Customer satisfaction analysis method and system based on big data

Abstract

The invention provides a customer satisfaction analysis method and system based on big data, and belongs to the field of data analysis. The problem of low customer satisfaction analysis efficiency is solved; the method specifically comprises the following steps: S1, obtaining order data; s2, calculating a sales coefficient of each product according to the order data; according to the questionnaire information and the sales coefficient, the satisfaction degree of the customer for each product is analyzed, and hot products and unpopular products are determined; s3, obtaining the product type of the hot product as a quasi type; order data of quasi-type products sold by different enterprises are obtained, SARIMA regression is used, order number differentiation of hot products in different seasons is analyzed, and a sale scheme of the hot products is formulated; s4, summarizing and feeding back unpopular products and sales schemes; according to the method, the order data and the questionnaire data of different types of products are analyzed, the satisfaction degree of each type of product of a customer is analyzed from a relatively objective perspective, and reference is provided for enterprise production.

Description

Technical Field

[0001] The present invention discloses a customer satisfaction analysis method and system based on big data and relates to the field of data analysis. Background Art

[0002] The existing methods or systems for customer satisfaction analysis have the following deficiencies: Data errors and anomalies: There may be errors and outliers in customer satisfaction data, such as incorrect filling, malicious comments, etc.; these errors and outliers will interfere with and affect the data analysis results, leading to wrong decisions and conclusions, and require data processing methods that extract information from multiple dimensions.

[0003] High technical threshold: Existing customer satisfaction analysis systems involve complex technologies and algorithms, such as data mining, machine learning, natural language processing, etc. These technologies require professional technicians to develop and maintain, and place high demands on the company's technical capabilities.

[0004] High application cost: Building a customer satisfaction analysis system requires a large amount of hardware equipment, such as servers, storage devices, network equipment, etc. The procurement, installation and maintenance costs of these hardware equipment are high; at the same time, in the process of customer satisfaction analysis, companies need a large number of data statistics talents, and the labor cost is high. Summary of the invention

[0005] In view of the shortcomings of the prior art, the purpose of the present invention is to provide a customer satisfaction analysis method and system based on big data, aiming to solve the problem of low efficiency of customer satisfaction analysis.

[0006] In order to achieve the above object, the present invention is implemented through the following technical solutions: The customer satisfaction analysis method based on big data includes: Step S1: Obtain the number of products and the number of orders for each product as order data; Step S2: Calculate the sales coefficient of each product based on the order data; collect questionnaire information from different customers on each product; analyze customer satisfaction with each product based on the questionnaire information and the sales coefficient, and determine popular products and unpopular products; Step S3: Obtain product types of popular products as quasi-types; obtain order data of quasi-type products sold by different companies, use SARIMA regression to analyze the differentiation of order numbers of popular products in different seasons, and formulate sales plans for popular products; Step S4: Summarize unpopular products and sales plans and provide feedback.

[0007] Furthermore, the specific steps of step S2 are as follows: Step S21: record the total number of products as pr, and the total number of days in the past three months as td; The number of orders for the first product from the 1st to the tdth day is denoted as og （1,1） ~og （td,1） ; Similarly, the number of orders for the prth product from the 1st to the tdth day is recorded as og （1,pr） ~og （td,pr） ; Step S22: Calculate og （1,1） ~og （td,1） The average value of aog （1） As the average number of orders for product 1; Similarly, og （1,pr） ~og （td,pr） The average value of aog （pr） As the average number of orders for the prth product; Define an empty matrix (td×pr) and fill it with og （1,1） ~og （td,pr） , we get the matrix BO: ; According to the matrix BO, the sales coefficients of the first to pr products are calculated and obtained as bu （1） ~bu （pr） ; Step S23: define scoring indicators: product quality, product price, after-sales service, and delivery time; Count the product quality ratings of the 1st to prth products by the 1st to usth customers, recorded as: ra （1,1） ~ra （us,pr） ; Product price rating, denoted as: rb （1,1） ~rb （us,pr） ; After-sales service rating, denoted as: rc （1,1） ~rc （us,pr） ; Arrival time score, recorded as: rd （1,1） ~rd （us,pr） ; us represents the number of customers who participated in the questionnaire survey; Step S24: Extract the product quality ratings of the first to the usth customers for the first product. （1,1） ~ra （us,1） , product price rating rb （1,1） ~rb （us,1） , after-sales service rating rc （1,1） ~rc （us,1） , arrival time score rd （1,1） ~rd （us,1） ; Calculate the acceptance coefficient ac of the first product （1） .

[0008] Furthermore, the step S2 further includes: Step S25: Repeat the calculation of ac （1） The same steps are used to calculate the recognition coefficients of the second to pr products, and we get ac （2） ~ac （pr） ; Calculate ac （1） ~ac （pr） The average value aac; Get the sales coefficients bu for the 1st to prth products （1） ~bu （pr） ; Define the relationship A-8-1: ;bu （m） and （n） , respectively represent the sales coefficients of the mth and nth products; Mark the products that satisfy relation A-8-1 as high-selling products; Step S26: Obtain the recognition coefficient gac of high-selling products （1） ~gac （rs） ;rs represents the number of high-selling products; Define the relationship A-8-2: gac （w） ≥aac; gac （w） represents the recognition coefficient of the wth high-selling product; Count the high-selling products that satisfy the relationship A-8-2 as best-selling products; Get the recognition coefficient eac of best-selling products （1） ~eac （re） ;re represents the number of best-selling products; Step S27: Obtain the sales coefficient bl of the remaining products （1） ~bl （pi） , recognition coefficient oac （1） ~oac （pi） ; pi represents the amount of remaining products; Extract bl （1） ~bl （pi） The maximum value blm and the minimum value bll in eac （1） ~eac （re） The maximum value is eam, and the minimum value is eal; Define calculation formula A-8-3: ; cr （f） ,bl （f） and oac （f） , respectively represent the satisfaction coefficient, sales coefficient and recognition coefficient of the f-th remaining product; Calculate the satisfaction coefficient cr of the remaining products from the 1st to the pith （1） ~cr （pi） ; Products with a satisfaction coefficient less than 0 are extracted as unpopular products, and products with a satisfaction coefficient greater than 1 are extracted as popular products.

