CN104715409A - Method and system for electronic commerce user purchasing power classification - Google Patents
Method and system for electronic commerce user purchasing power classification Download PDFInfo
- Publication number
- CN104715409A CN104715409A CN201510126137.6A CN201510126137A CN104715409A CN 104715409 A CN104715409 A CN 104715409A CN 201510126137 A CN201510126137 A CN 201510126137A CN 104715409 A CN104715409 A CN 104715409A
- Authority
- CN
- China
- Prior art keywords
- class
- user
- purchasing power
- commodity
- center
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/06—Buying, selling or leasing transactions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Finance (AREA)
- Accounting & Taxation (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Databases & Information Systems (AREA)
- Data Mining & Analysis (AREA)
- Development Economics (AREA)
- Economics (AREA)
- Marketing (AREA)
- Strategic Management (AREA)
- General Business, Economics & Management (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a method and system for electronic commerce user purchasing power classification. The method includes the steps that based on price and sale volume distribution, the level of commodities of the same type is determined to be the high level, the level of other commodities of the same type is determined to be the non-high level, and the commodities at the non-high level are divided into x levels from high to low sequentially according to the price from high to low, wherein x is a preset natural number larger than or equal to 1; the ratio of the commodities at each level in overall commodities purchased by each user is calculated, the ratios are vectorized based on all the users, purchasing power vectors of all the users are acquired and are x+1 dimension vectors, and each dimension corresponds to one level; the purchasing power vectors are clustered, x+1 point clusters related to the purchasing power vectors are acquired, each point cluster corresponds to one level, and the level corresponding to the point cluster where the corresponding purchasing power vector of each user is located serves as the purchasing power level of the user. By the adoption of the method and system, user purchasing power classification is more accurate.
Description
Technical field
The present invention designs ecommerce correlative technology field, particularly a kind of e-commerce user purchasing power sorting technique and system.
Background technology
Along with the develop rapidly of electric firm industry, the personalized shopping need meeting user also becomes extremely urgent.Browsing user in the process of shopping recommends rational commodity greatly will promote Consumer's Experience to user.But the behind of the shopping commending system of property needs a large amount of user tag to support one by one.Wherein, the purchasing power label of user is essential.For the user that purchasing power is high, when selecting same class commodity, what often buy is the commodity that quality and price are all higher, and the user that such as purchasing power is high wants to buy a mobile phone, and he can buy the high-end handsets in apple or Samsung brand; And for the low user of purchasing power, such as he wants to buy an earphone, he can buy tens yuan or 20 yuan can meet the low-end product generally used.Thus, the definition for the purchasing power of a user is, user buy same class commodity time, the height of paying ability.
Purchasing power for electric business's field user is distinguished, and existing technology based on the method for user to high, medium and low commodity purchasing number of times accounting, is divided into high, normal, basic three class user mostly.Specific practice is: to each end level category, and the commodity that price segment is in the highest 20% are defined as high-end commodity, and the commodity that price segment is in minimum 20% are low side commodity, and the commodity of middle 60% are middle-end commodity.Then calculate the number of times accounting of the high-end commodity of purchase of each user, buy the number of times accounting of middle-end commodity, buy the number of times accounting of low side commodity.Finally see that user is maximum in the purchase number of times accounting of which grade commodity, then this user is divided in the middle of this purchase class user group.Finally obtain the high, medium and low Three Estate of purchasing power.
The shortcoming of prior art mainly contains four aspects:
1) when distinguishing price segment and being high, medium and low, the price segment of the commodity used, but in actual conditions, the purchase situation of the high price section of a lot of category is very sparse, even there is no sales volume, thus such clean cut rule be easy to cause result set to distribute unbalanced.
2) calculate user when each class buys commodity accounting, calculate and buy frequency accounting, do not add the price factor of commodity itself, cause accuracy rate to reduce.Such as, although a user has bought the commodity of the top grade of certain category a lot, but the price of this category is very low (diaper, household articles etc.) inherently, so he is bought in the expensive customer groups such as high-grade mobile phone, computer with other, nature can be unfair.
3) when obtaining the high, medium and low three kinds of commodity purchasing accountings of user, directly with accounting maximum determine final purchasing power grade, can accuracy rate be reduced.Such as, a user A is (0.8,0.2,0) in the accounting of purchase high-, middle-and low-end commodity respectively, and user B is (0.4,0.3,0.3) in the accounting of purchase high-, middle-and low-end commodity respectively.According to existing determination methods, user A and B is the user that purchasing power is high, and actual observation, we can be very easy to find, and user B similarly is not high-end user.This is because the purchase accounting of user also also should play a role the purchasing power that user is final in distribution, and simple maximum accounting rule judgment, not science.
4) grade of existing purchasing power is generally divided into high, normal, basic three ranks, classifies less, uses underaction.
Summary of the invention
Based on this, be necessary, for the inaccurate technical matters of the classification of prior art to user's purchasing power, to provide a kind of e-commerce user purchasing power sorting technique and system.
