An event detection association system and method based on multivariate data

Abstract

The present invention discloses an event detection association system and method based on multivariate data, which relates to the field of event detection association technology, including: S1-1, obtaining historical order and order data in a store; S1-2, based on the historical order data of the store, analyzing the influence of customers who have ordered in the store on customers who have not ordered, and training a scoring model for dishes; S1-3, obtaining real-time order and order data in the store; S1-4, using the scoring model of the dishes to calculate the score of each dish, and recommending dishes to customers according to the dish score after the customer logs in to the ordering system. It can save customers' waiting time in line and improve their dining experience; improve the operating efficiency of the restaurant; analyze the influence of the physical objects of the ordered dishes on the customer's order, recommend dishes when the customer orders, and improve the accuracy of the dish recommendation.

Description

Technical Field

[0001] The present invention relates to the technical field of event detection association, and in particular to an event detection association system and method based on multivariate data. Background Art

[0002] Event detection association usually aims to identify and predict events that may affect operational efficiency by integrating and analyzing information from different data sources. This method can be applied to the catering industry to optimize resource allocation, reduce risks and improve decision-making quality. Scanning QR codes to order food in stores is a convenient and fast way to order food. By scanning QR codes or using mobile phone APPs to order food, you can save time waiting in line and improve dining efficiency. The advantages of scanning QR codes to order food include: convenience and time saving; you can order food in advance to avoid waiting in line; customers can learn about a variety of food options and promotions to meet the needs of different customers; you can realize self-service ordering, self-service settlement and other functions to improve the operating efficiency of the restaurant; however, scanning QR codes to order food also has some shortcomings. The existing food recommendation system usually only puts dishes with high sales on the recommendation page, only considering the overall situation, without considering the impact of individuals; for this reason, how to recommend food to customers based on the actual situation in the store has become an urgent problem to be solved. Summary of the invention

[0003] The object of the present invention is to provide an event detection and association system and method based on multivariate data to solve the problems raised in the above background technology.

[0004] In one aspect of the present invention, a method for event detection and association based on multivariate data is provided, comprising:

[0005] S1-1, obtain the historical order and order data in the store;

[0006] S1-2, based on the store’s historical order data, analyze the impact of customers who have ordered in the store on customers who have not ordered, and train a scoring model for dishes;

[0007] S1-3, obtain real-time order and order data in the store;

[0008] S1-4, using the dish scoring model to calculate the score of each dish, and after the customer logs into the ordering system, making dish recommendations for the customer based on the dish score.

[0009] In step S1-2, the score of the dish is calculated using the following formula:

[0010] In the formula, i is a positive integer between [1, a], which is used to represent different dishes, a is the total number of dishes; k is an integer greater than or equal to 0, which represents the number of times the table has been ordered by customers; The score of the i-th dish when the customer orders for the k+1th time at the table; It represents the influence of customers who have ordered food in the store on customers who have not ordered food when the customer orders food at the table for the k+1th time; is the average score of the i-th dish on all tables in the store when the customer orders for the k+1th time at the table; w1 is the inertia weight, which indicates the influence of the i-th dish ordered last time on the current order; w2 is the individual weight, which indicates the weight of the i-th dish score coming from the customers who have ordered; w3 is the group weight, which indicates the weight of the i-th dish score coming from other tables;

[0011] After dish i is ordered by the customer, the score of dish i when it is ordered at the table for the k+1th time Updated to When ordering at the table for the k+1th time, the other dishes b that were not ordered have a score of and The same, no update is performed; the value range of b is also a positive integer between [1, a], and b is not equal to i.

[0012] When dish i is ordered by a customer, the score of dish i on the table is updated, and the scores of other dishes on the table are not updated, and the scores of dishes on other tables are not updated either; the scores of dishes are calculated and recommended based on the table, and at the same time, The item reflects the influence of the physical objects of the dishes around the table on the customers' orders. This item ensures that the recommended dishes for a single meal group will not deviate from the overall recommended dishes, thereby achieving the effect of recommending dishes based on the real-time situation in the store.

