CN104616180A - Method for predicting hot sellers - Google Patents

Abstract

The invention discloses a method for predicting hot sellers. The method is achieved by three steps, namely data acquisition, data preprocessing and data analysis, wherein during the data acquisition, through Internet and by a vertical search engine technology, structured and unstructured network electronic business data can be acquired from the network; during the data preprocessing, by a noise-reducing and normalizing technology, the data is processed to remove interfering data so as to prepare for the following data analysis; during the data analysis, the preprocessed data is modeled and analyzed to obtain prediction of the hot sellers and the prediction is displayed in a graphical form. Compared with the prior art, the method for predicting hot sellers has the following advantages: by the method, the data can be fully utilized; through processing and analysis on the electronic business data, an electronic business user is helped to purposefully analyze hot-selling trends of goods so as to achieve a purpose of effectively predicting the shelf goods to earn more profits.

Description

A kind of hot item Forecasting Methodology

Technical field

The present invention relates to computer network technical field of electronic commerce, specifically a kind of hot item Forecasting Methodology.

Background technology

Along with the fast development of ecommerce, people more and more get used to using e-commerce website to buy commodity.But seller, while earning profit by Internet channel, but cannot be undertaken predicting later hot item by existing internet information, reach the object of the more profits of earning.How to help by a kind of effective mechanism the tendency that seller analyzes merchandise sales fast, to hold the type of added commodity accurately, be focus and the difficult point of current ecommerce research.

Data mining refers to the process excavating effective knowledge from the mass data leaving database, data warehouse or other information banks in.Data mining extracts implicit, valuable and intelligible information from mass data, to instruct the activity of people.Data mining technology mainly contains correlation rule, classifying rules, cluster analysis and sequence pattern etc.

Data mining is a kind of new business information treatment technology, and its principal feature extracts a large number of services data in business database, changes, analyzes and other modelling process, therefrom extracts the critical data of auxiliary commerce decision-making.In brief, data mining is the data analysing method of a class profound level in fact.

Data mining from a large amount of, incomplete, noisy, fuzzy, random real application data, extract lie in wherein, people are ignorant in advance but be the information of potentially useful and the process of knowledge.The synonym close with data mining has data to melt platform, data analysis and decision support etc.This definition comprises several layers implication: data source must be real, a large amount of, Noise; The found out that interested knowledge of user; The knowledge found will can accept, can understand, can use; Do not require the knowledge that the four seas that find to put are all accurate, only support specifically to pinpoint the problems.

Current e-commerce website is generally adopt the conventional art such as statistical technique and multidimensional analysis, with the experience of website design personnel, carries out prediction recommend the fast-selling trend of commodity.This commodity projection mode lacks accuracy, is difficult to reach remarkable result.These conventional arts are technology of checking type, are difficult to obtain be hidden in data knowledge behind.At present, also do not utilize data mining technology, carry out the fast-selling trend of commodity predicting the method for recommending.

Summary of the invention

Technical assignment of the present invention is to provide a kind of hot item Forecasting Methodology.

Technical assignment of the present invention realizes in the following manner, and the method realizes through data acquisition, data prediction and data analysis three step:

Data acquisition: utilize vertical search engine technology by internet, from network collection structuring and non-structured network electricity quotient data;

Data prediction: by noise reduction, normalization technology processes data, removes interfering data, for next step data analysis is prepared;

Data analysis: pretreated data are carried out modeling analysis, draws the prediction of hot item and shows with patterned form.

In described data acquisition, search for all inside and outside data messages relevant with business object, and therefrom select the data being applicable to data mining application.

Described noise reduction: filtered out the noise in data by data analysis algorithm, the data that analytical work is adopted are more accurate;

Normalization: when calculating the fancy grade of user to article, need to be weighted different behavioral datas; The data of each behavior are unified in an identical span, more accurate by being normalized the general trend making weighted sum obtain.

