A recommendation method and system based on behavior analysis

Abstract

The present invention relates to the field of data processing, and in particular to a recommendation method and system based on behavior analysis. The method includes: obtaining historical sales data of a vending machine and preprocessing the data; evaluating the binding degree of each pattern based on the single-item frequency of commodity sales in the preprocessed historical sales data and the combined frequency of patterns in the Apriori algorithm, and adjusting the preset minimum support threshold based on the binding degree, wherein the binding degree is inversely proportional to the preset minimum support threshold, obtaining a corrected frequent item set, and formulating a sales plan for commodity combinations based on the corrected frequent item set; obtaining the preference degree of each package based on the sales frequency of the package and the sales frequency of each commodity in the package, and adjusting the package based on the preference degree of the package; The present invention evaluates and quantifies the binding degree between commodities, dynamically adjusts the minimum support threshold, so as to formulate a package demand that better meets the needs of customers.

Description

Technical Field

[0001] The present invention relates to the field of data processing. In particular, it relates to a recommendation method and system based on behavior analysis. Background Art

[0002] With the deepening of the concept of new retail and the popularization of Internet of Things technology, the unmanned vending machine industry is transforming from the traditional "mechanized delivery" mode to the "intelligent service" mode. However, the rapid expansion of this industry has also exposed some significant pain points. Traditional unmanned vending machines rely on fixed product combinations and empirical replenishment strategies, making it difficult to adapt to the dynamic changes in consumer demand, resulting in a relatively low average sell-out rate of products. Against this background, the product combination recommendation technology based on big data has become the key to solving the industry's dilemmas.

[0003] "Behavior analysis" is a technology or method that observes, records, analyzes, and predicts the behavior patterns of individuals or groups. It can obtain valuable information and provide support for decision-making. Behavior analysis is widely used in many fields such as psychology, marketing, computer science, and sociology. In the unmanned vending machine industry, behavior analysis can optimize product combinations and recommendation strategies through in-depth research on consumer purchase behavior, thereby improving sales efficiency and customer satisfaction.

[0004] In the existing product recommendation technology, the Apriori algorithm is often used for product matching recommendation in unmanned vending machines. Currently, the Apriori algorithm has obvious limitations. For products that are often purchased together with other specific products in the purchase behavior but have a low self-selling frequency, the fixed minimum support threshold is difficult to identify them as frequent item sets. This results in the inability to effectively co-sell these products with other products, thus missing some potentially useful patterns and reducing the overall efficiency of product sales. Therefore, it is unable to cope with the diverse and dynamic needs of consumers. Summary of the Invention

[0005] To solve the problem that the Apriori algorithm is difficult to identify products with a low self-selling frequency but are often purchased together with other specific products, resulting in the inability to effectively co-sell and thus reducing the overall sales efficiency, the present invention provides solutions in the following aspects.

[0006] In a first aspect, a recommendation method based on behavior analysis includes: obtaining historical sales data of a vending machine and preprocessing the data; evaluating the binding degree of each pattern based on the single-item frequency of commodity sales and the combined frequency of patterns in the Apriori algorithm in the preprocessed historical sales data, and adjusting a preset minimum support threshold based on the binding degree, where the binding degree is inversely proportional to the preset minimum support threshold, obtaining a corrected frequent item set, and formulating a sales plan for commodity combinations based on the corrected frequent item set; obtaining the preference degree of each package based on the sales frequency of the package and the sales frequency of each commodity in the package, and adjusting the package based on the preference degree of the package; where the combined frequency refers to the frequency of a certain commodity or pattern being purchased in combination with other commodities, and the binding degree refers to the intensity of the frequency of a certain commodity being purchased together with other specific commodities relative to its frequency of being purchased alone, and a sales plan is a package.

[0007] By analyzing the historical sales data of the vending machine, it aims to solve the problem that it is difficult for the traditional Apriori algorithm to identify low-frequency and high-binding commodity combinations. First, preprocess the sales data to ensure data accuracy, evaluate the binding degree of each commodity combination pattern, and dynamically adjust the minimum support threshold according to this binding degree to ensure that commodity combinations that are sold infrequently alone but are often purchased together with other commodities can be identified as frequent item sets. Also, calculate the preference degree of each package based on the actual sales performance of the package and adjust the package accordingly. It is beneficial to improve sales efficiency, optimize inventory management, and enhance customer satisfaction, thereby promoting the overall performance and market competitiveness of the vending machine. It can more accurately formulate commodity combination sales plans to adapt to the diversification and dynamism of consumer demands.