[0013] When the customer orders at the table for the k+1th time, the impact of customers who have ordered in the store on customers who have not ordered The calculation is done by the following formula:

[0014] Where A is the random variable of the dish, P{A=i} represents the influence of other customers in the store choosing dish i on the choice of dish i by customers who have not ordered. When other customers do not choose dish i, P{A=i} is 0; inf j It represents the influence of the tables around the target table and the tables passed by on the moving path when the customer orders at the target table. j The relevant data is stored in the form of: r1…r cIt represents the vertical distance between other tables and the path on which the customer goes to the target table. Only the data with vertical distance less than the threshold L is stored. The threshold L is set according to the visual range of the human eye. c is the number of different vertical distances between other tables and the path stored. R1…R d The straight-line distance between the tables around the target table and the target table. Only the data with a straight-line distance less than the threshold L is stored. d is the number of different straight-line distances stored; n1…n c Indicates the vertical distance r1…r c The number of other tables that choose dish i when N1…N d Indicates that the straight-line distance is R1…R d The value range of j is a positive integer between [1, e], where e is the number of different influence matrices and j is used to distinguish different influence matrices.

[0015] When a customer logs into the ordering system at the dining table, the customer has already reached the dining table from the store entrance, so within the visible range of the moving path, the customer can see the actual dishes ordered by other diners; at the dining table, the customer can also see the actual dishes ordered by other diners within the visible range; the farther the distance, the lower the probability that the customer notices the dishes at other tables, so the probability of the customer noticing other tables is positively correlated with the vertical distance and the inverse of the straight-line distance; different dishes have different appeal, and dishes with rich colors or fragrance are more likely to attract customers to place orders, so the impact of the actual dishes on customer orders is determined based on the type of dishes.

[0016] inf j This is determined by:

[0017] From the influence matrix MAT j Get the vertical distance, straight-line distance and the corresponding number of other tables that choose dish i,

[0018] When customers see the physical object of dish i, they will be encouraged to order dish i; therefore, the influence matrix will be affected by the physical objects of dishes ordered by other diners in the store at the current moment; on different tables with the same vertical distance from the moving path, it can be assumed that the probability of customers noticing different tables is the same, and the more times dish i appears, the greater the probability of attracting customers. Therefore, the influence of the physical objects of dishes ordered by other diners in the store on customers is positively correlated with the number of times dish i appears. Combined with the individual weight w2, it can be determined that inf j .

[0019] P{A=i} is determined by the following steps:

[0020] B is the location path random variable, P{B=j} represents the influence matrix MAT of the surrounding tables and the tables passed by on the moving path when the customer orders at the target location. j The probability of influencing the target table; P{B=j} is the influence matrix MAT j The ratio of the number of occurrences to the total number of occurrences of all influencing matrices; P{A=i|B=j} indicates that the target table is affected by the matrix MAT j The probability of a customer ordering dish i when affected is: j The number of dishes i ordered by customers and the target table are affected by the matrix MAT j The ratio of the total number of times affected; P{B=j|A=i} is the number of times a customer orders dish i at the target table and the target table is affected by the matrix MAT j The ratio of the number of impacts to the total number of times dish i is ordered on the target table.

[0021] After other customers in the store choose dish i, the influence P{A=i} on the choice of dish i by customers who have not ordered, that is, the attractiveness of the actual dish, is affected by factors such as the color and smell of the dish, and has little to do with the real-time situation in the store. Therefore, P{A=i} is a constant. After it is determined, it can remain unchanged or be updated regularly, without the need for real-time updates like the influence matrix, which can reduce computing resource consumption. Formula The calculated value is the attractiveness of the actual dishes under different influence matrices at a table in the store. The average value calculated under all influence matrices at the table position is taken to get the attractiveness of the actual dishes at the table position. Then the average value of the attractiveness of the actual dishes calculated under all tables is taken to get the final value of the attractiveness of the dishes in the store, which can reduce randomness and improve the versatility of the dish scoring model.

[0022] In step S1-4, after the customer logs into the ordering system, making a dish recommendation for the customer based on the dish score also includes the following steps:

[0023] S6-1, obtain the order time in the store in real time, and find the actual ordered dishes that customers see on the idle tables in the store;

[0024] S6-2, based on the physical dishes that the customer has seen and has ordered, obtain the influence of the tables around the target table when the customer orders at the target table, as well as the tables that the customer passes by on the moving path when he or she goes to the target table on the customer's order. j ;

[0025] S6-3, inf j Substituting into the formula we get based on Get the scores of the dishes, select the dishes with high scores and present them on the recommendation page of the ordering system; when the user logs in to the ordering system, real-time dish recommendations can be made.

