Bayesian-based anchor commodity distribution method and system

Abstract

The invention relates to a Bayesian-based anchor commodity distribution method and system, and the method comprises the following steps: obtaining to-be-sold commodity information and to-be-distributed anchor information, calculating the probability that each anchor sells each type of commodities in to-be-distributed anchors, calculating the overall selling probability of each anchor, calculating the overall selling probability of the commodities, and carrying out the distribution of the commodities in the to-be-distributed anchors. And calculating the probability that each anchor in the to-be-allocated anchors is allocated to each type of commodity through a Bayesian algorithm, and allocating each commodity to the anchors in the corresponding random number interval according to the random number and the anchor allocation probability. According to the method, the distribution probability of the anchor is calculated by combining the historical sales data according to the Bayesian algorithm, so that automation of commodity distribution is realized, and the problems of low efficiency, incapability of combining historical sales records, subjectivity and prejudice of distribution personnel and the like in the prior art are solved.

Description

Technical Field

[0001] The present invention belongs to the technical field of network live broadcasting, and specifically relates to a method and system for allocating host products based on Bayesian analysis. Background Art

[0002] At present, the development of Internet technology has enriched people's shopping channels. On the basis of online shopping, online live shopping has emerged. Online live shopping is done by anchors showing products to customers in real time through live broadcasts, demonstrating product functions, etc. When watching shopping live broadcasts, customers can directly purchase the products displayed by the anchor in the anchor's live broadcast room. Now more and more live broadcasts tend to carry more comprehensive categories, and even open different category numbers to specialize in a certain category; but for a company, there are many types of goods and multiple anchors, which makes it a problem to assign a product to which anchor, especially for second-hand luxury goods. Due to their uniqueness and scarcity, and the fact that each second-hand luxury product is different in category, brand, price and quality, each second-hand luxury product needs to be sold one by one. Therefore, there needs to be a different way of distribution for such goods. However, the current distribution of such second-hand luxury goods is done manually, but manual distribution faces many problems:

[0003] (1) Inefficiency: Manual allocation of goods takes a lot of time, especially when there are a large number of goods and anchors, which may lead to a slow decision-making process.

[0004] (2) Subjectivity and bias: Manual allocation may be affected by the personal preferences of the allocator, resulting in some anchors receiving more high-quality goods while other anchors may be ignored. In addition, this allocation method may lead to inequality among anchors, affecting team morale and cooperation atmosphere.

[0005] (3) Insufficient data processing: Manual allocation usually lacks in-depth analysis of sales data, customer feedback and other information, which may lead to unreasonable product allocation. It is difficult to use historical data to analyze which products are more suitable for which anchors, so as to optimize allocation.

[0006] (4) Low product utilization: Some products may not be displayed by suitable anchors due to improper allocation, resulting in poor sales. Popular products may be concentrated in a few anchors, and other anchors cannot make full use of these resources, affecting overall sales performance.

[0007] (5) Increased communication costs: Communication barriers may arise between different teams (such as marketing, sales, customer service, etc.), resulting in poor information transmission and affecting allocation decisions. Summary of the invention

[0008] The purpose of the present invention is to provide a method for allocating anchor products based on Bayesian in order to overcome the defects of the above-mentioned prior art, comprising the following steps:

[0009] Obtain information on goods to be sold and information on anchors to be assigned, the goods information including the type of goods, quantity of goods and historical sales records of each type of goods, and the anchor information including the number of anchors to be assigned, anchors' historical sales records, number of fans, and live broadcast room turnover;

[0010] Calculate the probability of each anchor selling each type of product among the anchors to be assigned based on the product information and the anchor's historical sales records;

[0011] The overall sales probability of each anchor is obtained through weighted calculation based on the anchor's historical sales record, number of fans, and live broadcast room turnover in the anchor's information;

[0012] The overall sales probability of each product category is calculated through the historical sales records of each type of product;

[0013] Based on the three probabilities calculated above, the probability of each anchor being assigned to each type of product is calculated using the Bayesian algorithm;

[0014] According to the probability of each anchor being assigned to each type of product, a random number interval corresponding to the probability is assigned to each anchor;

[0015] For each product, a corresponding random number is generated, and it is determined in which random number interval the random number falls, and the product is allocated to the anchor corresponding to the random number interval. If the allocated product of the corresponding anchor is equal to the second preset value, a random number is re-generated for the product and reallocated.