[0008] Preferably, the preprocessing of the data includes:

[0009] According to the historical sales data, obtain all the types of commodities sold and the sales records, where the sales records include: sales time, sold commodities, and transaction amounts;

[0010] Unify the sales time into a standard time format, assign a unique code to each commodity, and convert the transaction amounts into a unified currency unit.

[0011] Preprocessing the historical sales data of the unmanned vending machine is a crucial step in implementing the behavior analysis and recommendation method. The effects of preprocessing are reflected in the following aspects: First, by unifying the sales time into a standard time format, the consistency and comparability of time data are ensured; Second, a unique code is assigned to each commodity, improving the accuracy of data processing and the tracking efficiency; Finally, the transaction amount is converted into a unified currency unit, facilitating financial analysis and comparison. Together, they ensure the quality and consistency of the data, laying a solid foundation for subsequent data analysis and pattern recognition, enabling the recommendation system to more accurately capture consumers' behavior patterns, formulate more effective commodity combination sales strategies, and ultimately improve sales efficiency and customer satisfaction.

[0012] Preferably, the combined frequency sequence includes:

[0013] Separate the combined frequencies of individual commodities and various combined sales according to the preprocessed historical sales data, and arrange the combined frequencies in descending order to form a combined frequency sequence.

[0014] Provide a clear perspective to observe and understand consumers' purchase behavior patterns. The combined frequency sequence reveals which commodity combinations are the most popular, helping us identify potential high-benefit sales combinations, facilitating the unmanned vending machine to more precisely customize commodity combinations and packages, optimize inventory management, reduce excess inventory, and at the same time increase the average customer price through bundled sales.

[0015] Preferably, obtaining the binding degree of each pattern includes:

[0016] Taking any commodity or pattern as the marked commodity or pattern, calculate the ratio between the maximum combined frequency of the marked commodity or pattern and the occurrence frequency of the marked commodity to obtain the association strength;

[0017] Calculate the difference between adjacent frequency values in the combined frequency sequence to obtain the frequency difference degree. Calculate the ratio between the number of times the marked commodity or pattern appears alone in the historical sales data and the total number of all different combinations participated by the marked commodity or pattern minus one to obtain the average distribution value;

[0018] Select the maximum value between the average distribution value and 1 as the weighting value, and take the product of the weighting value and the degree of association as the binding degree of the target commodity or pattern.

[0019] It can accurately identify and strengthen the correlation between commodities, thereby optimizing inventory management, improving sales efficiency, enhancing customer satisfaction, and supporting the adjustment of sales strategies according to market dynamics, providing data-driven commodity combinations and recommendation strategies for the unmanned vending machine to improve operational efficiency and market competitiveness.

[0020] Preferably, adjusting the preset minimum support threshold based on the binding degree includes:

[0021] Taking the ratio between the preset minimum support threshold and the binding degree as the significant minimum support threshold, and using the significant minimum support threshold to rescan the transaction data to identify significant frequent item sets.

[0022] It is beneficial to adapt to the diversification and dynamism of consumer demands, making the recommendation system more flexible and accurate, and ultimately promoting the growth of sales performance and the improvement of market competitiveness.

[0023] Preferably, the significant frequent item sets include:

[0024] Sorting out the historical sales data, counting the frequency of each commodity appearing alone to determine the frequent 1-item sets, generating larger candidate item sets step by step based on the frequent 1-item sets, and counting the appearance frequency of these candidate item sets by scanning the data set, screening out the qualified frequent item sets according to the set minimum support threshold, continuously generating larger item sets and screening until no new frequent item sets can be found, and obtaining the significant frequent item sets.

[0025] It can extract valuable patterns from the original data, that is: those combinations of commodities that are often purchased together. The identification of significant frequent item sets helps to optimize inventory management, highlighting which commodities are often purchased together by customers, so that replenishment and promotion can be carried out more targeted. In addition, these frequent item sets can also be used to formulate more effective marketing strategies and improve the customer experience, such as recommending commodity combinations that customers may be interested in through bundling sales or recommendation systems.