[0009] Furthermore, the specific steps of step S22 are as follows: Step S221: The average number of orders for the i-th product is recorded as aog （i） , the number of orders for the i-th product on the j-th day is denoted as og （j，i） ; Define calculation formula A-1: ;ag （j，i） Indicates og （j，i） A normalized value of ; Calculate the normalized values ​​of all elements in the matrix BO and get the matrix BO （1） : ; Step S222: Calculate the matrix BO （1） The average value of all elements in the first column Cag （1） ; The average value of all elements in the second column Cag （2） Similarly, the average value of all elements in the prth column is Cag （pr） ; Define calculation formula A-2: SC （i） Represents the matrix BO （1） The standard deviation of all elements in the i-th column; Calculate the standard deviation SC of all elements in column 1 （1） , the standard deviation SC of all elements in the second column （2） ; Similarly, the standard deviation SC of all elements in the prth column （pr） ; Step S223: Define calculation formula A-3: ; bg （j，i） Indicates ag （j，i） The secondary normalized value of ; Calculate the matrix BO （1） The quadratic normalization of all elements in gives the matrix BO （2） : ; Define calculation formula A-4-1: ; Define calculation formula A-4-2: ; R （j） Represents the matrix BO （2） The average value of all elements in the jth row, Cbg （i） Represents the matrix BO （2） The average value of all elements in the i-th column; Calculate the matrix BO （2） The average value Rbg of all elements in the 1st to tdth rows （1） ~Rbg （td） , the average value of all elements from the 1st to the prth column Cbg （1） ~Cbg （pr） .

[0010] Furthermore, the subsequent steps of step S223 are as follows: Step S224: Define calculation formula A-5: ; Among them, cg （j，i） Indicates bg （j，i） The three normalized values ​​of Calculate the matrix BO （2） The triple normalized values ​​of all elements in get the matrix BO （3） : ; The matrix BO （3） Split by column to get the matrix CBO （1） 、CBO （2） ~Matrix CBO （pr） ; Matrix CBO （1） for: ; Matrix CBO （2） for: ; By analogy, the matrix CBO （pr） for: ; Step S225: define matrix CBO （1） 、CBO （2） ~Matrix CBO （pr） The parameter coefficient λ （1） , （2） ~λ （pr） and the offset coefficient μ （1） , μ （2） ~μ （pr） ; Use Intel MKL library functions to calculate λ （1） ~λ （pr） and μ （1） ~μ （pr） , so that the matrix CBO （1） ~Matrix CBO （pr）The following mathematical relationship is satisfied: ; Among them, II （td） represents an all-one matrix (td×1); Step S226: Calculate the sales coefficient bu of the first product （1） , （1） =λ （1） / μ （1） ; The sales coefficient of the second product is bu （2） , （2） =λ （2） / μ （2） ; By analogy, the sales coefficient of the prth product is bu （pr） , （pr） =λ （pr） / μ （pr） .

[0011] Furthermore, the specific steps of step S24 are as follows: Step S241: Calculate the total score of the first customer （1） Similarly, the total score of the us-th customer is （us） ; Calculate ara （1） ~ara （us） of and Ara; Step S242: According to ra （1,1） ~ra （us,1） , calculate the weighted coefficient λa of the product quality ratings of the 1st to the usth customers （1） ~λa （us） and the offset coefficient va; Step S243: Repeat the calculation of λa （1） ~λa （us） The same steps as va, calculate the weighted coefficient λb of the product price rating of the 1st to the usth customers （1） ~λb （us） and offset coefficient vb; Calculate the weighted coefficient λc for after-sales service rating （1） ~λc （us） and offset coefficient vc; Calculate the weighted coefficient λd for the score of arrival time （1） ~λd （us） and the offset coefficient vd; Step S244: corresponding to the empty matrix (us×4), and fill in ra in sequence （1,1） ~ra （us,1） , rb （1,1） ~rb （us,1） , rc（1,1） ~rc （us,1） andrd （1,1） ~rd （us,1） , we get the matrix CO: ; Split the matrix CO by columns to obtain the matrix CA, matrix CB, matrix CC and matrix CD; Matrix CA: ; Matrix CB: ; Matrix CC: ; Matrix CD: .

[0012] Furthermore, the subsequent steps of step S244 are as follows: Step S245: Calculate the path coefficient β of the product quality score corresponding to the first product （ra） : ; Where T represents the transpose of the matrix, × represents matrix multiplication, and -1 represents the inverse of the matrix; The path coefficient of product price rating β （rb） : ; Path coefficient β of after-sales service rating （rc） : ; Path coefficient of arrival time score β （rd） : ; Step S246: Calculate the recognition coefficient ac of the first product （1） ; Define calculation formula A-7: ; Among them, λa （t） , λb （t） ,λc （t） and λd （t） , respectively represent the weighted coefficients of the t-th customer's rating of product quality, product price, after-sales service, and delivery time.