A kind of e-commerce user purchasing power sorting technique, comprising:
Commodity class determining step, comprise: based on price and sales volume distribution, the commodity in same category are determined that class is high-grade commodity, the class of other commodity in same category is defined as non-top grade, be that the commodity of non-top grade are divided into x class from high to low from high to low successively according to price by class, wherein, x be default be more than or equal to 1 natural number;
User buys accounting calculation procedure, comprise: calculate the accounting that each user buys each class commodity, described accounting is carried out vectorization based on each user, obtain the purchasing power vector of each user, described purchasing power vector is x+1 dimensional vector, and every one dimension is corresponding with a class;
User's classifying step, comprise: cluster computing is carried out to described purchasing power vector, obtain x+1 the point bunch about purchasing power vector, each point bunch corresponding respectively class, using the purchasing power class of purchasing power corresponding for user vector point bunch corresponding class as user.
A kind of e-commerce user purchasing power categorizing system, comprising:
Commodity class determination module, for: based on price and sales volume distribution, the commodity in same category are determined that class is high-grade commodity, the class of other commodity in same category is defined as non-top grade, be that the commodity of non-top grade are divided into x class from high to low from high to low successively according to price by class, wherein, x be default be more than or equal to 1 natural number;
User buys accounting computing module, for: calculate the accounting that each user buys each class commodity, described accounting is carried out vectorization based on each user, obtain the purchasing power vector of each user;
User's classifying module, for: cluster computing is carried out to described purchasing power vector, obtain x+1 the point bunch about purchasing power vector, each point bunch corresponding respectively class, using the purchasing power class of purchasing power corresponding for user vector point bunch corresponding class as user.
The present invention, by delimiting the intelligence of commodity class, makes the delimitation of commodity class more reasonable, and determines user's purchasing power based on the commodity class that intelligence divides, and classify based on user's purchasing power, it is more accurate to make the classification of user's purchasing power.
Accompanying drawing explanation
Fig. 1 is the workflow diagram of a kind of e-commerce user purchasing power of the present invention sorting technique;
Fig. 2 is the workflow diagram carrying out high-grade goods division;
Fig. 3 is the workflow diagram that the present invention carries out cluster computing;
Fig. 4 is the construction module figure of a kind of e-commerce user purchasing power of the present invention categorizing system.
Embodiment
Below in conjunction with the drawings and specific embodiments, the present invention will be further described in detail.
Be illustrated in figure 1 the workflow diagram of a kind of e-commerce user purchasing power of the present invention sorting technique, comprise:
Step S101, comprise: based on price and sales volume distribution, the commodity in same category are determined that class is high-grade commodity, the class of other commodity in same category is defined as non-top grade, be that the commodity of non-top grade are divided into x class from high to low from high to low successively according to price by class, wherein, x be default be more than or equal to 1 natural number;
Step S102, comprising: calculate the accounting that each user buys each class commodity, described accounting is carried out vectorization based on each user, and obtain the purchasing power vector of each user, described purchasing power vector is x+1 dimensional vector, and every one dimension is corresponding with a class;
Step S103, comprise: cluster computing is carried out to described purchasing power vector, obtain x+1 the point bunch about purchasing power vector, each point bunch corresponding respectively class, using the purchasing power class of purchasing power corresponding for user vector point bunch corresponding class as user.
The present invention proposes a kind of method more reasonably dividing user's purchasing power grade.The present invention be not simply only according to commodity price to distinguish commodity class, but carry out comprehensive assessment commodity class according to commodity price and corresponding Sales Volume of Commodity, because commodity class finally determines the purchasing power class of user, be exactly therefore the comprehensive assessment to user's purchasing power grade to the comprehensive assessment of commodity class.By the division methods automatically regulated based on price and sales volume distribution high-grade goods, the level evaluation of correspondingly user's purchasing power is made to also achieve automatic adjustment, make the classification of user's purchasing power more accurate, thus improve the experience of user in website widely.
Wherein in an embodiment, described step S101, specifically comprises: the class of the commodity of a% before price segment in same category is defined as top grade, and a adopts and confirms with the following method:
More than three or three spans are selected to be treat selective value between 0 ~ 100, maximum treats that selective value is as MAXIMUM SELECTION value, minimumly treat that selective value is minimum selective value, other treat that selective value is middle selective value, m is made to be the price statistics value of the front y% of price in same category, n is made to be the price statistics value having the front y% of price in the commodity of sales volume in same category in nearest preset time period, if m is greater than n and exceedes preset first threshold value, then a is selected to be MAXIMUM SELECTION value, if n is greater than m and exceedes default Second Threshold, then selection a is the minimum value in minimum selective value, other situations, selection a is one in middle selective value, wherein, y is less than MAXIMUM SELECTION value and is greater than minimum selective value.