[0026] Preferably, obtaining the preference degree of each package includes:

[0027] Taking any package as the target package, calculating the ratio between the selling frequency of the target package in the package sales data and the minimum individual selling frequency of all commodities in the package to obtain the package performance intensity ratio;

[0028] Taking the package performance intensity ratio as the preference degree of each package.

[0029] Preferably, obtaining the preference degree of each package further includes:

[0030] Taking any package as the target package, calculating the ratio between the selling frequency of the target package in the package sales data and the minimum individual selling frequency of all commodities in the package to obtain the package performance intensity ratio;

[0031] Calculating the sum of the historical selling frequencies of each commodity in the target package and the package selling frequency, and then dividing it by the average value of the sum of the individual selling frequencies of the corresponding commodities to obtain the package sales performance index;

[0032] Take the product of the package performance intensity ratio and the package sales performance index as the preference degree of the target package.

[0033] Preferably, the package adjustment based on the preference degree of the package includes:

[0034] Preset a preference degree threshold for the package. In response to the preference degree of the package being less than the preference degree threshold, it is considered that the package is not suitable for combined sales and should be cancelled. On the contrary, if it is greater than or equal to the preference degree threshold, it is considered that the package is suitable for combined sales.

[0035] In a second aspect, a recommendation system based on behavior analysis includes: a processor and a memory. The memory stores computer program instructions, and when the computer program instructions are executed by the processor, the above-mentioned recommendation method based on behavior analysis is implemented.

[0036] The present invention has the following effects:

[0037] 1. By analyzing the single-item frequency and combined frequency of products in historical sales data, the present invention evaluates and quantifies the binding degree between products, and then dynamically adjusts the minimum support threshold. It can identify those products that, although have a low single-sale frequency, are often purchased together with other specific products. By including these products in the combined sales plan, the average customer price and total sales amount can be more effectively increased, thereby improving the overall sales efficiency.

[0038] 2. By formulating a product combined sales plan through significant frequent item sets, the present invention can reduce inventory backlogs, ensure that the products in the inventory are more in line with consumers' purchase habits and preferences, which is conducive to reducing inventory costs and at the same time increasing inventory turnover rate, enabling the unmanned vending machine to more flexibly respond to market changes.

[0039] 3. The recommendation method based on behavior analysis of the present invention can provide product combinations that better meet customers' needs, thereby enhancing the shopping experience of customers. By accurately recommending product packages that customers may be interested in, the trust and satisfaction of customers towards the unmanned vending machine can be increased, promoting customer loyalty and bringing more potential customers to the vending machine. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 It is a flowchart of the method from step S1 to step S3 in a recommendation method based on behavior analysis according to an embodiment of the present invention.

[0041] Figure 2 It is a structural block diagram of a recommendation system based on behavior analysis according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0042] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments.

[0043] Referring to Figure 1 , a recommendation method based on behavior analysis includes steps S1 - S3, specifically as follows:

[0044] S1: Obtain the historical sales data of the vending machine and preprocess the data;

[0045] Collect the historical sales data within one month or the most recent two hundred times.

[0046] According to the historical sales data, obtain all the types of sold goods and detailed sales records, including but not limited to: information such as sales time, sold goods, and transaction amount;

[0047] Unify the sales time into a standard time format, assign a unique code to each commodity, and convert the transaction amount into a unified currency unit to ensure the consistency and comparability of the data.

[0048] It is also necessary to further preprocess the historical sales data, including: removing duplicate records, filling or deleting missing values, and correcting incorrect data, so as to improve the data quality, and aggregate the data according to the time, commodity, and transaction dimensions, so as to better analyze the sales trend of commodities and the opportunity of combined sales.

[0049] Through these preprocessing steps, we can ensure the accuracy and integrity of the data, lay a solid foundation for the subsequent analysis of combined commodity sales. Based on high-quality data, we can further analyze the correlation between commodities, design an effective combined commodity sales plan, thereby increasing the average customer price, reducing inventory, and ultimately improving the overall sales efficiency of the vending machine.