[0026] In another aspect of the present invention, an event detection association system based on multivariate data is provided, comprising: a multivariate data acquisition module, a data storage module, a data analysis module and a dish recommendation module; the output end of the multivariate data acquisition module is interconnected with the data segments of the data analysis module and the data storage module, and is used to obtain the ordering, order placement and location path information of each table in the store; the output end of the data storage module is interconnected with the input end of the data analysis module, and is used to store the historical ordering and order placement data of all tables in the store; the output end of the data analysis module is interconnected with the input end of the dish recommendation module, and the scores of all dishes are calculated based on the customer's ordering location and the food placement status of other tables at the current moment, and the scores of the dishes are sent to the dish recommendation module; the dish recommendation module recommends dishes to customers in the recommendation page of the ordering system according to the scores of each dish.

[0027] The multivariate data acquisition module is also used to determine the moving path from the store entrance to each target table, and the vertical distance between other tables on the moving path and the moving path; determine the straight-line distance between other tables around each target table and the target table.

[0028] The data analysis module also includes a location path calculation unit, a dish impact calculation unit, a dish score calculation unit and a weight update unit; the location path calculation unit is used to obtain the matrix MAT affecting the target table. j The dish impact calculation unit determines the impact of the actual dish on the customer's order based on the historical order and placement data; the dish score calculation unit obtains the scores of all dishes based on the target table influence matrix and the impact of the actual dish on the customer's order; the weight update unit adjusts the inertia weight, individual weight and group weight according to the effect of the dish recommendation system.

[0029] The data analysis module calculates the impact of actual dishes on customer orders P{A=i} based on the historical impact matrix, order placement and ordering results. P{A=i} is affected by the type of dish and remains unchanged after determination. The impact matrix is ​​affected by the order placement situation in the store and the location of the target table. The impact matrix of the unmanned table at the current moment is determined in real time. The recommended dishes for each table are determined based on the impact matrix. When the customer logs into the ordering system at the table, dish recommendations are made to the customer in real time.

[0030] Compared with the prior art, the beneficial effects achieved by the present invention are: by scanning a QR code or using a mobile phone APP to place an order, the customer's waiting time can be saved, thereby improving the customer's dining experience; self-service ordering, self-service settlement and other functions can be realized, thereby improving the restaurant's operating efficiency; the impact of the actual ordered dishes on the customer's order is analyzed, and dishes are recommended when the customer orders, thereby speeding up the user's ordering speed; at the same time, it can also improve the problem that customers are attracted by the dishes in the store but are affected by not knowing the names of the dishes, and the ordered dishes and location paths are analyzed in real time to improve the accuracy of dish recommendations and thereby improve the efficiency of customer ordering. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0032] Figure 1 It is a structural diagram of an event detection association system based on multivariate data according to an embodiment of the present invention. DETAILED DESCRIPTION

[0033] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0034] In the embodiments of the present invention, please refer to Figure 1 , a structural schematic diagram of an event detection association system based on multivariate data is provided, comprising: a multivariate data acquisition module, a data storage module, a data analysis module and a dish recommendation module; the output end of the multivariate data acquisition module is interconnected with the data segments of the data analysis module and the data storage module, and is used to obtain the order, order placement and location path information of each table in the store; the output end of the data storage module is interconnected with the input end of the data analysis module, and is used to store the historical order and order placement data of all tables in the store; the output end of the data analysis module is interconnected with the input end of the dish recommendation module, and the scores of all dishes are calculated based on the customer's ordering location and the food placement status of other tables at the current moment, and the scores of the dishes are sent to the dish recommendation module; the dish recommendation module recommends dishes to customers in the recommendation page of the ordering system according to the scores of each dish.

[0035] The multivariate data acquisition module is also used to determine the moving path from the store entrance to each target table, as well as the vertical distance between other tables on the moving path and the moving path; and determine the straight-line distance between other tables around each target table and the target table.

[0036] The data analysis module also includes a location path calculation unit, a dish impact calculation unit, a dish score calculation unit and a weight update unit; the location path calculation unit is used to obtain the matrix MAT affecting the target table. j The dish impact calculation unit determines the impact of the actual dish on the customer's order based on the historical order and placement data; the dish score calculation unit obtains the scores of all dishes based on the target table influence matrix and the impact of the actual dish on the customer's order; the weight update unit adjusts the inertia weight, individual weight and group weight according to the effect of the dish recommendation system.