First, to mark all commodity, because the purchasing power of user finally will be divided into X+1 grade by purchasing power model, preferably, X+1 is 5, then the purchasing power of user be divided into height, higher, in, in low, low five grades, so in the process marked commodity, all commodity also will be also designated as X class by us, are preferably top grade, higher gear, middle-grade, low and middle-grade, low-grade five class.Because the clean cut method mentioned in the introduction can make the high-grade goods of a lot of category very sparse.So divide high-grade goods, the present embodiment regulates division number percent automatically according to the price of each category commodity and sales volume distribution situation.For a category, such as three grades of categories, the commodity choosing a% before being in price segment are high-grade goods, and wherein, a is preferably three, can select 5,10 or 20, and namely a% can be 5%, 10 or 20%.The concrete value of a needs first to calculate two indices m, n, wherein m is the price statistics value that this category price is positioned at front y%, n is price statistics value (the not deduplication having the commodity price of sales volume to be positioned at front y% in this category nearly a period of time, if commodity are bought repeatedly, also count).Y is preferably 10%, the mxm. after the price statistics value of front y% is preferably and is removed by the commodity of front y% in surplus commodities.Like this, if m>>n, before this category price is described, 10% Sales Volume of Commodity is bad, and the threshold value of high-end commodity should improve, i.e. a=20%; If n>>m, before the price of this category is described, 10% Sales Volume of Commodity is fine, should get less, reduces threshold value, i.e. a=5%; Other situation a=10%.Just accomplish that the judgment threshold of high-end commodity can carry out self-regulation according to the actual conditions of selling thus.The value of m should between the centre of a value, but not necessarily intermediate value.Here value all can change.Concrete process flow diagram is as shown in Figure 2:
Step S201, calculates the price statistics value m that this category price is positioned at before 10 10%;
Step S202, calculates in this category nearly a period of time and has the commodity price of sales volume to be positioned at the price statistics value of front 10%;
Step S203, if m>>n, namely m is greater than n and exceedes first threshold, and first threshold can get larger scope, then an a%=20%, otherwise performs step S204;
Step S204, if n>>m, and n is greater than m and exceedes Second Threshold, and Second Threshold can get larger scope, then an a%=5%, otherwise a%=10.
For high-end commodity, supplementing of some systematicness can also be added, the commodity that such as luxurious category, high-end nonessential product (as smart machine etc.) or cargo price are very high.
The class of remaining commodity is non-high-end, be divided into X class (X be greater than 1 natural number), preferably, can be divided into higher, in, in low, low four class, wherein, low-grade goods can get the commodity of rear 20% of price segment, higher shelves, middle-grade, low and middle-grade be front 1/3 after removing top grade and low grade respectively, middle 1/3 and below 1/3.
Wherein in an embodiment, described step S102, specifically comprises:
For each user, calculate this user each class the amount of placing an order with dutiable value is taken the logarithm after product as the purchase volume of this user at this class, calculate the purchase volume summation of all class of this user, calculate each user accounts for the purchase volume summation of this user ratio in the purchase volume of each class buys each class commodity accounting as user;
For each user, the vector that the accounting each user being bought each class commodity is tieed up as an x+1, obtains the purchasing power vector of each user.
Calculate the accounting of each user at each class commodity purchasing, the vector of an X+1 dimension can be obtained, be preferably five dimensional vector (x
1, x
2, x
3, x
4, x
5), be called purchasing power vector.Wherein x
irepresent the accounting that this user is the commodity purchasing of i-th grade at class.At each class accounting x of calculating
itime, calculate the accounting buying commodity frequency j (i.e. the amount of placing an order) at each class, but calculate user each class buy commodity the frequency × ln (price) (namely the amount of placing an order and price take the logarithm after product) accounting.Such as certain user has bought two commodity A, B altogether in high-grade goods, the number of times bought respectively is k1, k2, and price is respectively p1, p2, then user buys as k1 × ln (p1)+k2 × ln (p2) at top grade, wherein, ln represents the logarithm to numerical value in bracket.Calculate this user after the purchase volume of each class according to this method, then the purchase volume obtaining each class accounts for the accounting of total purchase volume.Here the price adding commodity gets log as weight, thus solves second shortcoming of the prior art mentioned in background technology.Add the price factor of commodity itself, even if user has bought a lot of relatively high-grade commodity in low price category, also can be low because of price weight, adjusted.
As shown in Figure 3, wherein in an embodiment, described step S103, specifically comprises:
Step S301, comprising: from the purchasing power vector of all users, the purchasing power vector of a random selecting x+1 user is as the center, execution step S302;
Step S302, comprising: calculate the Euclidean distance of remaining all purchasing power vector to x+1 center respectively, is incorporated into by each purchasing power vector in the point minimum with the Euclidean distance at center bunch respectively, execution step S303;
Step S303, comprising: to all purchasing power vector calculation in x+1 point bunch about the arithmetical mean of each dimension as the center of this point bunch, perform step S304;
Step S304, comprising: the European cluster with each center is recalculated at the center that the purchasing power of all users vector obtains according to step S304, is incorporated into respectively by each purchasing power vector in the point minimum with the Euclidean distance at center bunch, performs step S305;
Step S305, comprising: if the center of each point bunch no longer changes, then perform step S306, otherwise perform step S303;
Step S306, according to the class corresponding to the center calculation corresponding point bunch of x+1 point bunch.
The present embodiment is not that the quantity accounting size buying certain class commodity according to user simply decides user's purchasing power, but carries out cluster analysis to divide user's purchasing power grade to the data of all users.