[0050] S2: Evaluate the binding degree of each pattern according to the single-item frequency of commodity sales in the preprocessed historical sales data and the combined frequency of patterns in the Apriori algorithm, adjust the preset minimum support threshold based on the binding degree, where the binding degree is inversely proportional to the preset minimum support threshold, obtain the corrected frequent item set, and formulate a sales plan for commodity combinations based on the corrected frequent item set.

[0051] Among them, the combined frequency sequence refers to the frequency of a certain commodity or pattern being purchased in combination with other commodities, and the binding degree refers to the intensity of the frequency of a certain commodity being purchased together with other specific commodities relative to its frequency of being purchased alone. A sales plan is a kind of package.

[0052] Obtain the combined frequency sequence:

[0053] Obtain the combination frequencies of individual products and various combined sales respectively based on historical sales data;

[0054] Exemplarily, for product the frequency of individual sales is 100 times, for product and product the frequency of combined sales is 50 times, for product and product the frequency of combined sales is 30 times, for product and product and product the frequency of combined sales is 20 times, obtaining the different combination frequencies of product .

[0055] Arrange the combination frequencies in descending order to form a combination frequency sequence.

[0056] In addition, in another embodiment, it further includes performing clustering analysis on historical sales data:

[0057] Convert the purchase behavior of each user into a feature vector. For example, take the number of purchases of each product as a feature, such as product categories: product 、product 、product 、product ;

[0058] User 1 purchased product 2 times, product 1 time, product 3 times, product 0 times, and the feature vector is ;

[0059] User 2 purchased product 1 time, product 2 times, product 0 times, product 1 time, and the feature vector is .

[0060] In addition, the purchase time feature can also be added. For example, divide the time period into weekdays and non - weekdays (Monday to Friday, weekends), set the purchase feature on weekdays to 1, and on non - weekdays to 0.

[0061] Use clustering algorithms (such as K-Means, hierarchical clustering, etc.) to cluster the purchase behaviors of users, obtaining several clusters. For each cluster, calculate the frequency of the combination of products purchased by all users in it. Extract the purchase records of all users from each cluster to generate all possible product combinations, count the number of times each product combination appears in the cluster, and sort the combination frequencies in descending order to form a combination frequency sequence.

[0062] It should be noted that if a certain product shows a significantly higher sales frequency with a specific product only in the combined sales with many other products, it can be considered that the product has a high binding degree (BindingDegree). This phenomenon indicates that the product shows a strong dependence in the sales scenario and is more suitable for being sold together with the bound product rather than being sold alone as an independent item.

[0063] Obtain the binding degree of each pattern, including:

[0064] Taking any product or pattern as the marked product or pattern, calculate the ratio between the maximum combination frequency of the marked product or pattern and the appearance frequency of the marked product to obtain the association strength;

[0065] Calculate the difference between adjacent frequency values in the combination frequency sequence to obtain the frequency difference degree. Calculate the ratio between the number of times the marked product or pattern appears alone in the historical sales data and the total number of all different combinations participated by the marked product or pattern minus one to obtain the average distribution value;

[0066] Select the maximum value between the average distribution value and 1 as the weighting value, and take the product of the weighting value and the association degree as the binding degree of the target product or pattern.

[0067] Specifically, if the frequency of a product or a pattern appearing in combination with another product in the historical sales data is much greater than the combination frequency with other products, it is considered that the product or pattern has a high binding degree. For any product or pattern, combine it with other single products. Combine the th product or pattern with the th product among all other products (the th product does not belong to the th pattern and is not the th product either), and arrange the frequencies in descending order to obtain the combination frequency order sequence of the th product or pattern , and record the th frequency value as . According to the appearance frequency of the th product or pattern combined with each other product, obtain the binding degree.

[0068] Specifically, the binding degree satisfies the following relational expression:

[0069] ;

[0070] In the formula, represents the binding degree of the th kind of commodity or mode, represents, represents the th kind of commodity or mode's occurrence frequency, , respectively represent the frequency values of the th and th combinations in the combined frequency order sequence of the th kind of commodity or mode, represents the number of commodity types included in the th kind of commodity or mode.

