CN108960719A - Selection method and apparatus and computer readable storage medium - Google Patents
Selection method and apparatus and computer readable storage medium Download PDFInfo
- Publication number
- CN108960719A CN108960719A CN201810693419.8A CN201810693419A CN108960719A CN 108960719 A CN108960719 A CN 108960719A CN 201810693419 A CN201810693419 A CN 201810693419A CN 108960719 A CN108960719 A CN 108960719A
- Authority
- CN
- China
- Prior art keywords
- commodity
- feature
- line
- data
- added
- 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.)
- Granted
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
- G06Q10/00—Administration; Management
- G06Q10/08—Logistics, e.g. warehousing, loading or distribution; Inventory or stock management
- G06Q10/087—Inventory or stock management, e.g. order filling, procurement or balancing against orders
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- 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
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0201—Market modelling; Market analysis; Collecting market data
-
- 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
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0201—Market modelling; Market analysis; Collecting market data
- G06Q30/0203—Market surveys; Market polls
Abstract
The invention discloses a kind of selection method and apparatus and computer readable storage mediums, are related to data processing field.Selection method includes: the characteristic value that commodity data on the line according to alternative commodity and the user behavior data for being less than user on the line of preset value to distance under the line in selection shop determine alternative commodity；The characteristic value of alternative commodity is inputted into disaggregated model, wherein disaggregated model is to train using the characteristic value of training sample data as input, sales volume information as mark value；Determined whether according to the classification results of disaggregated model to select alternative commodity to selection shop.It is thus possible to improve selection efficiency and accuracy rate from the commodity for quickly determining shop under line on magnanimity line in commodity and needing to adopt pin.
Description
Technical field
The present invention relates to data processing field, in particular to a kind of selection method and apparatus and computer-readable storage medium
Matter.
Background technique
With the diversification of consumer demand, electric business switchs to on-line off-line fusion from pure on-line selling, gradually becomes
A kind of electric business form of increasingly mainstream.The merchandise resources of Xian Xia shops mainly pass through line upper mounting plate and supply, and need from magnanimity commodity
Middle selection particular commodity.
What the commodity that line is put up at an inn at present were mainly taken is the artificial selection mode for adopting pin personnel, i.e., generally by adopting pin people
Member business experience, carry out supplier select, commodity audit etc., to be investigated to the commodity various aspects on platform, therefrom
Selection participates in the commodity for being suitble to shops's positioning and corresponding category.
Summary of the invention
Inventor it is found after analysis that, due to quantity in stock unit (the Stock Keeping on current large-scale electric business platform
Unit, referred to as: SKU) mostly in the millions, therefore the workload of artificial selection mode is huge to quantity.Further, since artificial choosing
There are subjective bias for product, do not have Comprehensive Evaluation index, cause artificial selection mode probability of failure higher, and then lead to sales performance
It is bad.
One technical problem to be solved by the embodiment of the invention is that: how to improve under line the efficiency of shop selection and accurate
Rate.
First aspect according to some embodiments of the invention provides a kind of selection method, comprising: according to alternative commodity
Commodity data and the user behavior data for being less than user on the line of preset value to distance under the line in selection shop determine alternative on line
The characteristic value of commodity；The characteristic value of alternative commodity is inputted into disaggregated model, wherein disaggregated model is with the spy of training sample data
Value indicative is as input, sales volume information as mark value training；Determined whether according to the classification results of disaggregated model for selection
Shop selects alternative commodity.
In some embodiments, selection method further include: obtain commodity data and user behavior number on the line for training
According to；Fusion Features are carried out to commodity data on the line of acquisition and user behavior data according to commodity sign, generate number of training
According to, wherein training sample data include the corresponding characteristic value of each commodity sign and sales volume information；Using training sample data
Characteristic value is trained disaggregated model as mark value as input, corresponding sales volume information.
In some embodiments, disaggregated model is categorised decision tree, and the classification node in categorised decision tree is the spy of commodity
Sign, leaf node presentation class result.
In some embodiments, selection method further include: obtain commodity data and user behavior number on the line for training
According to；Fusion Features are carried out to commodity data on the line of acquisition and user behavior data according to commodity sign, generate number of training
According to, wherein training sample data include the corresponding characteristic value of each commodity sign and sales volume information；Disaggregated model is instructed
Practice, comprising: according to training sample data corresponding to site position to be added, calculate the feature being not added into categorised decision tree
Information gain；The maximum feature of information gain is added to site position to be added, as classification node；Repeat above calculate
The step of information gain, addition feature, until all features are all added in categorised decision tree.
