:left;direction:ltr"> (E(T2, F2), D)<h2 style=";text-align:left;direction:ltr"> e3 <h2 style=";text-align:left;direction:ltr"> (E(T3, F3)) Among them, D e1 , D e2 , D e3 They represent the data sets extracted from the e-commerce platform, offline system, and logistics system, respectively. T1, T2, and T3 represent the original data tables, respectively. F1, F2, and F3 are the extraction conditions of the three different platforms, including the screening time interval and order status, respectively. E(·) represents the extraction operation; The extracted data is formatted and cleaned, and the format standardization formula is: D t1 =T(D e1 ,S1),D t2 =T(D e2 ,S2),D t3 =T(D e3 ,S3), Among them, D t1 , D t2 , D t3 They represent the transformed data sets of different platforms, S1, S2, and S3 are the transformation rules of the corresponding platforms, and T(·) represents the standardization operation. Perform data cleaning, the cleaning formula is: D c1 =C(D t1 ,R1),D c2 =C(D t2 ,R2),D c3 =C(D t3 ,R3), Among them, D c1 , D c2 , D c3 are the cleaned data sets, R1, R2, and R3 are the cleaning rules of different platforms, and C(·) represents the cleaning operation; The converted and cleaned data is stored in the data management module for data loading. The data loading formula is: DB=L(D c1 ,P1,I1)∪L(D c2 ,P2,I2)∪L(D c3 ,P3,I3), Where DB represents the target database, P1, P2, and P3 represent the partitioning rules of each data set, I1, I2, and I3 represent the indexing strategies of each data set, and L(·) represents the loading operation.

5. The furniture sales analysis system based on big data as claimed in claim 4, characterized in that: The steps of using time series analysis and clustering algorithm based on the integrated sales data to mine the demand trends and regional distribution patterns of different furniture products and predict the sales situation are as follows: To perform time series analysis: Arrange the historical sales data in chronological order to form a time series. The time series definition formula is: S p ={x p，1 ,x p，2 ,...,x p，n }, Among them, S p represents the time series of product p, x p，t represents the sales volume of product p at time t, t∈1,2,...,n, n is the total length of the time series, The additive model is used to decompose the time series into trend, seasonal and random components. The decomposition formula is: x p，t =T p，t +S p，t +R p，t , Among them, T p，t is the trend component, S p，t is the seasonal component, R p，t is a random component, Using linear regression model to predict trend component T p，t , the trend prediction formula is: T p，t =a p +b p t, Among them, α p , β p is the regression coefficient, and t is the time point.

6. The furniture sales analysis system based on big data according to claim 5, characterized in that: The step of mining the demand trends and regional distribution patterns of different furniture products and predicting sales conditions also includes: Perform cluster analysis: Extract the feature vector and construct the feature vector with the sales data of each region as the dimension. The feature vector is defined as: vp,r=[vp,r,1,vp,r,2,...,vp,r,m], Among them, v p，r is the characteristic vector of product p in region r, v p，r，i represents the sales volume of product p at the i-th time point in region r, i∈1,2,...m, m represents the total time points, Clustering is performed by minimizing the intra-cluster variance, and the clustering goal is: Where k is the number of clusters, C i represents the i-th cluster, v represents the specific cluster, μ i is the cluster center.

7. The furniture sales analysis system based on big data according to claim 6, characterized in that: In the step of predicting the sales situation, the final prediction result is output as H p，r : H p，r =f(T p，t ,S p，t ),T peak =max(S p，t ),R main =argmax(H p，r ), Among them, H p，r represents the demand heat of product p in region r, f(·) represents the heat demand function, T peak represents the peak demand time point, R main Indicates areas of high demand.

8. The furniture sales analysis system based on big data according to claim 7, characterized in that: The steps of adjusting the regional inventory configuration according to the demand forecast result are: To define inventory requirements: The inventory demand of each region r is calculated based on the demand forecast results. The inventory demand calculation formula is: Q r =∑ p (H p，r ·w p ), Among them, Q r represents the total inventory demand in region r, H p，r is the demand heat of product p in region r, w p is the weight factor of product p, which is used to indicate the priority of different products. The goal is to meet the needs of areas with high demand while balancing the inventory allocation in other areas. The optimization goal is defined and the optimization objective function is minZ: minZ=∑ r |Q r -S r |, Among them, Z represents the total deviation between all regional demands and actual inventory, S r is the existing inventory in region r.

9. The furniture sales analysis system based on big data according to claim 8, characterized in that: The step of adjusting the regional inventory configuration according to the demand forecast result also includes: Introduce multiple constraints for inventory adjustments, including: Total inventory unchanged constraint: ∑ r S′ r =∑ r S r , Among them, S′ r is the adjusted inventory level, Regional minimum inventory constraints: S′ r ≥L r , Among them, L r is the minimum inventory requirement for region r; Using the linear programming method to solve the adjustment plan, the overall inventory adjustment model is: minZ=∑ r |Q r -S′ r |, ∑ r S′ r =∑ r S r ,S′ r ≥L r ,S′ r ≤U r , Among them, U r is the maximum inventory limit for region r; Calculate the adjusted inventory S′ for each region r , output allocation plan, adjust inventory calculation formula: S′ r =S r +D r , in, Δ r is the inventory adjustment in region r, solved by linear programming.

10. The furniture sales analysis system based on big data according to claim 9, characterized in that: The step of marking the unsalable products online and offline by using the data of inventory reallocation in step S3, combined with the browsing and ordering data of the e-commerce platform and the sales records of the offline system collected in real time, is as follows: By comparing inventory and sales data, we can identify unsalable products. The unsalable coefficient is calculated as follows: Among them, K p，r represents the unsalable coefficient of product p in region r, S′ p，r is the inventory of product p in region r, M p，r is the sales volume of product p in region r, Determine the unsalable threshold and set the unsalable threshold θ. If K p，r >θ, then Flag(p, r) = 1, otherwise, Flag(p, r) = 0, Among them, Flag(p, r) indicates whether product p is a slow-selling product in region r, 1 means slow-selling, 0 means not slow-selling, Combined with the real-time collected sales records, the unsalable status is dynamically updated. The dynamic unsalable judgment formula is: Among them, ΔM p，r For the real-time increase in sales data, According to the marking results, the promotion budget is allocated for the staff to implement the promotion plan.

Images