CN106296290A - Personalized order recommendation method based on big data and data mining - Google Patents

Abstract

The invention discloses a personalized order recommendation method based on big data and data mining. By mining the user data and behaviors, the retail user can quickly find the cigarettes suitable for self-sale, the tobacco enterprise can recommend the cigarettes to the suitable retail user, the tobacco enterprise is helped to market the commodities, the economic benefit is increased, and meanwhile, the time and the energy of the user for selecting the commodities are saved.

Description

A kind of customized orders based on big data and data mining recommends method

Technical field

The present invention relates to data analysis technique field, be specifically related to a kind of personalization based on big data and data mining and order Single recommendation method.

Background technology

Society all can produce the information of magnanimity every day now.In the face of so a lot of people of multi information does not knows that what information is Oneself needs.

Along with the development of tobacco business ecommerce, have accumulated substantial amounts of user data and behavioral data, the most effectively Utilize these data to improve experience and the economic benefit of tobacco enterprise of user, change traditional retail customer ordering method.

Data mining technology is one of focus of technical development of computer in recent years.By the mass data to historical accumulation Effective excavation, it appeared that the rule hidden or pattern, provide for decision-making and support, and these rules or pattern are to depend on Obtain by simple data query, or can not obtain within the acceptable time.These rules or pattern can be further Knowledge is become under the identification of professional.Task faced by data mining is complicated, generally includes classification, predicts, associates Rule discovery and cluster analysis etc..

Web data is excavated and is set up on the basis of being analyzed substantial amounts of network data, uses corresponding data mining Algorithm, in the concrete extraction of application model enterprising row data, data screening, data conversion, data mining and pattern analysis, After make the reasoning of inductive, the prediction personalized behavior of client and user habit, thus help to carry out decision-making and management, subtract The risk of few decision-making.

Web data is excavated and is related to multiple field, in addition to data mining, further relates to computer network, data base and data bins The technology such as storage, artificial intelligence, information retrieval, visualization, natural language understanding.

Summary of the invention

The technical problem to be solved in the present invention is: the present invention provides a kind of personalization based on big data and data mining to order Single recommendation method, by the excavation to user data and behavior, allows retail customer can be quickly found out the suitable Medicated cigarette oneself sold, Allow tobacco enterprise that Medicated cigarette can be recommended applicable retail customer.

The technical solution adopted in the present invention is:

A kind of customized orders based on big data and data mining recommends method, and described method is by calculating between user Similarity, find the user having similar Running Characteristic to user, similarity based on user to recommend similar users to purchase to user The commodity bought, improve the accuracy recommended.

In described method, similar users generation process is as follows:

To a user, find the user having similar Running Characteristic to this user, and the business that this group user was bought Product comprise one group of purchasing history as the Recommendations of candidate, a user, commodity that i.e. user bought and quantity, use and use The quantity purchase vector at family identifies user, and use that Pearson relevance metric formula calculates between user vector two-by-two is similar Degree sim (x, y):

$sim(x,y)=\frac{{\Σ}_{p\∈P_{xy}}(r_{x,p}-{\stackrel{\‾}{r}}_x)(r_{y,p}-{\stackrel{\‾}{r}}_y)}{\sqrt{{\Σ}_{p\∈P_{xy}}{(r_{x,p}-{\stackrel{\‾}{r}}_x)}^2{\Σ}_{p\∈P_{xy}}{(r_{y,p}-{\stackrel{\‾}{r}}_y)}^2}}$

Wherein: Pxy represents one group of commodity that user x, y purchased jointly；

To a user, select with its similarity ranking top n user from high to low as similar users.

Described method being predicted, commodity scoring process is as follows:

To a user, buy the quantity of commodity according to similar users, it was predicted that the probability of these user's candidate's commodity, and root The weighted sum of one group of similar users candidate's commodity is calculated according to following formula:

r_u,p=k ∑_u′∈U sim(u,u′)×r_u′,p

Wherein k represents normalization factor；

Finally by using probabilistic model modeling, calculate user and like the probability of certain part candidate's commodity.

The invention have the benefit that

The present invention, by the excavation to user data and behavior, allows retail customer can be quickly found out the suitable volume oneself sold Cigarette, allows tobacco enterprise that Medicated cigarette can be recommended applicable retail customer, is helping tobacco enterprise to promote the sale of goods, is increasing economic well-being of workers and staff While, also save the time and efforts of picking commodities for users.