[0016] Furthermore, the commodities are classified according to commodity category, brand, price and quality.

[0017] Furthermore, the calculation of the probability of each anchor among the anchors to be assigned selling each type of goods according to the anchors' historical sales records specifically includes the following steps:

[0018] Obtain information on products to be sold and sales records of each anchor to be assigned;

[0019] Extract the sales records of each anchor in the first preset time period, and calculate the probability of the anchor selling each type of goods to be sold in the first preset time period according to the sales records in the period;

[0020] Extracting the sales records of each anchor in the second preset time period, and calculating the probability of the anchor selling each type of goods to be sold in the second preset time period according to the sales records in the period;

[0021] The probability of the first preset time period and the probability of the second preset time period are weighted and summed to obtain the probability P(B i |A j ),

[0022] Determine the probability P(B) of each anchor selling each type of product i |A j ) is greater than the first preset value. If less than, the value is set to the first preset value; if greater than or equal to, it remains unchanged.

[0023] Furthermore, the probability of each anchor selling each type of commodity among the anchors to be assigned is calculated based on the anchor's historical sales records, and the calculation formula is:

[0024] P 1 (B i |A j )=x 1ij / Y 1ij

[0025] P 2 (B i |A j )=x 2ij / Y 2ij

[0026] P(B i |A j )=αP 1 (B i |A j )+βP 2 (B i |A j )

[0027] Among them, x 1ij is the number of products of type i sold by anchor j in the first preset time period, x 2ij is the number of products of type i sold by anchor j in the second preset time period, Y 1ij is the number of commodities of type i assigned to anchor j in the first preset time period, Y 2ij is the number of commodities of type i assigned to anchor j in the second preset time period, P 1 (B i |A j ) is the probability that anchor j sells product of type i in the first preset time period, P 2 (B i |A j ) is the probability that anchor j sells product of type i in the second preset time period, and α and β are weight values.

[0028] Furthermore, the first preset time period is nearly one month, the second preset time period is nearly half a year to nearly one month, and α is greater than β.

[0029] Furthermore, the overall sales probability of each anchor is obtained by weighted calculation based on the anchor's historical sales record, number of fans, and live broadcast room turnover in the anchor information. The calculation formula is:

[0030] E i =aS j +bF j +cW j

[0031]

[0032] Among them, S j is the sales volume of all products in the first preset time period of anchor j, F j is the number of fans of anchor j, W j is the live streaming volume of anchor j in the first preset time period, P(A j ) is the overall selling probability of anchor j, n is the number of anchors to be assigned, and a, b, c are weight values.

[0033] Furthermore, the overall sales probability of each commodity is calculated by using the historical sales records of each type of commodity, and the calculation formula is:

[0034] P(B i )=x i / Y i

[0035] Among them, x i is the historical sales quantity of product type i, Y i is the total number of historical items of type i, P(B i ) is the overall sales probability of product type i.

[0036] Furthermore, the three probabilities obtained by the above calculation are used to calculate the probability of each anchor among the anchors to be assigned to each type of goods by using the Bayesian algorithm, which specifically includes: according to the three probabilities obtained by the above calculation, the initial probability of each anchor being assigned to each type of goods is calculated by using the Bayesian algorithm, and the probability of each anchor is normalized to obtain the final probability. The calculation formula is:

[0037]

[0038] Among them, P(A j |B i ) is the probability that anchor j is assigned to product type i,

[0039] Furthermore, according to the probability of each anchor being assigned to each type of product, a random number interval corresponding to the probability is allocated to each anchor. Specifically, the probability of anchor j being assigned to type i product is P(A j |B i ), then the random number interval of anchor j for product type i is The method generates a corresponding random number, and the random number value range is [0,100].

[0040] Another aspect of the present invention provides a host commodity allocation system based on Bayesian, comprising:

[0041] A data acquisition module is used to obtain information about products to be sold and information about anchors to be assigned;

[0042] The anchor allocation probability calculation module is used to calculate the probability of each anchor among the anchors to be allocated being allocated to each type of product;

[0043] The product allocation module allocates each product using random numbers based on the probability of each anchor being assigned to each type of product.