[0013] Furthermore, the specific steps of step S242 are as follows: Step S2421: Set λa after k updates （1） ~λa （us） The value of ar （1） （k） ~ar （us） （k） ; Calculate ra （1,1） The value raa after k updates （1,1） （k）,raa （1,1） （k） =ra （1,1） ×ar （1） （k） ; Similarly, （us,1） The value raa after k updates （us,1） （k） ,raa （us,1） （k） =ra （us,1） ×ar （us） （k） ; Step S2422: Calculate ra （1,1） ~ra （us,1） The k-times weighted sum of （k） : ; Among them, ra （t,1） represents the calculation of the product quality score of the t-th customer for the first product, ar （t） （k） It represents the value of the weighted coefficient of the product quality score of the t-th customer after k updates; the value range of t is: 1~us; Step S2423: Calculate ra （1,1） ~ra （us,1） The corresponding k-times updated mean is denoted as rajj （k） , the variance is recorded as ss （ra） （k） ; Calculate ra （1,1） ~ra （us,1） The mean of is denoted as raj; based on raj and rajj, calculate ra （1,1） ~ra （us,1） About raa （1,1） ~raa （us,1） The k-th covariance of （ra） （k） ; Define calculation formula A-6-1: ; Among them, raa （t,1） （k） Indicates ra （t,1） k-times updated value, Aλ （k） Represents the combined weight after k updates.

[0014] Furthermore, the subsequent steps of step S2423 are as follows: Step S2424: ar （1） ~ar （us） The initial value of is set to 0.5; Set the value of k to 1, and（k） Update until relation A-6-2 is satisfied; The relationship A-6-2 is: ; Among them, Aλ （k+1） represents the combined weight after (k+1) updates; ε represents the error tolerance; The combined weight satisfying the relation A-6-2 is denoted as Bλ; Step S2425: Weighting coefficient λa of the first customer's product quality rating （1） for: ; The weighted coefficient λa of the second customer's product quality rating （2） for: ; Similarly, the weighted coefficient λa of the product quality score of the usth customer is （us） for: ; Step S2426: Calculate the offset coefficient va, and define the calculation formula A-6-3: ; λa （t） Represents the weighted coefficient of the product quality score of the t-th customer.

[0015] The customer satisfaction analysis system based on big data includes: Order acquisition module: used to obtain the number of products and the number of orders for each product as order data; Data analysis module: satisfaction analysis submodule and sales analysis submodule; Satisfaction analysis submodule: used to calculate the sales coefficient of each product based on order data; collect questionnaire information from different customers on each product; analyze customer satisfaction with each product based on questionnaire information and sales coefficient, and determine popular products and unpopular products; Sales analysis submodule: used to obtain the product types of popular products as quasi-types; obtain order data of quasi-type products sold by different companies, use SARIMA regression to analyze the differentiation of order numbers of popular products in different seasons, and formulate sales plans for popular products; User interaction module: used to summarize unpopular products and sales plans and provide feedback.

[0016] Compared with the prior art, the present invention has the following beneficial effects: Accurately identify the root cause of the problem: By analyzing the relevant data of sensory orders and questionnaires, the present invention can help enterprises to discover the key factors that affect customer satisfaction, such as product quality problems, poor service processes, unreasonable prices, etc.; this precise problem location helps enterprises to formulate targeted improvement measures, thereby improving the scientificity and accuracy of decision-making.

[0017] Predicting customer behavior trends: The big data analysis of the present invention not only focuses on current data, but also predicts future customer behavior trends through the analysis of historical data, enabling companies to plan ahead, meet future customer needs, and improve customer satisfaction and loyalty.

[0018] Continuously improve service processes: Through continuous analysis of customer satisfaction data, the present invention can help enterprises identify bottlenecks and inefficient links in the service process and optimize them in a targeted manner. This continuously improved service process can enhance the overall customer experience and strengthen the competitiveness of enterprises. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Other features, objects and advantages of the present invention will become more apparent from the detailed description of non-limiting embodiments made with reference to the following drawings: Figure 1 It is a schematic diagram of the method of the present invention; Figure 2 It is a schematic diagram of the system of the present invention; Figure 3 Schematic diagram of the questionnaire survey of the present invention. DETAILED DESCRIPTION

[0020] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0021] Embodiment 1: See also Figure 1 and Figure 3 , customer satisfaction analysis methods based on big data include: Step S1: Obtain the number of products (provided by the enterprise) and the number of orders for each product (in the past three months) as order data; Step S2: Calculate the sales coefficient of each product based on the order data; (distribute online or offline questionnaires for each product to customers) collect questionnaire information from different customers on each product; analyze customer satisfaction with each product based on the questionnaire information and sales coefficient, and determine popular products and unpopular products; The specific steps of step S2 are as follows: Step S21: record the total number of products as pr; record the number of orders for each product (in the past three months) as order data; record the total number of days in the past three months as td; The number of orders for the first product on the first, second, and last td days is denoted as og （1,1） og （2,1） ~og （td,1） ; The number of orders for the second product on the first, second, and last day is og （1,2） og （2,2） ~og （td,2） ; Similarly, the number of orders for the prth product on the 1st, 2nd, and tdth days is recorded as og （1,pr） og （2,pr） ~og （td,pr） ; It should be noted that the "order" in the present invention means a "valid order", that is, an order that completes a complete transaction; if an order is withdrawn or interrupted by the consumer during the transaction process, the order will not be counted as a "valid order".