[0037] The data analysis module calculates the impact of actual dishes on customer orders P{A=i} based on the historical impact matrix, order placement and ordering results. P{A=i} is affected by the type of dish and remains unchanged after determination. The impact matrix is ​​affected by the order placement situation in the store and the location of the target table. The impact matrix of the unmanned table at the current moment is determined in real time. The recommended dishes for each table are determined based on the impact matrix. When the customer logs into the ordering system at the table, dish recommendations are made to the customer in real time.

[0038] In an embodiment of the present invention, a method for event detection and association based on multivariate data is provided, comprising:

[0039] S1-1, obtain the historical order and order data in the store;

[0040] S1-2, based on the store’s historical order data, analyze the impact of customers who have ordered in the store on customers who have not ordered, and train the scoring model for dishes:

[0041] In the formula, i is a positive integer between [1, a], which is used to represent different dishes, a is the total number of dishes; k is an integer greater than or equal to 0, which represents the number of times the table has been ordered by customers; The score of the i-th dish when the customer orders for the k+1th time at the table; It represents the influence of customers who have ordered food in the store on customers who have not ordered food when the customer orders food at the table for the k+1th time; is the average score of the i-th dish on all tables in the store when the customer orders for the k+1th time at the table; w1 is the inertia weight, which indicates the influence of the i-th dish ordered last time on the current order; w2 is the individual weight, which indicates the weight of the i-th dish score coming from the customers who have ordered; w3 is the group weight, which indicates the weight of the i-th dish score coming from other tables;

[0042] The initial values ​​of w1, w2 and w3 can be set to 1, and the weights can be adjusted according to the effect of the dish recommendation; if there is a large deviation between the dishes recommended by the table and the dishes recommended by the entire restaurant, for example, more than half of the recommended dishes are inconsistent with the dishes recommended on other tables, the group weight w3 can be increased and the inertia weight w1 and individual weight w2 can be reduced; if the influence of the actual dishes is high, for example, there is a situation where diners ask employees what dishes other people in the store have ordered, the value of the individual weight w2 can be increased; w1, w2 and w3 can be adjusted according to the situation.

[0043] The initial scores of the same dishes on each table are the same, and the initial scores of different dishes are determined according to the sales volume of the dishes; for example, the sales volume of dishes in categories 1, 2, and 3 are 100, 200, and 300 respectively, then the initial scores of dishes in categories 1, 2, and 3 on all tables can be set to 100 / 6, 200 / 3, and 300 / 6. Only one dish score is updated on each table at a time, and the scores of other dishes remain unchanged except for the updated dishes; when a user orders multiple dishes at the same time, the scores of the dishes are updated one by one; the scores of dishes on different tables are updated separately, and uploaded to the data storage unit after the update is completed, which affects the overall value;

[0044] After dish i is ordered by the customer, the score of dish i when it is ordered at the table for the k+1th time Updated to When ordering at the table for the k+1th time, the other dishes b that were not ordered have a score of and The same, no update is performed; the value range of b is also a positive integer between [1, a], and b is not equal to i.

[0045] The calculation is done by the following formula:

[0046] Where A is the random variable of the dish, P{A=i} represents the influence of other customers in the store choosing dish i on the choice of dish i by customers who have not ordered. When other customers do not choose dish i, P{A=i} is 0; inf jIt represents the influence of the tables around the target table and the tables passed by on the moving path when the customer orders at the target table. j The relevant data is stored in the form of: r1…r c It represents the vertical distance between other tables and the path on which the customer goes to the target table. Only the data with vertical distance less than the threshold L is stored. The threshold L is set according to the visual range of the human eye. c is the number of different vertical distances between other tables and the path stored. R1…R d The straight-line distance between the tables around the target table and the target table. Only the data with a straight-line distance less than the threshold L is stored. d is the number of different straight-line distances stored; n1…n c Indicates the vertical distance r1…r c The number of other tables that choose dish i when N1…N d Indicates that the straight-line distance is R1…R d The value range of j is a positive integer between [1, e], where e is the number of different influence matrices and j is used to distinguish different matrices.