After carrying out cluster by the purchasing power vector of clustering method to user, obtain X+1 point bunch, be preferably five points bunch.Through observation, obviously can find out that each point bunch represents a kind of power of buying crowd of class.The user of the purchasing power vector included by each point bunch is sorted out respectively and is saved in database, thus the classification of completing user purchasing power.
Preferably, described step S306, specifically comprises:
Obtain x+1 point bunch center, perform following sub-step successively according to class sequence:
High-grade chooser step, comprising: select current class to be determined for high-grade;
Class determination sub-step, comprising: never to determine in the point bunch of class in the heart, and the current of the point corresponding to center maximum for the dimension corresponding with current class to be determined bunch is defined as current class to be determined;
Other, when selecting sub-step, comprising: if do not determine the point bunch of class in addition, then select next order class of current class to be determined as current class to be determined, perform class determination sub-step, otherwise terminate.
Wherein in an embodiment, also comprise:
Classification recommendation step, comprising: when receiving the visit information of user, obtains the purchasing power class of user, recommends the commodity with corresponding class according to purchasing power class to user.
When after the classification completing user's purchasing power, when old user's access websites, just can be known the purchasing power situation of this user by database, recommend the commodity of corresponding price according to the class of user's purchasing power.The user that such as purchasing power class is high is browsing mobile phone, just recommends his high-end, smart mobile phone; Contrary, the user that a purchasing power grade is very low, then recommend the mobile phone that his some prices are low, practical.
Be illustrated in figure 4 the construction module figure of a kind of e-commerce user purchasing power of the present invention categorizing system, comprise:
Commodity class determination module 401, for: based on price and sales volume distribution, the commodity in same category are determined that class is high-grade commodity, the class of other commodity in same category is defined as non-top grade, be that the commodity of non-top grade are divided into x class from high to low from high to low successively according to price by class, wherein, x be default be more than or equal to 1 natural number;
User buys accounting computing module 402, for: calculate the accounting that each user buys each class commodity, described accounting is carried out vectorization based on each user, obtain the purchasing power vector of each user, described purchasing power vector is x+1 dimensional vector, and every one dimension is corresponding with a class;
User's classifying module 403, for: cluster computing is carried out to described purchasing power vector, obtain x+1 the point bunch about purchasing power vector, each point bunch corresponding respectively class, using the purchasing power class of purchasing power corresponding for user vector point bunch corresponding class as user.
Wherein in an embodiment, described commodity class determination module, specifically for: the class of the commodity of a% before price segment in same category is defined as top grade, and a adopts and confirms with the following method:
More than three or three spans are selected to be treat selective value between 0 ~ 100, maximum treats that selective value is as MAXIMUM SELECTION value, minimumly treat that selective value is minimum selective value, other treat that selective value is middle selective value, m is made to be the price statistics value of the front y% of price in same category, n is made to be the price statistics value having the front y% of price in the commodity of sales volume in same category in nearest preset time period, if m is greater than n and exceedes preset first threshold value, then a is selected to be MAXIMUM SELECTION value, if n is greater than m and exceedes default Second Threshold, then selection a is the minimum value in minimum selective value, other situations, selection a is one in middle selective value, wherein, y is less than MAXIMUM SELECTION value and is greater than minimum selective value.
Wherein in an embodiment, described user buys accounting computing module, specifically for:
For each user, calculate this user each class the amount of placing an order with dutiable value is taken the logarithm after product as the purchase volume of this user at this class, calculate the purchase volume summation of all class of this user, calculate each user accounts for the purchase volume summation of this user ratio in the purchase volume of each class buys each class commodity accounting as user;
For each user, the vector that the accounting each user being bought each class commodity is tieed up as an x+1, obtains the purchasing power vector of each user.
Wherein in an embodiment, described user's classifying module, specifically for:
Center initialization submodule, for: from the purchasing power vector of all users, the purchasing power vector of a random selecting x+1 user is as the center, and sub-step is initially sorted out in execution;
Initial classification submodule, for: calculate the Euclidean distance of remaining all purchasing power vector to x+1 center respectively, each purchasing power vector is incorporated into respectively in the point minimum with the Euclidean distance at center bunch, implementation center's renewal sub-step;
Center upgrades submodule, for: to all purchasing power vector calculation in x+1 point bunch about the arithmetical mean of each dimension as the center of this point bunch, perform and sort out renewal sub-step;
Sort out and upgrade submodule, for: the European cluster with each center is recalculated at the center that the purchasing power of all users vector obtains according to classification renewal sub-step, each purchasing power vector is incorporated into respectively in the point minimum with the Euclidean distance at center bunch, perform convergence and judge sub-step;
Convergence judges submodule, for: if the center of each point bunch no longer changes, then perform class determination submodule, otherwise implementation center's renewal submodule;
Class determination submodule, for: according to the class corresponding to the center calculation corresponding point bunch of x+1 point bunch.
Wherein in an embodiment, described class determination submodule, specifically for:
Obtain x+1 point bunch center, perform following submodule successively according to class sequence:
High-grade chooser module, for: select current class to be determined for high-grade;
Class determination submodule, for: never to determine in the point bunch of class in the heart, the current of the point corresponding to center maximum for the dimension corresponding with current class to be determined bunch is defined as current class to be determined;
Other are when selecting submodule, for: if do not determine the point bunch of class in addition, then select next order class of current class to be determined as current class to be determined, perform class determination sub-step, otherwise terminate.