[0022] Step S22: Calculate og （1,1） ~og （td,1） The average value of aog （1） As the average number of orders for the first product; （1,2） ~og （td,2） The average value of aog （1） as the average number of orders for the second product; and so on, （1,pr） ~og （td,pr） The average value of aog （pr） As the average number of orders for the prth product; Define an empty matrix (td×pr) and fill it with og （1,1） ~og （td,pr） , we get the matrix BO: ; According to the matrix BO, the sales coefficients of the first to pr products are calculated and obtained as bu （1） ~bu （pr） ; Step S221: The average number of orders for the i-th product is recorded as aog （i） , the number of orders for the i-th product on the j-th day is denoted as og （j，i） ; The value range of i is: 1 ~ pr, the value range of j is: 1 ~ td; Define calculation formula A-1: Among them, ag （j，i） Indicates og （j，i） A normalized value of ; (According to formula A-1) calculate the normalized values ​​of all elements in the matrix BO and obtain the matrix BO（1） : ; Step S222: Calculate the matrix BO （1） All elements in the first column (i.e., ag （1,1） ~ag （td,1） ), denoted as Cag （1） ; All elements in the second column (i.e. ag （1,2） ~ag （td,2） ), denoted as Cag （2） ; And so on, all the elements in the prth column (i.e. ag （1,pr） ~ag （td,pr） ), denoted as Cag （pr） ; Define calculation formula A-2: ; Among them, SC （i） Represents the matrix BO （1） The standard deviation of all elements in the i-th column; (According to formula A-2) calculate all elements in column 1 (i.e. ag （1,1） ~ag （td,1） ) standard deviation SC （1） , all elements in the second column (i.e. ag （1,2） ~ag （td,2） ) standard deviation SC （2） ; And so on, all the elements in the prth column (i.e. ag （1,pr） ~ag （td,pr） ) standard deviation SC （pr） ; Step S223: Define calculation formula A-3: ; Among them, bg （j，i） Indicates ag （j，i） The secondary normalized value of ; (According to formula A-3) Calculate the matrix BO （1） The quadratic normalization of all elements in gives the matrix BO （2） : ; Define calculation formula A-4-1: ; Define calculation formula A-4-2: ; Among them, Rbg （j） Represents the matrix BO （2） The average value of all elements in the jth row, Cbg （i） Represents the matrix BO （2） The average value of all elements in the i-th column; (According to calculation formula A-4-1 and calculation formula A-4-1), calculate the matrix BO （2） The average value Rbg of all elements in the 1st to tdth rows （1） ~Rbg （td） , the average value of all elements from the 1st to the prth column Cbg （1） ~Cbg （pr） ; Step S224: Define calculation formula A-5: ; Among them, cg （j，i） Indicates bg （j，i） The three normalized values ​​of (According to formula A-5) Calculate the matrix BO （2） The triple normalized values ​​of all elements in get the matrix BO （3） : ; The matrix BO （3） Split by column to get the matrix CBO （1） 、CBO （2） ~Matrix CBO （pr） ; Matrix CBO （1） for: ; Matrix CBO （2） for: ; By analogy, the matrix CBO （pr） for: ; Step S225: define matrix CBO （1） 、CBO （2） ~Matrix CBO （pr） The parameter coefficient λ （1） , （2） ~λ （pr） and the offset coefficient μ （1） , μ （2） ~μ （pr） ; Use Intel MKL library functions to calculate λ （1） ~λ （pr） and μ （1） ~μ （pr） , so that the matrix CBO （1） ~Matrix CBO （pr） The following mathematical relationship is satisfied: ; Among them, II （td） represents an all-one matrix (td×1); Step S226: Calculate the sales coefficient bu of the first product （1） ,（1） =λ （1） / μ （1） ; The sales coefficient of the second product is bu （2） , （2） =λ （2） / μ （2） ; By analogy, the sales coefficient of the prth product is bu （pr） , （pr） =λ （pr） / μ （pr） ; Step S23: Define scoring indicators: product quality, product price, after-sales service, and delivery time; (the upper limit of the score is 10 points) Send a questionnaire (online or offline) to us customers (for products 1 to pr); us represents the number of customers who participated in the questionnaire survey; Count the product quality ratings of the 1st to prth products by the 1st to usth customers, recorded as: ra （1,1） ~ra （us,pr） ; Product price rating, denoted as: rb （1,1） ~rb （us,pr） ; After-sales service rating, denoted as: rc （1,1） ~rc （us,pr） ; Arrival time score, recorded as: rd （1,1） ~rd （us,pr） ; Step S24: Extract the product quality ratings of the first, second, and up to the usth customers for the first product. （1,1） ,ra （2,1） ~ra （us,1） , product price rating rb （1,1） ,rb （2,1） ~rb （us,1） , after-sales service rating rc （1,1） rc （2,1） ~rc （us,1） , arrival time score rd （1,1） ,rd （2,1） ~rd （us,1） ; Calculate the acceptance coefficient ac of the first product （1） ; Step S241: Calculate the total score of the first customer (i.e., （1,1） ,rb （1,1） rc （1,1） and （1,1） and), written as ara （1） ; The total score of the second customer (i.e. （2,1） ,rb（2,1） rc （2,1） and （2,1） and), written as ara （2） ; Similarly, the total score of the usth customer (i.e. （us,1） ,rb （us,1） rc （us,1） and （us,1） and), written as ara （us） ; Calculate ara （1） ~ara （us） The sum of is denoted as Ara; Step S242: According to ra （1,1） ~ra （us,1） , calculate the weighted coefficient λa of the product quality rating of the first to the usth customers (of the first product) （1） ~λa （us） and the offset coefficient va; Step S2421: Set λa after k updates （1） ~λa （us） The value of ar （1） （k） ~ar （us） （k） ; (k is a natural number) Calculate ra （1,1） The value raa after k updates （1,1） （k） ,raa （1,1） （k） =ra （1,1） ×ar （1） （k） ; ra （2,1） The value raa after k updates （2,1） （k） ,raa （2,1） （k） =ra （2,1） ×ar （2） （k） ; By analogy, （us,1） The value raa after k updates （us,1） （k） ,raa （us,1） （k） =ra （us,1） ×ar （us） （k） ; Step S2422: (based on ar （1） （k） ~ar （us） （k）) Calculate ra （1,1） ~ra （us,1） The k-times weighted sum of （k） : ; Among them, ra （t,1） represents the calculation of the product quality score of the t-th customer for the first product, ar （t） （k） It represents the value of the weighted coefficient of the product quality score of the t-th customer (for the first product) after k updates; the value range of t is: 1~us; Step S2423: Calculate ra （1,1） ~ra （us,1） (ie raa （1,1） （k） ~raa （us,1） （k） ) The corresponding k-times updated mean is denoted as rajj （k） , the variance is recorded as ss （ra） （k） ; Calculate ra （1,1） ~ra （us,1） The mean of is denoted as raj; based on raj and rajj, calculate ra （1,1） ~ra （us,1） About raa （1,1） ~raa （us,1） The k-th covariance of （ra） （k） ; Define calculation formula A-6-1: ; Among them, raa （t,1） （k） Indicates