[0071] It should be noted that represents taking the maximum value of all possible combined frequencies . A larger value indicates that the occurrence frequency of the th kind of commodity or mode in the th combination is much higher than its individual occurrence frequency, which indicates that the th kind of commodity has a strong correlation or binding relationship with other commodities in the th combination; conversely, a smaller value indicates a weaker correlation of the th kind of commodity with other commodities in the th combination.

[0072] For example, if the combined frequency of commodity and commodity is much higher than the individual occurrence frequency of commodity , it means that when consumers purchase commodity , they often purchase commodity at the same time.

[0073] reflects the frequency value obtained by evenly distributing the individual occurrence frequency of the th kind of commodity or mode to all its possible combinations. This is because when calculating the combined frequency difference, we usually do not consider the situation where the th kind of commodity or mode appears individually, so it is only distributed to combinations. ​

[0074] , it ensures that the weighted value of the difference is not less than 1, avoiding underestimating the binding degree due to too small a difference, which helps to more accurately identify the correlation between products, optimize product recommendations, inventory management, and marketing strategies.

[0075] It means starting from the second combined frequency because is the highest combined frequency and there is no previous combined frequency for comparison, so the difference is calculated starting from the second combined frequency.

[0076] By calculating the weighted sum of the combined frequency differences, the correlation between products can be more accurately identified and product recommendations optimized. Understanding the weighted sum of the combined frequency differences can help optimize inventory management to ensure sufficient inventory of products that are often purchased together. Based on the weighted sum of the combined frequency differences, more effective promotional activities can be designed, such as bundling sales or preferential packages.

[0077] Take the ratio between the preset minimum support threshold and the binding degree as the significant minimum support threshold, and use the significant minimum support threshold to rescan the transaction data to identify significant frequent item sets.

[0078] Obtain significant frequent item sets, including:

[0079] Sort out the historical sales data, count the frequency of each product appearing alone to determine the frequent 1-item set. Based on the frequent 1-item set, gradually generate larger candidate item sets, and count the occurrence frequency of these candidate item sets by scanning the data set. Screen out the qualified frequent item sets according to the set minimum support threshold, continuously generate larger item sets and screen them until no new frequent item sets can be found, and then stop to obtain the significant frequent item sets.

[0080] Exemplarily, the transaction data is: Transaction 1: , Transaction 2: , Transaction 3: , Transaction 4: and Transaction 5: ; Set the minimum support threshold to 0.4 (i.e., appear in at least 2 transactions). Generate the following frequent 1-item set:

[0081] Product appears in Transactions 1, 2, 3, and 5, a total of 4 times, and the support degree is , which is greater than 0.4, so product is frequent.

[0082] Product appears in Transactions 1, 2, and 4, a total of 3 times, and the support degree is , which is greater than 0.4, so product is frequent.

[0083] Product appears in transactions 1, 3, and 4, a total of 3 times, with a support of , greater than 0.4, so product is frequent.

[0084] Product appears in transaction 5, a total of 1 time, with a support of , less than 0.4, so product is not frequent.

[0085] Therefore, the frequent 1-itemsets are: , , .

[0086] The frequent 2-itemsets generated are:

[0087] Product combination appears 2 times, with a support of , equal to 0.4, so product B is frequent.

[0088] Product combination appears 2 times, with a support of , equal to 0.4, so product B is frequent.

[0089] Product combination appears 2 times, with a support of , equal to 0.4, so product B is frequent.

[0090] Product combination appears 1 time, with a support of , less than 0.4, so product D is not frequent.

[0091] Therefore, the frequent 2-itemsets are , , .

[0092] The frequent 3-itemsets generated are:

[0093] Product combination : 1 time, less than the minimum support threshold, so there are no frequent itemsets in the frequent 3-itemsets;

[0094] As a result, the frequent itemsets: , , , , , .

[0095] In this implementation, the preset minimum support threshold is , which can be adjusted according to specific circumstances.

[0096] By extracting frequent item sets from historical sales data, calculating the binding degree of each product or pattern, and forming a combined frequency sequence. This information can be used to optimize product recommendations, inventory management, and marketing strategies, thereby improving the sales efficiency and customer satisfaction of the vending machine.

[0097] S3: Obtain the preference degree of each package according to the sales frequency of the package and the sales frequency of each product in the package, and adjust the package based on the priority degree of the package.