The technology that described Modeling analysis adopts is: Bayesian network technology, correlation rule, cluster analysis, Horting diagram technology and collaborative filtering.

A kind of hot item Forecasting Methodology of the present invention compared to the prior art, data can be made to be fully utilized, carry out the fast-selling trend of the commercial family of auxiliary electrical to commodity by the process to electric quotient data, analysis and carry out autotelic analysis, reach the effective prediction to added commodity, earn the object of more profits.

Accompanying drawing explanation

Accompanying drawing 1 is a kind of FB(flow block) of hot item Forecasting Methodology.

Embodiment

Embodiment 1:

The method realizes through data acquisition, data prediction and data analysis three step:

Data acquisition: utilize vertical search engine technology by internet, from network collection structuring and non-structured network electricity quotient data; In data acquisition, search for all inside and outside data messages relevant with business object, and therefrom select the data being applicable to data mining application.

Data prediction: by noise reduction, normalization technology processes data, removes interfering data, for next step data analysis is prepared;

Noise reduction: filtered out the noise in data by data analysis algorithm, the data that analytical work is adopted are more accurate;

Normalization: when calculating the fancy grade of user to article, need to be weighted different behavioral datas; The data of each behavior are unified in an identical span, more accurate by being normalized the general trend making weighted sum obtain.

Data analysis: pretreated data are carried out modeling analysis, draws the prediction of hot item and shows with patterned form.

The each system of last mode, makes system run all right.

Embodiment 2:

The method realizes through data acquisition, data prediction and data analysis three step:

Data acquisition: utilize vertical search engine technology by internet, from network collection structuring and non-structured network electricity quotient data; In data acquisition, search for all inside and outside data messages relevant with business object, and therefrom select the data being applicable to data mining application, after collection, unification is stored in raw data warehouse.

Data prediction: by noise reduction, normalization technology processes data, removes interfering data, for next step data analysis is prepared;

Noise reduction: filtered out the noise in data by data analysis algorithm, the data that analytical work is adopted are more accurate;

Normalization: when calculating the fancy grade of user to article, need to be weighted different behavioral datas; The data of each behavior are unified in an identical span, more accurate by being normalized the general trend making weighted sum obtain.

Data analysis: pretreated data acquisition Bayesian network technology, correlation rule, cluster analysis, Horting diagram technology and collaborative filtering are carried out modeling analysis, draw the prediction of hot item and show with patterned form, being presented to user.

The each system of last mode, makes system run all right.

By embodiment above, described those skilled in the art can be easy to realize the present invention.But should be appreciated that the present invention is not limited to above-mentioned several embodiments.On the basis of disclosed embodiment, described those skilled in the art can the different technical characteristic of combination in any, thus realizes different technical schemes.

Claims (4)

1. a hot item Forecasting Methodology, is characterized in that, the method realizes through data acquisition, data prediction and data analysis three step:

Data acquisition: utilize vertical search engine technology by internet, from network collection structuring and non-structured network electricity quotient data;

Data prediction: by noise reduction, normalization technology processes data, removes interfering data, for next step data analysis is prepared;

Data analysis: pretreated data are carried out modeling analysis, draws the prediction of hot item and shows with patterned form.

2. a kind of hot item Forecasting Methodology according to claim 1, is characterized in that, in described data acquisition, searches for all inside and outside data messages relevant with business object, and therefrom selects the data being applicable to data mining application.

3. a kind of hot item Forecasting Methodology according to claim 1, it is characterized in that, described noise reduction: filtered out the noise in data by data analysis algorithm, the data that analytical work is adopted are more accurate;

Normalization: when calculating the fancy grade of user to article, need to be weighted different behavioral datas; The data of each behavior are unified in an identical span, more accurate by being normalized the general trend making weighted sum obtain.

4. a kind of hot item Forecasting Methodology according to claim 1, is characterized in that, the technology that described Modeling analysis adopts is: Bayesian network technology, correlation rule, cluster analysis, Horting diagram technology and collaborative filtering.