ra （t,1） k-times updated value, Aλ （k） Represents the combined weight after k updates; Step S2424: ar （1） ~ar （us） The initial value is set to 0.5; (customers can adjust the initial value according to actual needs) Set the value of k to 1, and (according to the rules corresponding to steps S2421 to S2423) （k） Update until relation A-6-2 is satisfied; The relationship A-6-2 is: ; Among them, Aλ （k+1） represents the combined weight after (k+1) updates; ε represents the allowable error value; (the value of ε is 0.01, and customers or relevant technical personnel can adjust the value of ε according to actual needs) The combined weight satisfying the relation A-6-2 is denoted as Bλ; Step S2425: Weighting coefficient λa of the first customer's product quality rating (for the first product) （1） for: ; The weighting coefficient λa of the second customer's product quality rating (for the first product) （2） for: ; Similarly, the weighted coefficient λa of the product quality score of the usth customer (for the first product) is （us） for: ; Step S2426: Define calculation formula A-6-3: ; Among them, λa （t） represents the weighted coefficient of the product quality score of the t-th customer (for the first product); (According to calculation formula A-6-3) calculate the offset coefficient va; Step S243: Repeat the calculation of λa （1） ~λa （us） The same steps as va, calculate the weighted coefficient λb of the product price rating of the first to the usth customers (of the first product) （1） ~λb （us） and offset coefficient vb; Weighting coefficient λc for after-sales service rating (for product 1) （1） ~λc （us） and offset coefficient vc; Weighting coefficient λd for the score of arrival time (for product 1) （1） ~λd （us） and the offset coefficient vd; Step S244: corresponding to the empty matrix (us×4), and fill in ra in sequence （1,1） ~ra （us,1） , rb （1,1） ~rb （us,1） , rc （1,1） ~rc （us,1） andrd （1,1） ~rd （us,1） , we get the matrix CO: ; Split the matrix CO by columns to obtain the matrix CA, matrix CB, matrix CC and matrix CD; Matrix CA: ; Matrix CB: ; Matrix CC: ; Matrix CD: ; Step S245: Calculate the path coefficient β of the product quality score corresponding to the first product (the overall score of the first product) （ra） : ; Where T represents the transpose of the matrix, × represents matrix multiplication, and -1 represents the inverse of the matrix; The path coefficient of product price rating (overall rating of product 1) β （rb） : ; Path coefficient β of after-sales service rating (overall rating of the first product) （rc） : ; Path coefficient β of the delivery time score (overall score of the first product) （rd） : ; Step S246: Calculate the recognition coefficient ac of the first product (according to formula A-7) （1） ; Define calculation formula A-7: ; Among them, λa （t） , λb （t） ,λc （t） and λd （t） , respectively represent the weighted coefficients of the t-th customer’s rating of product quality, product price, after-sales service, and delivery time (for the first product); Step S25: Repeat the calculation of ac （1） The same steps are used to calculate the recognition coefficients of the second to pr products, and we get ac （2） ~ac （pr） ; Calculate ac （1） ~ac （pr） The average value of is denoted as aac; Get the sales coefficients bu for the 1st to prth products （1） ~bu （pr） ; Define the relationship A-8-1: ; Among them, bu （m） and （n） , respectively represent the sales coefficients of the mth and nth products; m and n∈{1~pr} and m≠n; (According to bu （1） ~bu （pr） ) Mark the products that satisfy the relationship A-8-1 as high-selling products; Step S26: Obtain the recognition coefficient of high-selling products, denoted as gac （1） ~gac （rs） ; Among them, rs represents the number of high-selling products; Define the relationship A-8-2: gac （w） ≥aac; where gac （w） represents the recognition coefficient of the wth high-selling product; the value range of w is: 1~rs; (According to GAC （1） ~gac （rs） ) Count the high-selling products that satisfy the relationship A-8-2 as best-selling products; Get the recognition coefficient of the best-selling product, denoted as eac （1） ~eac （re） ; Among them, re represents the number of best-selling products; Step S27: (Among the 1st to prth products, remove the best-selling products to obtain the remaining products) (In bu （1） ~bu （pr） Neutralize ac （1） ~ac （pr） (middle) Eliminate the sales coefficient and recognition coefficient corresponding to popular high-selling products, and get bl （1） ~bl （pi） and oac （1） ~oac （pi） ; Among them, pi represents the number of remaining products (after removing the best-selling products), pi = pr-re; Extract bl （1） ~bl （pi） The maximum value is denoted as blm, and the minimum value is denoted as bll; Extract eac （1） ~eac （re） The maximum value is recorded as eam, and the minimum value is recorded as eal; Define calculation formula A-8-3: ; Among them, cr （f） ,bl （f） and oac （f） , respectively represent the satisfaction coefficient, sales coefficient and recognition coefficient of the fth remaining product; the value range of f is: 1~pi; (According to calculation formula A-8-3) calculate the satisfaction coefficient of the remaining products from the 1st to the pith type, and get cr （1） ~cr （pi） ; Among the remaining products from the 1st to the pith type, products with a satisfaction coefficient less than 0 are selected as unpopular products, and products with a satisfaction coefficient close to or greater than 1 are selected as popular products (a satisfaction coefficient close to 1 means a satisfaction coefficient ≥ 0.97) Make popular products and best-sellers popular products; Step S3: Obtain product types of popular products as quasi-types; obtain order data of quasi-type products sold by different companies (in the past three years), use SARIMA regression to analyze the differentiation of order numbers of popular products in different seasons, and formulate sales plans for popular products; It should be noted that the “quasi-type” in the present invention refers to the type of goods; For example, if company A is an enterprise using the present invention, and if the popular product of company A is a ** brand smartphone, then the quasi-type is "smartphone"; the "order data" acquired in step S3 is: the order data (or sales data) of smartphones of other companies (not companies A) on the market in the past three years; Step S4: Summarize the sales plans of unpopular products and (popular products) and provide feedback.