[0098] It should be noted that if the sales frequency of the combined package is too low after the package is sold, or the sales frequency of the package seriously affects the individual sales frequency of each product in it, it is considered that the preference degree of this package option is relatively low, and such a package should be cancelled by increasing the minimum support threshold.

[0099] Obtaining the preference degree of each package includes:

[0100] Taking any package as the target package, calculate the ratio between the sales frequency of the target package in the package sales data and the minimum individual sales frequency of all products in the package to obtain the package performance intensity ratio;

[0101] Use the package performance intensity ratio as the preference degree of each package.

[0102] Specifically, after launching the product package for sale, collect the sales records within one month or 100 times as the package sales data, and record the number of frequent item sets as , each frequent item set corresponds to a combined sales plan for a product package, and record the th package as the th product set .

[0103] Specifically, the preference degree of the package satisfies the following relational expression:

[0104] ;

[0105] In the formula, represents the preference degree of the th package, represents the sales frequency of the th package in the package sales data, represents the minimum value among the individual sales frequencies of all products in the th package, represents the sales frequency of the th product in the historical sales data.

[0106] That is to say, It reflects the individual sales situation of the "least popular" product in the package, indicating the th power of the intensity of the sales frequency of the th package relative to the individual sales frequency of the "least popular" product in the package. A larger ratio indicates that the sales frequency of the th package is much higher than the individual sales frequency of any product in the package, indicating that the market performance of the

[0107] In addition, in another embodiment, it further includes:

[0108] Taking any package as the target package, calculate the ratio between the sales frequency of the target package in the package sales data and the minimum individual sales frequency of all products in the package to obtain the package performance intensity ratio;

[0109] Calculate the sum of the historical sales frequencies of each product in the target package and the package sales frequency, and then divide it by the average value of the sum of the individual sales frequencies of the corresponding products to obtain the package sales performance index;

[0110] Take the product of the package performance intensity ratio and the package sales performance index as the preference degree of the target package.

[0111] Obtain the sales frequencies of each package and each individual product. Denote the sales frequency of the th product as , denote the sales frequency of the th package as , and denote the number of product types included in the combined sales of the th package as .

[0112] Specifically, the preference degree of the package satisfies the following relational expression:

[0113] ;

[0114] In the formula, represents the preference degree of the th package, represents the sales frequency of the th package in the package sales data, represents the minimum value among the individual sales frequencies of all products in the th package, represents the sales frequency of the th product in the historical sales data, represents the number of product types included in the th package, represents the set of products corresponding to the th package, represents the th product's sales frequency in historical sales data.

[0115] That is to say, represents the strength of the comprehensive performance of the th product in historical sales and package sales relative to its individual sales frequency. If is larger, it indicates that the package sales method helps the sales efficiency of each product in the package more, that is, the higher the preference level of the package. On the contrary, is smaller, it indicates that the package sales method helps the sales efficiency of each product in the package less, that is, the lower the preference level of the package.

[0116] By comparing the individual sales frequencies of each product in the package, it is possible to understand which products in the package have poor market performance and further analyze the reasons or adjust the marketing strategy.

[0117] Preset a preference level threshold for the package. In response to the preference level of the package being less than the preference level threshold, it is considered that the package is not suitable for combined sales and should be cancelled. On the contrary, if it is greater than or equal to the preference level threshold, it is considered that the package is suitable for combined sales.

[0118] Exemplarily, the preference level threshold is 0.7 and can be adjusted according to specific circumstances.

[0119] The present invention also provides a recommendation system based on behavior analysis. As Figure 2 shown, the system includes a processor and a memory. The memory stores computer program instructions, which, when executed by the processor, implement a recommendation method based on behavior analysis according to the first aspect of the present invention. The system also includes a communication bus and a communication interface and other components well known to those skilled in the art. Their settings and functions are known in the art, so they will not be described in detail here.

[0120] It should be noted that for those of ordinary skill in the art, without departing from the concept of the present invention, several modifications and improvements can still be made, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the present invention patent shall be subject to the appended claims.