Wherein in an embodiment, also comprise:
Classification recommending module, for: when receiving the visit information of user, obtaining the purchasing power class of user, recommending the commodity with corresponding class according to purchasing power class to user.
The above embodiment only have expressed several embodiment of the present invention, and it describes comparatively concrete and detailed, but therefore can not be interpreted as the restriction to the scope of the claims of the present invention.It should be pointed out that for the person of ordinary skill of the art, without departing from the inventive concept of the premise, can also make some distortion and improvement, these all belong to protection scope of the present invention.Therefore, the protection domain of patent of the present invention should be as the criterion with claims.
Claims (12)
1. an e-commerce user purchasing power sorting technique, is characterized in that, comprising:
Commodity class determining step, comprise: based on price and sales volume distribution, the commodity in same category are determined that class is high-grade commodity, the class of other commodity in same category is defined as non-top grade, be that the commodity of non-top grade are divided into x class from high to low from high to low successively according to price by class, wherein, x be default be more than or equal to 1 natural number;
User buys accounting calculation procedure, comprise: calculate the accounting that each user buys each class commodity, described accounting is carried out vectorization based on each user, obtain the purchasing power vector of each user, described purchasing power vector is x+1 dimensional vector, and every one dimension is corresponding with a class;
User's classifying step, comprise: cluster computing is carried out to described purchasing power vector, obtain x+1 the point bunch about purchasing power vector, each point bunch corresponding respectively class, using the purchasing power class of purchasing power corresponding for user vector point bunch corresponding class as user.
2. e-commerce user purchasing power sorting technique according to claim 1, is characterized in that, described commodity class determining step, specifically comprises: the class of the commodity of a% before price segment in same category is defined as top grade, and a adopts and confirms with the following method:
More than three or three spans are selected to be treat selective value between 0 ~ 100, maximum treats that selective value is as MAXIMUM SELECTION value, minimumly treat that selective value is minimum selective value, other treat that selective value is middle selective value, m is made to be the price statistics value of the front y% of price in same category, n is made to be the price statistics value having the front y% of price in the commodity of sales volume in same category in nearest preset time period, if m is greater than n and exceedes preset first threshold value, then a is selected to be MAXIMUM SELECTION value, if n is greater than m and exceedes default Second Threshold, then selection a is the minimum value in minimum selective value, other situations, selection a is one in middle selective value, wherein, y is less than MAXIMUM SELECTION value and is greater than minimum selective value.
3. e-commerce user purchasing power sorting technique according to claim 1, it is characterized in that, described user buys accounting calculation procedure, specifically comprises:
For each user, calculate this user each class the amount of placing an order with dutiable value is taken the logarithm after product as the purchase volume of this user at this class, calculate the purchase volume summation of all class of this user, calculate each user accounts for the purchase volume summation of this user ratio in the purchase volume of each class buys each class commodity accounting as user;
For each user, the vector that the accounting each user being bought each class commodity is tieed up as an x+1, obtains the purchasing power vector of each user.
4. e-commerce user purchasing power sorting technique according to claim 1, is characterized in that, described user's classifying step, specifically comprises:
Center initialization sub-step, comprising: from the purchasing power vector of all users, the purchasing power vector of a random selecting x+1 user is as the center, and sub-step is initially sorted out in execution;
Initial classification sub-step, comprising: calculate the Euclidean distance of remaining all purchasing power vector to x+1 center respectively, is incorporated into by each purchasing power vector in the point minimum with the Euclidean distance at center bunch respectively, implementation center's renewal sub-step;
Center upgrades sub-step, comprising: to all purchasing power vector calculation in x+1 point bunch about the arithmetical mean of each dimension as the center of this point bunch, perform and sort out renewal sub-step;
Sort out and upgrade sub-step, comprise: the European cluster with each center is recalculated at the center that the purchasing power of all users vector obtains according to classification renewal sub-step, each purchasing power vector is incorporated into respectively in the point minimum with the Euclidean distance at center bunch, perform convergence and judge sub-step;
Convergence judges sub-step, comprising: if the center of each point bunch no longer changes, then perform class determination sub-step, otherwise implementation center's renewal sub-step;
Class determination sub-step, comprising: according to the class corresponding to the center calculation corresponding point bunch of x+1 point bunch.
5. e-commerce user purchasing power sorting technique according to claim 4, is characterized in that, described class determination sub-step, specifically comprises:
Obtain x+1 point bunch center, perform following sub-step successively according to class sequence:
High-grade chooser step, comprising: select current class to be determined for high-grade;
Class determination sub-step, comprising: never to determine in the point bunch of class in the heart, and the current of the point corresponding to center maximum for the dimension corresponding with current class to be determined bunch is defined as current class to be determined;
Other, when selecting sub-step, comprising: if do not determine the point bunch of class in addition, then select next order class of current class to be determined as current class to be determined, perform class determination sub-step, otherwise terminate.