[0023] It should be noted that the "product quality, product price, after-sales service, and delivery time" in the present invention are only used as examples for demonstration, and customers or relevant technical personnel can define or adjust more product rating indicators according to actual needs.

[0024] Actual example 2: See also Figure 2 , the customer satisfaction analysis system based on big data includes: Order acquisition module: used to obtain the number of products (provided by the enterprise) and the number of orders for each product (in the past three months) as order data; Data analysis module: satisfaction analysis submodule and sales analysis submodule; Satisfaction analysis submodule: used to calculate the sales coefficient of each product based on order data; (distribute online or offline questionnaires for each product to customers) collect questionnaire information from different customers on each product; analyze customer satisfaction with each product based on questionnaire information and sales coefficient, and determine popular and unpopular products; Sales analysis submodule: used to obtain the product types of popular products as quasi-types; obtain the order data of quasi-type products sold by different companies (in the past three years), use SARIMA regression to analyze the differentiation of the number of orders for popular products in different seasons, and formulate sales plans for popular products; User interaction module: used to summarize and provide feedback on sales plans for unpopular products and (popular products).

[0025] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technical personnel in this field according to actual conditions. For example, if there are weight coefficients and proportional coefficients, their set sizes are to quantify each parameter to obtain a specific value, which is convenient for subsequent comparison. Regarding the size of the weight coefficient and the proportional coefficient, it is sufficient as long as it does not affect the proportional relationship between the parameter and the quantized value.

[0026] Finally, it should be noted that the above-described embodiments are only specific implementations of the present invention, which are used to illustrate the technical solutions of the present invention, rather than to limit them. The protection scope of the present invention is not limited thereto. Although the present invention is described in detail with reference to the above-described embodiments, ordinary technicians in the field should understand that any technician familiar with the technical field can still modify the technical solutions recorded in the above-described embodiments within the technical scope disclosed by the present invention, or can easily think of changes, or make equivalent replacements for some of the technical features therein; and these modifications, changes or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A customer satisfaction analysis method based on big data, characterized in that: The method comprises: Step S1: Obtain the number of products and the number of orders for each product as order data; Step S2: Calculate the sales coefficient of each product based on the order data; collect questionnaire information from different customers on each product; analyze customer satisfaction with each product based on the questionnaire information and the sales coefficient, and determine popular products and unpopular products; Step S3: Obtain product types of popular products as quasi-types; obtain order data of quasi-type products sold by different companies, use SARIMA regression to analyze the differentiation of order numbers of popular products in different seasons, and formulate sales plans for popular products; Step S4: Summarize unpopular products and sales plans and provide feedback.