[0047] The influence matrix is ​​affected by the type and distance of dishes. For Category 1 dishes, when the diners at the other two tables with straight-line distances of 2, 2, and 1 meters order Category 1, Category 2, and Category 1 dishes respectively, in the influence matrix, the number corresponding to the column with a straight-line distance of 2 meters is increased by 1, and the number corresponding to the column with a straight-line distance of 1 meter is increased by 1; for Category 2 dishes, the number corresponding to the column with a straight-line distance of 2 meters is increased by 1; scores are calculated for all dishes that can be ordered at the table.

[0048] inf j This is determined by:

[0049] From the influence matrix MAT j Get the vertical distance, straight-line distance and the corresponding number of other tables that choose dish i,

[0050] P{A=i} is determined by the following steps:

[0051] B is the location path random variable, P{B=j} represents the influence matrix MAT of the surrounding tables and the tables passed by on the moving path when the customer orders at the target location. j The probability of influencing the target table; P{B=j} is the influence matrix MAT j The ratio of the number of occurrences to the total number of occurrences of all influencing matrices; P{A=i|B=j} indicates that the target table is affected by the matrix MATj The probability of a customer ordering dish i when the target table is affected by the matrix MAT j The number of dishes i ordered by customers and the target table are affected by the matrix MAT j The ratio of the total number of times affected; P{B=j|A=i} is the number of times a customer orders dish i at the target table and the target table is affected by the matrix MAT j The ratio of the number of impacts to the total number of times dish i is ordered on the target table.

[0052] S1-3, obtain real-time order and order data in the store;

[0053] S1-4, using the dish scoring model to calculate the score of each dish, after the customer logs into the ordering system, the dish is recommended to the customer based on the dish score:

[0054] Get the order time in the store in real time and find the actual ordered dishes that customers see on the idle tables in the store;

[0055] Based on the physical dishes that customers have seen, we can get the influence of the tables around the target table and the tables on the path of the customer on the customer's order when the customer orders at the target table. j ;

[0056] Inf j Substituting into the formula we get based on Get the scores of the dishes, select the dishes with high scores and present them on the recommendation page of the ordering system; when the user logs in to the ordering system, real-time dish recommendations can be made.

[0057] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0058] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. An event detection association method based on multivariate data, characterized in that: The following steps are involved: S1-1, obtain the historical order and order data in the store; S1-2, based on the store's historical order data, analyze the impact of customers who have ordered in the store on customers who have not ordered, and train the dish scoring model; the dish score is calculated using the following formula: In the formula, i is a positive integer between [1, a], which is used to represent different dishes, a is the total number of dishes; k is an integer greater than or equal to 0, which represents the number of times the table has been ordered by customers; The score of the i-th dish when the customer orders for the k+1th time at the table; It represents the influence of customers who have ordered food in the store on customers who have not ordered food when the customer orders food at the table for the k+1th time; is the average score of the i-th dish on all tables in the store when the customer orders for the k+1th time at the table; w1 is the inertia weight, which indicates the influence of the i-th dish ordered last time on the current order; w2 is the individual weight, which indicates the weight of the i-th dish score derived from the part of customers who have ordered; w3 is the group weight, which indicates the weight of the score of the i-th dish from other tables; After dish i is ordered by the customer, the score of dish i when it is ordered at the table for the k+1th time Updated to When ordering at the table for the k+1th time, the other dishes b that were not ordered have a score of and The same, no update is performed; the value range of b is also a positive integer between [1, a], and b is not equal to i; When the customer orders at the table for the k+1th time, the impact of customers who have ordered in the store on customers who have not ordered The calculation is done by the following formula: Where A is the random variable of the dish, P{A=i} represents the influence of other customers in the store choosing dish i on the choice of dish i by customers who have not ordered. When other customers do not choose dish i, P{A=i} is 0; inf j It represents the influence of the tables around the target table and the tables passed by on the moving path when the customer orders at the target table. j The relevant data is stored in the form of: r1…r c It represents the vertical distance between other tables and the path on which the customer goes to the target table. Only the data with vertical distance less than the threshold L is stored. The threshold L is set according to the visual range of the human eye. c is the number of different vertical distances between other tables and the path stored. R1…R d The straight-line distance between the tables around the target table and the target table. Only the data with a straight-line distance less than the threshold L is stored. d is the number of different straight-line distances stored; n1…n c Indicates the vertical distance r1…r c The number of other tables that choose dish i when N1…N d Indicates that the straight-line distance is R1…R d , the number of other tables that choose dish i; the value range of j is a positive integer between [1, e], e is the number of different influence matrices, and j is used to distinguish different matrices; inf j This is determined by: From the influence matrix MAT j Get the vertical distance, straight-line distance and the corresponding number of other tables that choose dish i, P{A=i} is determined by the following steps: B is the location path random variable, P{B=j} represents the influence matrix MAT of the surrounding tables and the tables passed by on the moving path when the customer orders at the target location. j The probability of influencing the target table; P{B=j} is the influence matrix MAT j The ratio of the number of occurrences to the total number of occurrences of all influence matrices; P{A=i|B=j} indicates that the target table is affected by the matrix MAT j The probability of a customer ordering dish i when the target table is affected by the matrix MAT j The number of dishes i ordered by customers and the target table are affected by the matrix MAT j The ratio of the total number of times affected; P{B=j|A=i} is the number of times a customer orders dish i at the target table and the target table is affected by the matrix MAT j The ratio of the number of times affected to the total number of times dish i was ordered on the target table; S1-3, obtain real-time order and order data in the store; S1-4, using the dish scoring model to calculate the score of each dish, and after the customer logs into the ordering system, making dish recommendations for the customer based on the dish score.