[0044] Compared with the prior art, the present invention has the following advantages:

[0045] (1) The present invention combines historical sales data and calculates the distribution probability of the anchor according to the Bayesian algorithm, thereby realizing the automation of commodity distribution, solving the problems of low efficiency, inability to combine historical sales records, and subjectivity and bias of the distribution personnel in the prior art.

[0046] (2) The present invention solves the problem of low commodity utilization rate by setting a basic initial value for the probability of each anchor selling a commodity. Some commodities may not be displayed by suitable anchors due to improper allocation, resulting in poor sales. Popular commodities may be concentratedly allocated to a few anchors, and other anchors cannot make full use of these resources, affecting overall sales performance.

[0047] (3) The present invention provides a method that integrates the above method into a system through computer technology to realize the automation of commodity distribution, thereby solving the problems of low efficiency and increased communication costs in live commodity distribution, and possible communication barriers between different teams (such as marketing, sales, customer service, etc.), resulting in poor information transmission and affecting distribution decisions. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 The figure is a flow chart of the method of the present invention. DETAILED DESCRIPTION

[0049] 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 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 should fall within the scope of protection of the present invention.

[0050] It is known that there are x types of goods to be sold tomorrow, and there are a total of y goods. There are m anchors who will broadcast live tomorrow. Assuming that all goods can be sold tomorrow, how to allocate the goods to suitable anchors? To this end, this embodiment provides a Bayesian anchor product allocation method.

[0051] Bayes' theorem is a fundamental theorem in probability theory that describes how to update your understanding of the probability of an event using existing information. The theorem is named after the 18th-century mathematician Thomas Bayes. Bayes' theorem is widely used in statistics, machine learning, medicine, finance and many other fields.

[0052] The formula for Bayes' theorem is:

[0053] The mathematical expression of Bayes' theorem is as follows:

[0054]

[0055] in:

[0056] P(A|B): The conditional probability of event A occurring given event B.

[0057] P(B|A): The conditional probability of event B occurring given event A.

[0058] P(A): The prior probability of event A occurring.

[0059] P(B): The total probability of event B occurring.

[0060] Based on the above Bayesian theorem, Figure 1 As shown, the specific process of this embodiment is as follows:

[0061] Step 1: Get the information of the products to be sold tomorrow and the information of the anchors to be assigned. The product information includes the product type, quantity and historical sales records of each type of product. The anchor information includes the number of anchors to be assigned, the anchor's historical sales record, the number of fans, and the flow of the live broadcast room. The products are classified according to the brand, price and quality of the products.

[0062] Step 2: Calculate the probability of each anchor selling each type of product among the anchors to be assigned based on the product information and the anchor's historical sales records, and use this probability as P(B|A) in Bayesian. The specific process includes:

[0063] Obtain information on products to be sold and sales records of each anchor to be assigned;

[0064] Extract the sales records of each anchor in the past month, and calculate the probability of the anchor selling each type of goods to be sold in the past month based on the sales records in this period;

[0065] Extract the sales records of each anchor in the past six months to the previous month, and calculate the probability of the anchor selling each type of goods to be sold in the past six months to the previous month based on the sales records in this period;

[0066] The probability of the past month and the probability of the past six months to the previous month are weighted and summed to obtain the probability P(B i |A j ),

[0067] Determine the probability P(B) of each anchor selling each type of product i |A j ) is greater than the first preset value. If less than, the value is set to the first preset value; if greater than or equal to, it remains unchanged.

[0068] The calculation formula is:

[0069] P 1 (B i |A j )=x 1ij / Y 1ij

[0070] P 2 (B i |A j )=x 2ij / Y 2ij

[0071] P(B i |A j )=αP 1 (B i |A j )+βP 2 (B i |A j )

[0072] Among them, x 1ij is the number of products of type i sold by anchor j in the first preset time period, x 2ijis the number of products of type i sold by anchor j in the second preset time period, Y 1ij is the number of commodities of type i assigned to anchor j in the first preset time period, Y 2ij is the number of commodities of type i assigned to anchor j in the second preset time period, P 1 (B i |A j ) is the probability that anchor j sells product of type i in the first preset time period, P 2 (B i |A j ) is the probability that anchor j sells product of type i in the second preset time period, and α and β are weight values.