2. The customer satisfaction analysis method based on big data according to claim 1 is characterized in that: The specific steps of step S2 are as follows: Step S21: record the total number of products as pr, and the total number of days in the past three months as td; The number of orders for the first product from the 1st to the tdth day is denoted as og （1,1） ~og （td,1） ; Similarly, the number of orders for the prth product from the 1st to the tdth day is recorded as og （1,pr） ~og （td,pr） ; Step S22: Calculate og （1,1） ~og （td,1） The average value of aog （1） As the average number of orders for product 1; Similarly, og （1,pr） ~og （td,pr） The average value of aog （pr） As the average number of orders for the prth product; Define an empty matrix (td×pr) and fill it with og （1,1） ~og （td,pr） , we get the matrix BO: ； According to the matrix BO, the sales coefficients of the first to pr products are calculated and obtained as bu （1） ~bu （pr） ; Step S23: define scoring indicators: product quality, product price, after-sales service, and delivery time; Count the product quality ratings of the 1st to prth products by the 1st to usth customers, recorded as: ra （1,1） ~ra （us,pr） ; Product price rating, denoted as: rb （1,1） ~rb （us,pr） ; After-sales service rating, denoted as: rc （1,1） ~rc （us,pr） ; Arrival time score, recorded as: rd （1,1） ~rd （us,pr） ; us represents the number of customers who participated in the questionnaire survey; Step S24: Extract the product quality ratings of the first to the usth customers for the first product. （1,1） ~ra （us,1） , product price rating rb （1,1） ~rb （us,1） , after-sales service rating rc （1,1） ~rc （us,1） , arrival time score rd （1,1） ~rd （us,1） ; Calculate the acceptance coefficient ac of the first product （1） .

3. The customer satisfaction analysis method based on big data according to claim 2 is characterized in that: The step S2 further comprises: Step S25: Repeat the calculation of ac （1） The same steps are used to calculate the recognition coefficients of the second to pr products, and we get ac （2） ~ac （pr） ; Calculate ac （1） ~ac （pr） The average value aac; Get the sales coefficients bu for the 1st to prth products （1） ~bu （pr） ; Define the relationship A-8-1: ;bu （m） and （n） , respectively represent the sales coefficients of the mth and nth products; Mark the products that satisfy relation A-8-1 as high-selling products; Step S26: Obtain the recognition coefficient gac of high-selling products （1） ~gac （rs） ;rs represents the number of high-selling products; Define the relationship A-8-2: gac （w） ≥aac; gac （w） represents the recognition coefficient of the wth high-selling product; Count the high-selling products that satisfy the relationship A-8-2 as best-selling products; Get the recognition coefficient eac of best-selling products （1） ~eac （re） ;re represents the number of best-selling products; Step S27: Obtain the sales coefficient bl of the remaining products （1） ~bl （pi） , recognition coefficient oac （1） ~oac （pi） ; pi represents the amount of remaining products; Extract bl （1） ~bl （pi） The maximum value blm and the minimum value bll in eac （1） ~eac （re） The maximum value is eam, and the minimum value is eal; Define calculation formula A-8-3: ； cr （f） ,bl （f） and oac （f） , respectively represent the satisfaction coefficient, sales coefficient and recognition coefficient of the f-th remaining product; Calculate the satisfaction coefficient cr of the remaining products from the 1st to the pith （1） ~cr （pi） ; Products with a satisfaction coefficient less than 0 are extracted as unpopular products, and products with a satisfaction coefficient greater than 1 are extracted as popular products.

4. The customer satisfaction analysis method based on big data according to claim 2 is characterized in that: The specific steps of step S22 are as follows: Step S221: The average number of orders for the i-th product is recorded as aog （i） , the number of orders for the i-th product on the j-th day is denoted as og （j，i） ; Define calculation formula A-1: ;ag （j，i） Indicates og （j，i） A normalized value of ; Calculate the normalized values ​​of all elements in the matrix BO and get the matrix BO （1） : ； Step S222: Calculate the matrix BO （1） The average value of all elements in the first column Cag （1） Similarly, the average value of all elements in the prth column is Cag （pr） ; Define calculation formula A-2: SC （i） Represents the matrix BO （1） The standard deviation of all elements in the i-th column; Calculate the standard deviation SC of all elements in column 1 （1） The standard deviation SC of all elements up to the prth column （pr） ; Step S223: Define calculation formula A-3: ;bg （j，i） Indicates ag （j，i） The secondary normalized value of ; Calculate the matrix BO （1） The quadratic normalization of all elements in gives the matrix BO （2） : ； Define calculation formula A-4-1: ; Define calculation formula A-4-2: ; R （j） Represents the matrix BO （2） The average value of all elements in the jth row, Cbg （i） Represents the matrix BO （2） The average value of all elements in the i-th column; Calculate the matrix BO （2） The average value Rbg of all elements in the 1st to tdth rows （1） ~Rbg （td） , the average value of all elements from the 1st to the prth column Cbg （1） ~Cbg （pr） .

5. The customer satisfaction analysis method based on big data according to claim 4 is characterized in that: The subsequent steps of step S223 are as follows: Step S224: Define calculation formula A-5: ; Among them, cg （j，i） Indicates bg （j，i） The three normalized values ​​of Calculate the matrix BO （2） The triple normalized values ​​of all elements in get the matrix BO （3） : ； The matrix BO （3） Split by column to get the matrix CBO （1） ~Matrix CBO （pr） ; Matrix CBO （1） for: ; Similarly, the matrix CBO （pr） for: ; Step S225: define matrix CBO （1） ~Matrix CBO （pr） The parameter coefficient λ （1） ~λ （pr） and the offset coefficient μ （1） ~μ （pr） ; Use Intel MKL library functions to calculate λ （1） ~λ （pr） and μ （1） ~μ （pr） , so that the matrix CBO （1） ~Matrix CBO （pr） The following mathematical relationship is satisfied: ； Among them, II （td） represents an all-one matrix (td×1); Step S226: Calculate the sales coefficient bu of the first product （1） , （1） =λ （1） / μ （1） ; The sales coefficient of the second product is bu （2） , （2） =λ （2） / μ （2） ; By analogy, the sales coefficient of the prth product is bu （pr） , （pr） =λ （pr） / μ （pr） .