In some embodiments, selection method further include: obtain commodity data and user behavior number on the line for training
According to；Fusion Features are carried out to commodity data on the line of acquisition and user behavior data according to commodity sign, generate number of training
According to, wherein training sample data include the corresponding characteristic value of each commodity sign and sales volume information；Disaggregated model is instructed
Practice, comprising: according to training sample data corresponding to site position to be added, calculate the feature being not added into categorised decision tree
Information gain；Preset information gain threshold is respectively less than in the information gain of each feature being not added into categorised decision tree
In the case where, to the leaf node of site position to be added addition presentation class result, wherein classification results are according to knot to be added
The corresponding mark value in point position determines；It is respectively less than not preset in the information gain for the feature being not added into categorised decision tree
In the case where information gain threshold, the maximum feature of information gain is added to site position to be added, as classification node；Weight
Multiple above the step of calculating information gain, addition feature, until all features are all added in categorised decision tree.
In some embodiments, selection method further include: the characteristic value for testing commodity is input in categorised decision tree；Root
The predictablity rate of categorised decision tree is determined according to the classification results of categorised decision tree and the mark value of test commodity；In response to prediction
Accuracy rate is lower than preset accuracy rate threshold value, executes re -training categorised decision tree after following at least one operation: modification information
Gain threshold；More new feature；Update training sample data.
In some embodiments, selection method further include: obtain multiple training sample data subsets, wherein each training
The corresponding commodity of sample data subset belong to same category；Each training sample data trained disaggregated model is respectively adopted,
Obtain multiple subclassification models；According to the predictablity rate of each subclassification model, the weight of each subclassification model is determined, with
Just determined whether according to the weighing computation results of the classification results of each subclassification model to select alternative commodity to selection shop.
In some embodiments, selection method further includes executing following at least one Feature Selection method: screening out correlation
Greater than a feature in the different characteristic of preset value；Screen out the feature for being less than preset value with the correlation of sales volume information.
In some embodiments, commodity data comprises at least one of the following on line: bought on line feature, browse on line it is special
Sign, search characteristics on line add shopping cart feature, brand sales volume feature on line, evaluating characteristic on line, primary attribute on line on line
Feature is bought on the line of user in preset range under feature, line；User behavior data comprises at least one of the following: user location is special
It is special that sign, user buy feature, user's browsing feature, user's search characteristics, user's addition shopping cart feature, user's purchase brand
Sign, user base attributive character.
The second aspect according to some embodiments of the invention provides a kind of selection device, comprising: alternative product features value
Determining module is configured as commodity data on the line according to alternative commodity, is less than preset value with to distance under the line in selection shop
Line on the user behavior data of user determine the characteristic values of alternative commodity；Alternative product features value input module, is configured as
The characteristic values of alternative commodity is inputted into disaggregated model, wherein disaggregated model be using the characteristic value of training sample data as input,
Sales volume information is as mark value training；Selection determining module is configured as being determined whether according to the classification results of disaggregated model
To select alternative commodity to selection shop.
In some embodiments, selection device further include: disaggregated model training module is configured as obtaining for training
Commodity data and user behavior data on line；Commodity data on the line of acquisition and user behavior data are carried out according to commodity sign
Fusion Features generate training sample data, wherein training sample data include the corresponding characteristic value of each commodity sign and sales volume
Information；Disaggregated model is carried out as mark value as input, corresponding sales volume information using the characteristic value of training sample data
Training.
In some embodiments, disaggregated model is categorised decision tree, and the classification node in categorised decision tree is the spy of commodity
Sign, leaf node presentation class result.
In some embodiments, selection device further include: categorised decision tree training module is configured as obtaining for training
Line on commodity data and user behavior data；According to commodity sign to commodity data on the line of acquisition and user behavior data into
Row Fusion Features generate training sample data, wherein training sample data include the corresponding characteristic value of each commodity sign and pin
Measure information；Disaggregated model is trained, comprising: according to training sample data corresponding to site position to be added, calculate not
Be added to the information gain of the feature in categorised decision tree, by the maximum feature of information gain be added to site position to be added,
As classification node, above the step of calculating information gain, addition feature is repeated until all features are all added to categorised decision
In tree.