Detailed description of the invention

Below according to detailed description of the invention, the present invention is further described:

Embodiment 1:

A kind of customized orders based on big data and data mining recommends method, and described method is by calculating between user Similarity, find the user having similar Running Characteristic to user, similarity based on user to recommend similar users to purchase to user The commodity bought, improve the accuracy recommended.

Embodiment 2

On the basis of embodiment 1, in method described in the present embodiment, similar users generation process is as follows:

To a user, find the user having similar Running Characteristic to this user, and the business that this group user was bought Product comprise one group of purchasing history as the Recommendations of candidate, a user, commodity that i.e. user bought and quantity, use and use The quantity purchase vector at family identifies user, and use that Pearson relevance metric formula calculates between user vector two-by-two is similar Degree sim (x, y):

$sim(x,y)=\frac{{\Σ}_{p\∈P_{xy}}(r_{x,p}-{\stackrel{\‾}{r}}_x)(r_{y,p}-{\stackrel{\‾}{r}}_y)}{\sqrt{{\Σ}_{p\∈P_{xy}}{(r_{x,p}-{\stackrel{\‾}{r}}_x)}^2{\Σ}_{p\∈P_{xy}}{(r_{y,p}-{\stackrel{\‾}{r}}_y)}^2}}$

Wherein: Pxy represents one group of commodity that user x, y purchased jointly；

To a user, select with its similarity ranking top n user from high to low as similar users.

Embodiment 3

On the basis of embodiment 2, method described in the present embodiment being predicted, commodity scoring process is as follows:

To a user, buy the quantity of commodity according to similar users, it was predicted that the probability of these user's candidate's commodity, and root The weighted sum of one group of similar users candidate's commodity is calculated according to following formula:

r_u,p=k ∑_u′∈U sim(u,u′)×r_u′,p

Wherein k represents normalization factor；

Described method, by using probabilistic model to model, calculates user and likes the probability of certain part candidate's commodity.

Embodiment is merely to illustrate the present invention, and not limitation of the present invention, about the ordinary skill of technical field Personnel, without departing from the spirit and scope of the present invention, it is also possible to make a variety of changes and modification, the most all equivalents Technical scheme fall within scope of the invention, the scope of patent protection of the present invention should be defined by the claims.

Claims (3)

1. a customized orders based on big data and data mining recommends method, it is characterised in that: described method is by meter Calculating the similarity between user, find the user having similar Running Characteristic to user, similarity based on user is recommended to user The commodity that similar users is bought, improve the accuracy recommended.

A kind of customized orders based on big data and data mining the most according to claim 1 recommends method, its feature Being, in described method, similar users generation process is as follows:

To a user, find the user having similar Running Characteristic to this user, and the commodity that this group user was bought are made For the Recommendations of candidate, a user comprises one group of purchasing history, commodity that i.e. user bought and quantity, uses user's Quantity purchase vector identifies user, uses Pearson relevance metric formula to calculate similarity sim between user vector two-by-two (x,y):

$sim\left(x,y\right)=\frac{{\mathrm{\Σ}}_{p\∈P_{xy}}\left(r_{x,p}-{\stackrel{\‾}{r}}_x\right)\left(r_{y,p}-{\stackrel{\‾}{r}}_y\right)}{\sqrt{{\mathrm{\Σ}}_{p\∈P_{xy}}{\left(r_{x,p}-{\stackrel{\‾}{r}}_x\right)}^2{\mathrm{\Σ}}_{p\∈P_{xy}}{\left(r_{y,p}-{\stackrel{\‾}{r}}_y\right)}^2}}$

Wherein: Pxy represents one group of commodity that user x, y purchased jointly；

To a user, select with its similarity ranking top n user from high to low as similar users.

A kind of customized orders based on big data and data mining the most according to claim 2 recommends method, its feature It is, described method being predicted, commodity scoring process is as follows:

To a user, buy the quantity of commodity according to similar users, it was predicted that the probability of these user's candidate's commodity, and according under State formula and calculate the weighted sum of one group of similar users candidate's commodity:

r_u,p=k Σ_u′∈U sim(u,u′)×r_u′,p

Wherein k represents normalization factor；

Finally by using probabilistic model modeling, calculate user and like the probability of certain part candidate's commodity.