[0073] And α is greater than β, because the sales data of the past month can better reflect the real sales ability of the anchor, but the sales data of the anchor in the past six months should also be taken into account. Through comprehensive consideration, the sales ability of the anchor is calculated.

[0074] Step 3: Based on the anchor’s historical sales record, number of fans, and live broadcast room turnover in the anchor information, a weighted calculation is performed to obtain the overall sales probability of each anchor, and this probability is used as the prior probability P(A) of event A in the Bayesian formula.

[0075] The calculation formula is:

[0076] E i =aS j +bF j +cW j

[0077]

[0078] Among them, S j is the sales volume of all products in the first preset time period of anchor j, F j is the number of fans of anchor j, W j is the live streaming volume of anchor j in the first preset time period, P(A j ) is the overall selling probability of anchor j, n is the number of anchors to be assigned, and a, b, c are weight values.

[0079] Because the prior probability P(A) of event A in the Bayesian formula refers to our initial estimate of the probability of an event or hypothesis before considering any new evidence, the total sales, total live broadcast flow, and number of fans of the anchor in the past month are taken into comprehensive consideration, and this data is used as the overall sales probability of the anchor, as well as the prior probability of the Bayesian formula.

[0080] Step 4: Calculate the overall sales probability of each product through the historical sales records of each type of product, and use this probability as the prior probability P(B) of event B in the Bayesian formula.

[0081] The calculation formula is:

[0082] P(B i )=x i / Yi i

[0083] Among them, x i is the historical sales quantity of product type i, Y i is the total number of historical items of type i, P(B i ) is the overall sales probability of product type i.

[0084] Step 5: Based on the three probabilities calculated above, the probability of each anchor being assigned to each type of product is calculated by the Bayesian algorithm, specifically including: based on the three probabilities calculated above, the initial probability of each anchor being assigned to each type of product is calculated by the Bayesian algorithm, and the probability of each anchor is normalized to obtain the final probability. The calculation formula is:

[0085]

[0086] Among them, P(A j |B i ) is the probability that anchor j is assigned to product type i,

[0087] Step 6: According to the probability of each anchor being assigned to each type of product, a random number interval corresponding to its probability is assigned to each anchor. Specifically, the probability of anchor j being assigned to type i product is P(A j |B i ), then the random number interval of anchor j for product type i is

[0088] For each product, a corresponding random number is generated, and the random number interval in which the random number falls is determined, and the product is assigned to the anchor corresponding to the random number interval. The generated random number ranges from [0,100].

[0089] This embodiment also provides a Bayesian-based anchor commodity allocation system, including:

[0090] A data acquisition module is used to obtain information about products to be sold and information about anchors to be assigned;

[0091] The anchor allocation probability calculation module is used to calculate the probability of each anchor among the anchors to be allocated being allocated to each type of product;

[0092] The product allocation module allocates each product using random numbers based on the probability of each anchor being assigned to each type of product.

[0093] If the above functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc., which can store program code.

[0094] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed by the present invention, and these modifications or replacements 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 method for allocating anchor products based on Bayesian, characterized in that: The following steps are involved: Obtain information on goods to be sold and information on anchors to be assigned, the goods information including the type of goods, quantity of goods and historical sales records of each type of goods, and the anchor information including the number of anchors to be assigned, anchors' historical sales records, number of fans, and live broadcast room turnover; Calculate the probability of each anchor selling each type of product among the anchors to be assigned based on the product information and the anchor's historical sales records; The overall sales probability of each anchor is obtained through weighted calculation based on the anchor's historical sales record, number of fans, and live broadcast room turnover in the anchor's information; The overall sales probability of each product category is calculated through the historical sales records of each type of product; Based on the three probabilities calculated above, the probability of each anchor being assigned to each type of product is calculated using the Bayesian algorithm; According to the probability of each anchor being assigned to each type of product, a random number interval corresponding to the probability is assigned to each anchor; For each product, a corresponding random number is generated, and it is determined in which random number interval the random number falls, and the product is allocated to the anchor corresponding to the random number interval. If the allocated product of the corresponding anchor is equal to the second preset value, a random number is re-generated for the product and reallocated.

2. A method for allocating anchor products based on Bayesian analysis according to claim 1, characterized in that: The commodities are classified according to commodity category, brand, price and quality.