6. e-commerce user purchasing power sorting technique according to claim 1, is characterized in that, also comprise:
Classification recommendation step, comprising: when receiving the visit information of user, obtains the purchasing power class of user, recommends the commodity with corresponding class according to purchasing power class to user.
7. an e-commerce user purchasing power categorizing system, is characterized in that, comprising:
Commodity class determination module, for: based on price and sales volume distribution, the commodity in same category are determined that class is high-grade commodity, the class of other commodity in same category is defined as non-top grade, be that the commodity of non-top grade are divided into x class from high to low from high to low successively according to price by class, wherein, x be default be more than or equal to 1 natural number;
User buys accounting computing module, for: calculate the accounting that each user buys each class commodity, described accounting is carried out vectorization based on each user, obtain the purchasing power vector of each user, described purchasing power vector is x+1 dimensional vector, and every one dimension is corresponding with a class;
User's classifying module, for: cluster computing is carried out to described purchasing power vector, obtain x+1 the point bunch about purchasing power vector, each point bunch corresponding respectively class, using the purchasing power class of purchasing power corresponding for user vector point bunch corresponding class as user.
8. e-commerce user purchasing power categorizing system according to claim 7, is characterized in that, described commodity class determination module, specifically for: the class of the commodity of a% before price segment in same category is defined as top grade, and a adopts and confirms with the following method:
More than three or three spans are selected to be treat selective value between 0 ~ 100, maximum treats that selective value is as MAXIMUM SELECTION value, minimumly treat that selective value is minimum selective value, other treat that selective value is middle selective value, m is made to be the price statistics value of the front y% of price in same category, n is made to be the price statistics value having the front y% of price in the commodity of sales volume in same category in nearest preset time period, if m is greater than n and exceedes preset first threshold value, then a is selected to be MAXIMUM SELECTION value, if n is greater than m and exceedes default Second Threshold, then selection a is the minimum value in minimum selective value, other situations, selection a is one in middle selective value, wherein, y is less than MAXIMUM SELECTION value and is greater than minimum selective value.
9. e-commerce user purchasing power categorizing system according to claim 7, it is characterized in that, described user buys accounting computing module, specifically for:
For each user, calculate this user each class the amount of placing an order with dutiable value is taken the logarithm after product as the purchase volume of this user at this class, calculate the purchase volume summation of all class of this user, calculate each user accounts for the purchase volume summation of this user ratio in the purchase volume of each class buys each class commodity accounting as user;
For each user, the vector that the accounting each user being bought each class commodity is tieed up as an x+1, obtains the purchasing power vector of each user.
10. e-commerce user purchasing power categorizing system according to claim 7, is characterized in that, described user's classifying module, specifically for:
Center initialization submodule, for: from the purchasing power vector of all users, the purchasing power vector of a random selecting x+1 user is as the center, and sub-step is initially sorted out in execution;
Initial classification submodule, for: calculate the Euclidean distance of remaining all purchasing power vector to x+1 center respectively, each purchasing power vector is incorporated into respectively in the point minimum with the Euclidean distance at center bunch, implementation center's renewal sub-step;
Center upgrades submodule, for: to all purchasing power vector calculation in x+1 point bunch about the arithmetical mean of each dimension as the center of this point bunch, perform and sort out renewal sub-step;
Sort out and upgrade submodule, for: the European cluster with each center is recalculated at the center that the purchasing power of all users vector obtains according to classification renewal sub-step, each purchasing power vector is incorporated into respectively in the point minimum with the Euclidean distance at center bunch, perform convergence and judge sub-step;
Convergence judges submodule, for: if the center of each point bunch no longer changes, then perform class determination submodule, otherwise implementation center's renewal submodule;
Class determination submodule, for: according to the class corresponding to the center calculation corresponding point bunch of x+1 point bunch.
11. e-commerce user purchasing power categorizing systems according to claim 10, is characterized in that, described class determination submodule, specifically for:
Obtain x+1 point bunch center, perform following submodule successively according to class sequence:
High-grade chooser module, for: select current class to be determined for high-grade;
Class determination submodule, for: never to determine in the point bunch of class in the heart, the current of the point corresponding to center maximum for the dimension corresponding with current class to be determined bunch is defined as current class to be determined;
Other are when selecting submodule, for: if do not determine the point bunch of class in addition, then select next order class of current class to be determined as current class to be determined, perform class determination sub-step, otherwise terminate.
12. e-commerce user purchasing power categorizing systems according to claim 7, is characterized in that, also comprise:
Classification recommending module, for: when receiving the visit information of user, obtaining the purchasing power class of user, recommending the commodity with corresponding class according to purchasing power class to user.