2. The event detection and association method based on multivariate data according to claim 1, characterized in that: In step S1-4, after the customer logs into the ordering system, making a dish recommendation for the customer based on the dish score also includes the following steps: S6-1, obtain the order time in the store in real time, and find the actual ordered dishes that customers see on the idle tables in the store; S6-2, based on the physical dishes that the customer has seen and has ordered, obtain the influence of the tables around the target table when the customer orders at the target table, as well as the tables that the customer passes by on the moving path when he or she goes to the target table on the customer's order. j ; S6-3, inf j Substituting into the formula we get based on Get the scores of the dishes, select the dishes with high scores and present them on the recommendation page of the ordering system; when the user logs in to the ordering system, real-time dish recommendations can be made.

3. An event detection and association system based on multivariate data, using an event detection and association method based on multivariate data as described in any one of claims 1-2, characterized in that: include: Multivariate data collection module, data storage module, data analysis module and dish recommendation module; The output end of the multivariate data acquisition module is interconnected with the data segments of the data analysis module and the data storage module, and is used to obtain the ordering, ordering and location path information of each table in the store; the output end of the data storage module is interconnected with the input end of the data analysis module, and is used to store the historical ordering and ordering data of all tables in the store; the output end of the data analysis module is interconnected with the input end of the dish recommendation module, and the scores of all dishes are calculated based on the customer's ordering location and the current delivery status of other tables, and the scores of the dishes are sent to the dish recommendation module; The dish recommendation module recommends dishes to customers on the recommendation page of the ordering system according to the scores of each dish.

4. The event detection and association system based on multivariate data according to claim 3, characterized in that: The multivariate data acquisition module is also used to determine the moving path from the store entrance to each target table, as well as the vertical distance between other tables on the moving path and the moving path; and determine the straight-line distance between other tables around each target table and the target table.

5. The event detection and association system based on multivariate data according to claim 3, characterized in that: The data analysis module also includes a location path calculation unit, a dish impact calculation unit, a dish score calculation unit and a weight update unit; the location path calculation unit is used to obtain the matrix MAT affecting the target table. j The dish impact calculation unit determines the impact of the actual dish on the customer's order based on the historical order and order data; the dish score calculation unit obtains the scores of all dishes based on the target table influence matrix and the impact of the actual dish on the customer's order; The weight updating unit adjusts the inertia weight, individual weight and group weight according to the effect of the dish recommendation system.

6. The event detection and association system based on multivariate data according to claim 5, characterized in that: The data analysis module calculates the impact of actual dishes on customer orders P{A=i} based on the historical impact matrix, order placement and ordering results. P{A=i} is affected by the type of dish and remains unchanged after determination. The impact matrix is ​​affected by the order placement situation in the store and the location of the target table. The impact matrix of the unmanned table at the current moment is determined in real time. The recommended dishes for each table are determined based on the impact matrix. When the customer logs into the ordering system at the table, dish recommendations are made to the customer in real time.

Images