3. A method for allocating anchor products based on Bayesian analysis according to claim 1, characterized in that: The method of calculating the probability of each anchor among the anchors to be assigned selling each type of goods according to the anchors' historical sales records specifically includes the following steps: Obtain information on products to be sold and sales records of each anchor to be assigned; Extract the sales records of each anchor in the first preset time period, and calculate the probability of the anchor selling each type of goods to be sold in the first preset time period according to the sales records in the period; Extracting the sales records of each anchor in the second preset time period, and calculating the probability of the anchor selling each type of goods to be sold in the second preset time period according to the sales records in the period; The probability of the first preset time period and the probability of the second preset time period are weighted and summed to obtain the probability P(B i |A j ), Determine the probability P(B) of each anchor selling each type of product i |A j ) is greater than the first preset value. If less than, the value is set to the first preset value; if greater than or equal to, it remains unchanged.

4. A method for allocating anchor products based on Bayesian analysis according to claim 3, characterized in that: The probability of each anchor selling each type of goods among the anchors to be assigned is calculated based on the anchor's historical sales records, and the calculation formula is: P1(B i |A j )=x 1ij / Y 1ij P2(B i |A j )=x 2ij / Y 2ij P(B i |A j )6αP1(B i |A j )+βP2(B i |A j ) Among them, x 1ij is the number of products of type i sold by anchor j in the first preset time period, x 2ij is the number of products of type i sold by anchor j in the second preset time period, Y 1ij is the number of commodities of type i assigned to anchor j in the first preset time period, Y 2ij is the number of commodities of type i allocated to anchor j in the second preset time period, P1(B i |A j ) is the probability that anchor j sells product of type i in the first preset time period, P2(B i |A j ) is the probability that anchor j sells product of type i in the second preset time period, and α and β are weight values.

5. A Bayesian anchor product allocation method according to claim 3 or 4, characterized in that: The first preset time period is nearly one month, the second preset time period is nearly half a year to nearly one month, and α is greater than β.

6. A method for allocating anchor products based on Bayesian analysis according to claim 1, characterized in that: The overall sales probability of each anchor is obtained by weighted calculation based on the anchor's historical sales record, number of fans, and live broadcast room flow in the anchor information. The calculation formula is: E i =aS j +bF j +cW j Among them, S j is the sales volume of all products in the first preset time period of anchor j, F j is the number of fans of anchor j, W j is the live streaming volume of anchor j in the first preset time period, P(A j ) is the overall selling probability of anchor j, n is the number of anchors to be assigned, and a, b, c are weight values.

7. A method for allocating anchor products based on Bayesian analysis according to claim 1, characterized in that: The overall sales probability of each commodity is calculated by the historical sales records of each type of commodity, and the calculation formula is: P(B i )=x i / Y i Among them, x i is the historical sales quantity of product type i, Y i is the total number of historical items of type i, P(B i ) is the overall sales probability of product type i.

8. A method for allocating anchor products based on Bayesian analysis according to claim 1, characterized in that: The three probabilities obtained by the above calculation are used to calculate the probability of each anchor among the anchors to be assigned to each type of goods by using the Bayesian algorithm, which specifically includes: according to the three probabilities obtained by the above calculation, the initial probability of each anchor being assigned to each type of goods is calculated by using the Bayesian algorithm, and the probability of each anchor is normalized to obtain the final probability. The calculation formula is: Among them, P(A j |B i ) is the probability that anchor j is assigned to product type i, 9. A method for allocating anchor products based on Bayesian analysis according to claim 1, characterized in that: According to the probability of each anchor being assigned to each type of product, a random number interval corresponding to the probability is allocated to each anchor. Specifically, the probability of anchor j being assigned to type i product is P(A j |B i ), then the random number interval of anchor j for product type i is The method generates a corresponding random number, and the random number value range is [0, 100].

10. A Bayesian-based anchor product allocation system, characterized in that: include: A data acquisition module is used to obtain information about the products to be sold and the anchor information to be assigned; The anchor allocation probability calculation module is used to calculate the probability of each anchor among the anchors to be allocated being allocated to each type of product; The product allocation module allocates each product using random numbers based on the probability of each anchor being assigned to each type of product.

Images