Priority Applications (3)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510126137.6A CN104715409A (en) | 2015-03-20 | 2015-03-20 | Method and system for electronic commerce user purchasing power classification |
| HK15109485.6A HK1208946A1 (en) | 2015-03-20 | 2015-09-25 | Method and system for classifying purchasing power of electronic commerce user |
| PCT/CN2016/076811 WO2016150354A1 (en) | 2015-03-20 | 2016-03-18 | Method and system for classifying users of e-commerce platform |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510126137.6A CN104715409A (en) | 2015-03-20 | 2015-03-20 | Method and system for electronic commerce user purchasing power classification |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN104715409A true CN104715409A (en) | 2015-06-17 |
Family
ID=53414708
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510126137.6A Pending CN104715409A (en) | 2015-03-20 | 2015-03-20 | Method and system for electronic commerce user purchasing power classification |
Country Status (3)
| Country | Link |
|---|---|
| CN (1) | CN104715409A (en) |
| HK (1) | HK1208946A1 (en) |
| WO (1) | WO2016150354A1 (en) |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105516353A (en) * | 2016-01-06 | 2016-04-20 | 沈思远 | Network information popularization method and system |
| CN105894019A (en) * | 2016-03-30 | 2016-08-24 | 北京京东尚科信息技术有限公司 | Database data classification method and apparatus |
| WO2016150354A1 (en) * | 2015-03-20 | 2016-09-29 | 北京京东尚科信息技术有限公司 | Method and system for classifying users of e-commerce platform |
| CN106204163A (en) * | 2016-07-29 | 2016-12-07 | 北京京东尚科信息技术有限公司 | A kind of determination method and apparatus of user property feature |
| CN106445941A (en) * | 2015-08-05 | 2017-02-22 | 北京奇虎科技有限公司 | Recommendation method and apparatus for objects provided by website |
| CN107133843A (en) * | 2017-04-25 | 2017-09-05 | 丹露成都网络技术有限公司 | A kind of Method of Commodity Recommendation based on collaborative filtering |
| CN107545470A (en) * | 2017-08-30 | 2018-01-05 | 北京京东尚科信息技术有限公司 | Data processing method and system |
| CN108109042A (en) * | 2017-12-22 | 2018-06-01 | 中国联合网络通信集团有限公司 | Business favor information method for pushing and device |
| CN109272373A (en) * | 2018-08-02 | 2019-01-25 | 阿里巴巴集团控股有限公司 | A kind of computer based brand recommended method |
| CN110020135A (en) * | 2017-11-09 | 2019-07-16 | 阿里巴巴集团控股有限公司 | A kind of demand determines method, resource recommendation method and relevant apparatus |
| CN110245977A (en) * | 2019-05-21 | 2019-09-17 | 中国平安人寿保险股份有限公司 | A kind of upgrading service method, equipment and computer readable storage medium |
| CN113016169A (en) * | 2019-01-15 | 2021-06-22 | 深圳市欢太科技有限公司 | Information pushing method and related product |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112348300A (en) * | 2020-03-05 | 2021-02-09 | 北京沃东天骏信息技术有限公司 | Method and device for pushing information |
| CN111709812A (en) * | 2020-06-19 | 2020-09-25 | 山东财经大学 | E-commerce platform commodity recommendation method and system based on user dynamic classification |
| CN115935068B (en) * | 2022-12-12 | 2023-09-05 | 杭州洋驼网络科技有限公司 | Commodity recommendation method and device for Internet platform |
| CN115953166B (en) * | 2022-12-27 | 2024-04-02 | 鑫恒绅企业服务(无锡)有限公司 | Customer information management method and system based on big data intelligent matching |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103412948A (en) * | 2013-08-27 | 2013-11-27 | 北京交通大学 | Cluster-based collaborative filtering commodity recommendation method and system |
| CN103426102A (en) * | 2013-08-02 | 2013-12-04 | 苏州两江科技有限公司 | Commodity feature recommending method based on body classification |
| EP2711887A1 (en) * | 2012-03-30 | 2014-03-26 | Rakuten, Inc. | Information processing device, information processing method, information processing program, and recording medium |
| CN103729383A (en) * | 2012-10-16 | 2014-04-16 | 阿里巴巴集团控股有限公司 | Push method and device for commodity information |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20010014868A1 (en) * | 1997-12-05 | 2001-08-16 | Frederick Herz | System for the automatic determination of customized prices and promotions |
| CN102567900A (en) * | 2011-12-28 | 2012-07-11 | 尚明生 | Method for recommending commodities to customers |
| CN104715409A (en) * | 2015-03-20 | 2015-06-17 | 北京京东尚科信息技术有限公司 | Method and system for electronic commerce user purchasing power classification |
-
2015
- 2015-03-20 CN CN201510126137.6A patent/CN104715409A/en active Pending
- 2015-09-25 HK HK15109485.6A patent/HK1208946A1/en unknown
-
2016
- 2016-03-18 WO PCT/CN2016/076811 patent/WO2016150354A1/en active Application Filing
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP2711887A1 (en) * | 2012-03-30 | 2014-03-26 | Rakuten, Inc. | Information processing device, information processing method, information processing program, and recording medium |
| CN103729383A (en) * | 2012-10-16 | 2014-04-16 | 阿里巴巴集团控股有限公司 | Push method and device for commodity information |
| CN103426102A (en) * | 2013-08-02 | 2013-12-04 | 苏州两江科技有限公司 | Commodity feature recommending method based on body classification |
| CN103412948A (en) * | 2013-08-27 | 2013-11-27 | 北京交通大学 | Cluster-based collaborative filtering commodity recommendation method and system |