Claims

1. A recommendation method based on behavior analysis, characterized in that, Including: Obtain the historical sales data of the vending machine and preprocess the data; Evaluate the binding degree of each pattern according to the single-item frequency of commodity sales in the preprocessed historical sales data and the combined frequency of patterns in the Apriori algorithm, including: obtaining the combined frequencies of individual commodities and various combined sales respectively from the preprocessed historical sales data, arranging the combined frequencies in descending order to form a combined frequency sequence; using any commodity or pattern as the marked commodity or pattern, calculating the ratio between the maximum combined frequency of the marked commodity or pattern and the occurrence frequency of the marked commodity to obtain the association strength; calculating the difference between adjacent frequency values in the combined frequency sequence to obtain the frequency difference degree, calculating the ratio between the number of times the marked commodity or pattern appears alone in the historical sales data and the total number of all different combinations participated by the marked commodity or pattern minus one to obtain the average distribution value; selecting the maximum value between the average distribution value and 1 as the weighting value, and taking the product of the weighting value and the association degree as the binding degree of the target commodity or pattern; Adjust the preset minimum support threshold based on the binding degree, where the binding degree is inversely proportional to the preset minimum support threshold, obtain the corrected frequent item set, and formulate a sales plan for commodity combinations based on the corrected frequent item set; Obtain the preference degree of each package according to the sales frequency of the package and the sales frequency of each commodity in the package, and adjust the package based on the preference degree of the package; Among them, the combined frequency refers to the frequency of a certain commodity or pattern being purchased in combination with other commodities, the binding degree refers to the intensity of the frequency of a certain commodity being purchased together with other specific commodities relative to its frequency of being purchased alone, and a sales plan is a package.

2. The recommended method based on behavior analysis according to claim 1, wherein The preprocessing of the data includes: According to the historical sales data, obtain all the types of commodities sold and the sales records, where the sales records include: sales time, sold commodities, and transaction amount; Unify the sales time into a standard time format, assign a unique code to each commodity, and convert the transaction amount into a unified currency unit.

3. The recommended method based on behavior analysis according to claim 1, characterized in that, The adjustment of the preset minimum support threshold based on the binding degree includes: Taking the ratio between the preset minimum support threshold and the binding degree as the significant minimum support threshold, and using the significant minimum support threshold to rescan the transaction data to identify the significant frequent item set.

4. The recommended method based on behavior analysis according to claim 3, wherein The identification of the significant frequent item set includes: Sort out the historical sales data, count the frequency of each commodity appearing alone to determine the frequent 1-item set, based on the frequent 1-item set, gradually generate larger candidate item sets, and count the occurrence frequencies of these candidate item sets by scanning the data set, screen out the qualified frequent item sets according to the set minimum support threshold, continuously generate larger item sets and screen until no new frequent item sets can be found, and obtain the significant frequent item set.

5. The recommended method based on behavior analysis according to claim 1, characterized in that The obtaining of the preference degree of each package includes: Taking any package as the target package, calculate the ratio between the sales frequency of the target package in the package sales data and the minimum single sales frequency of all commodities in the package to obtain the package performance strength ratio; Use the package performance intensity ratio as the preference degree of each package.

6. The recommended method based on behavior analysis according to claim 1, characterized in that The obtaining of the preference degree of each package further includes: Taking any package as the target package, calculating the ratio between the selling frequency of the target package in the package sales data and the minimum individual selling frequency of all the products in the package to obtain the package performance intensity ratio; Calculating the sum of the historical selling frequencies of each product in the target package and the package selling frequency, and then dividing it by the average value of the sum of the individual selling frequencies of the corresponding products to obtain the package sales performance index; Taking the product of the package performance intensity ratio and the package sales performance index as the preference degree of the target package.

7. A recommendation method based on behavior analysis according to claim 1, characterized in that, The package adjustment based on the preference degree of the package includes: Presetting a preference degree threshold for the package. In response to the preference degree of the package being less than the preference degree threshold, it is considered that the package is not suitable for combined sales and should be cancelled. On the contrary, if it is greater than or equal to the preference degree threshold, it is considered that the package is suitable for combined sales.

8. A recommendation system based on behavior analysis, characterized in that, Including: A processor and a memory, the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the recommendation method based on behavior analysis according to any one of claims 1-7 is implemented.

Images