6. The customer satisfaction analysis method based on big data according to claim 2 is characterized in that: The specific steps of step S24 are as follows: Step S241: Calculate the total score of the first customer （1） Similarly, the total score of the us-th customer is （us） ; Calculate ara （1） ~ara （us） of and Ara; Step S242: According to ra （1,1） ~ra （us,1） , calculate the weighted coefficient λa of the product quality ratings of the 1st to the usth customers （1） ~λa （us） and the offset coefficient va; Step S243: Repeat the calculation of λa （1） ~λa （us） The same steps as va, calculate the weighted coefficient λb of the product price ratings of the 1st to the usth customers （1） ~λb （us） and the offset coefficient vb; calculate the weighting coefficient λc for the after-sales service score （1） ~λc （us） and the offset coefficient vc; calculate the weighted coefficient λd for the arrival time score （1） ~λd （us） and the offset coefficient vd; Step S244: Fill in the empty matrix (us×4) with ra in sequence （1,1） ~ra （us,1） , rb （1,1） ~rb （us,1） , rc （1,1） ~rc （us,1） andrd （1,1） ~rd （us,1） , we get the matrix CO: ； Split the matrix CO by columns to obtain the matrix CA, matrix CB, matrix CC and matrix CD; Matrix CA: ; Matrix CB: ; Matrix CC: ; Matrix CD: .

7. The customer satisfaction analysis method based on big data according to claim 6 is characterized in that: The subsequent steps of step S244 are as follows: Step S245: Calculate the path coefficient β of the product quality score corresponding to the first product （ra） : ; Where T represents the transpose of the matrix, × represents matrix multiplication, and -1 represents the inverse of the matrix; The path coefficient of product price rating β （rb） : ; Path coefficient β of after-sales service rating （rc） : ; Path coefficient of arrival time score β （rd） : ; Step S246: Calculate the recognition coefficient ac of the first product （1） ; Define calculation formula A-7: ； Among them, λa （t） , λb （t） ,λc （t） and λd （t） , respectively represent the weighted coefficients of the t-th customer's rating of product quality, product price, after-sales service, and delivery time.

8. The customer satisfaction analysis method based on big data according to claim 6 is characterized in that: The specific steps of step S242 are as follows: Step S2421: Set λa after k updates （1） ~λa （us） The value of ar （1） （k） ~ar （us） （k） ; Calculate ra （1,1） The value raa after k updates （1,1） （k） =ra （1,1） × （1） （k） ; Similarly, （us,1） The value raa after k updates （us,1） （k） =ra （us,1） × （us） （k） ; Step S2422: Calculate ra （1,1） ~ra （us,1） The k-th weighted sum SQ （k） : ;ra （t,1） represents the calculation of the product quality score of the t-th customer for the first product, ar （t） （k） represents the value of the weighted coefficient of the product quality score of the t-th customer after k updates; Step S2423: Calculate ra （1,1） ~ra （us,1） The corresponding k-times updated mean is denoted as rajj （k） , the variance is recorded as ss （ra） （k） ; Calculate ra （1,1） ~ra （us,1） The mean of is denoted as raj; based on raj and rajj, calculate ra （1,1） ~ra （us,1） About raa （1,1） ~raa （us,1） The k-th covariance of （ra） （k） ; Define calculation formula A-6-1: ; Among them, raa （t,1） （k） Indicates ra （t,1） k-times updated value, Aλ （k） Represents the combined weight after k updates.

9. The customer satisfaction analysis method based on big data according to claim 8 is characterized in that: The subsequent steps of step S2423 are as follows: Step S2424: ar （1） ~ar （us） The initial value of is set to 0.5; Set the value of k to 1, and （k） Update until relation A-6-2 is satisfied; The relationship A-6-2 is: ; Among them, Aλ （k+1） represents the combined weight after (k+1) updates; ε represents the error tolerance; The combined weight satisfying the relation A-6-2 is denoted as Bλ; Step S2425: Weighting coefficient λa of the first customer's product quality rating （1） for: ； Similarly, the weighted coefficient λa of the product quality score of the usth customer is （us） for: ； Step S2426: Calculate the offset coefficient va, and define the calculation formula A-6-3: ; λa （t） Represents the weighted coefficient of the product quality score of the t-th customer.

10. A customer satisfaction analysis system based on big data, applicable to the customer satisfaction analysis method based on big data as claimed in any one of claims 1 to 9, characterized in that: The system comprises: Order acquisition module: used to obtain the number of products and the number of orders for each product as order data; Data analysis module: satisfaction analysis submodule and sales analysis submodule; Satisfaction analysis submodule: used to calculate the sales coefficient of each product based on order data; collect questionnaire information from different customers on each product; analyze customer satisfaction with each product based on questionnaire information and sales coefficient, and determine popular products and unpopular products; Sales analysis submodule: used to obtain the product types of popular products as quasi-types; obtain order data of quasi-type products sold by different companies, use SARIMA regression to analyze the differentiation of order numbers of popular products in different seasons, and formulate sales plans for popular products; User interaction module: used to summarize unpopular products and sales plans and provide feedback.

Images