Cited By (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2016150354A1 (en) * | 2015-03-20 | 2016-09-29 | 北京京东尚科信息技术有限公司 | Method and system for classifying users of e-commerce platform |
| CN106445941A (en) * | 2015-08-05 | 2017-02-22 | 北京奇虎科技有限公司 | Recommendation method and apparatus for objects provided by website |
| CN105516353B (en) * | 2016-01-06 | 2018-12-25 | 沈思远 | A kind of network information promotion method and system |
| CN105516353A (en) * | 2016-01-06 | 2016-04-20 | 沈思远 | Network information popularization method and system |
| CN105894019A (en) * | 2016-03-30 | 2016-08-24 | 北京京东尚科信息技术有限公司 | Database data classification method and apparatus |
| CN106204163A (en) * | 2016-07-29 | 2016-12-07 | 北京京东尚科信息技术有限公司 | A kind of determination method and apparatus of user property feature |
| CN107133843A (en) * | 2017-04-25 | 2017-09-05 | 丹露成都网络技术有限公司 | A kind of Method of Commodity Recommendation based on collaborative filtering |
| CN107545470B (en) * | 2017-08-30 | 2021-04-30 | 北京京东尚科信息技术有限公司 | Data processing method and system |
| CN107545470A (en) * | 2017-08-30 | 2018-01-05 | 北京京东尚科信息技术有限公司 | Data processing method and system |
| CN110020135A (en) * | 2017-11-09 | 2019-07-16 | 阿里巴巴集团控股有限公司 | A kind of demand determines method, resource recommendation method and relevant apparatus |
| CN110020135B (en) * | 2017-11-09 | 2023-01-31 | 阿里巴巴集团控股有限公司 | Demand determination method, resource recommendation method and related device |
| CN108109042A (en) * | 2017-12-22 | 2018-06-01 | 中国联合网络通信集团有限公司 | Business favor information method for pushing and device |
| CN109272373A (en) * | 2018-08-02 | 2019-01-25 | 阿里巴巴集团控股有限公司 | A kind of computer based brand recommended method |
| CN109272373B (en) * | 2018-08-02 | 2021-09-14 | 创新先进技术有限公司 | Brand recommendation method based on computer |
| CN113016169A (en) * | 2019-01-15 | 2021-06-22 | 深圳市欢太科技有限公司 | Information pushing method and related product |
| CN110245977A (en) * | 2019-05-21 | 2019-09-17 | 中国平安人寿保险股份有限公司 | A kind of upgrading service method, equipment and computer readable storage medium |
| CN110245977B (en) * | 2019-05-21 | 2023-10-20 | 中国平安人寿保险股份有限公司 | Service upgrading method, equipment and computer readable storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| WO2016150354A1 (en) | 2016-09-29 |
| HK1208946A1 (en) | 2016-03-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104715409A (en) | Method and system for electronic commerce user purchasing power classification | |
| CN103136683B (en) | Calculate method, device and product search method, the system of product reference price | |
| CN108205768A (en) | Database building method and data recommendation method and device, equipment and storage medium | |
| CN108256691A (en) | Refund Probabilistic Prediction Model construction method and device | |
| CN106708821A (en) | User personalized shopping behavior-based commodity recommendation method | |
| CN110335157A (en) | Insurance products recommended method, equipment and storage medium | |
| CN110580649A (en) | Method and device for determining potential value of commodity | |
| CN106651418A (en) | Method of recommending add-on item for special offer when spending enough by e-business | |
| CN104615721B (en) | For the method and system based on return of goods related information Recommendations | |
| CN107301592A (en) | The method and device excavated for commodity substitute | |
| CN102609422A (en) | Class misplacing identification method and device | |
| CN111738805B (en) | Behavior log-based search recommendation model generation method, device and storage medium | |
| CN107516246A (en) | Determination method, determining device, medium and the electronic equipment of user type | |
| CN104331818A (en) | Method and system capable of controlling differential pushing of merchant service information | |
| CN105468628B (en) | A kind of sort method and device | |
| CN104572775A (en) | Advertisement classification method, device and server | |
| CN109447762A (en) | Method of Commodity Recommendation and its device, server, commercial product recommending system | |
| CN113724042A (en) | Commodity recommendation method, commodity recommendation device, commodity recommendation medium and commodity recommendation equipment | |
| CN106776859A (en) | Mobile solution App commending systems based on user preference | |
| CN106251178A (en) | Data digging method and device | |
| CN107093122B (en) | Object classification method and device | |
| CN106447429A (en) | Message pushing method, apparatus and terminal | |
| CN106919995A (en) | A kind of method and device for judging user group's loss orientation | |
| CN106447397A (en) | Tobacco retail customer pricing method based on decision tree algorithm | |
| CN111861679A (en) | Commodity recommendation method based on artificial intelligence |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| REG | Reference to a national code |
Ref country code: HK Ref legal event code: DE Ref document number: 1208946 Country of ref document: HK |
|
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20150617 |
|
| REG | Reference to a national code |
Ref country code: HK Ref legal event code: WD Ref document number: 1208946 Country of ref document: HK |