In some embodiments, selection device further include: categorised decision tree training module is configured as obtaining for training
Line on commodity data and user behavior data；According to commodity sign to commodity data on the line of acquisition and user behavior data into
Row Fusion Features generate training sample data, wherein training sample data include the corresponding characteristic value of each commodity sign and pin
Measure information；Disaggregated model is trained, comprising: according to training sample data corresponding to site position to be added, calculate not
It is added to the information gain of the feature in categorised decision tree, in the information gain of each feature being not added into categorised decision tree
The leaf node of presentation class result is added in the case where respectively less than preset information gain threshold, to site position to be added,
In the case where the information gain for the feature being not added into categorised decision tree is not respectively less than preset information gain threshold, incite somebody to action
The maximum feature of information gain is added to site position to be added as classification node, repeats the above calculating information gain, addition
The step of feature, is all added in categorised decision tree until all features；Wherein, classification results are according to site position pair to be added
The mark value answered determines.
In some embodiments, selection device further include: disaggregated model testing and debugging module is configured as that commodity will be tested
Characteristic value be input in categorised decision tree；It is determined and is classified according to the mark value of the classification results of categorised decision tree and test commodity
The predictablity rate of decision tree；It is lower than preset accuracy rate threshold value in response to predictablity rate, executes following at least one operation
Re -training categorised decision tree afterwards: modification information gain threshold；More new feature；Update training sample data.
In some embodiments, selection device further include: multi-model training module is configured as obtaining multiple training samples
Data subset, wherein the corresponding commodity of each training sample data subset belong to same category；Each training sample is respectively adopted
Data subset train classification models obtain multiple subclassification models；According to the predictablity rate of each subclassification model, determine every
The weight of a sub- disaggregated model, so as to according to the weighing computation results of the classification results of each subclassification model determine whether for
Selection shop selects alternative commodity.
In some embodiments, selection device further include: Feature Selection module is configured as executing following at least one special
Sign screening technique: correlation is screened out greater than a feature in the different characteristic of preset value；Screen out the correlation with sales volume information
Less than the feature of preset value.
In some embodiments, commodity data comprises at least one of the following on line: bought on line feature, browse on line it is special
Sign, search characteristics on line add shopping cart feature, brand sales volume feature on line, evaluating characteristic on line, primary attribute on line on line
Feature is bought on the line of user in preset range under feature, line；User behavior data comprises at least one of the following: user location is special
It is special that sign, user buy feature, user's browsing feature, user's search characteristics, user's addition shopping cart feature, user's purchase brand
Sign, user base attributive character.
In terms of third according to some embodiments of the invention, a kind of selection device is provided, comprising: memory；And coupling
It is connected to the processor of the memory, the processor is configured to the instruction based on storage in the memory, before execution
State any one selection method.
The 4th aspect according to some embodiments of the invention, provides a kind of computer readable storage medium, stores thereon
There is computer program, which is characterized in that the program realizes any one aforementioned selection method when being executed by processor.
Some embodiments in foregoing invention have the following advantages that or the utility model has the advantages that the embodiment of the present invention can pass through line
Data under upper commodity data and line to the user around selection shop determine the characteristic value of alternative commodity, and using instruction in advance
Experienced disaggregated model classifies to alternative commodity, selects the alternative commodity to selection shop to determine whether, so as to
From the commodity for quickly determining shop under line on magnanimity line in commodity and needing to adopt pin, selection efficiency and accuracy rate are improved.
By referring to the drawings to the detailed description of exemplary embodiment of the present invention, other feature of the invention and its
Advantage will become apparent.
Detailed description of the invention
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, to embodiment or will show below
There is attached drawing needed in technical description to be briefly described, it should be apparent that, the accompanying drawings in the following description is only this
Some embodiments of invention without any creative labor, may be used also for those of ordinary skill in the art
To obtain other drawings based on these drawings.
Fig. 1 is the exemplary process diagram according to the selection method of some embodiments of the invention.
Fig. 2A and 2B is the exemplary process diagram according to the Feature Selection method of some embodiments of the invention.
Fig. 3 is the exemplary process diagram according to the disaggregated model training method of some embodiments of the invention.
Fig. 4 is the categorised decision tree schematic diagram in some embodiments of the invention.
Fig. 5 A and 5B are the exemplary process diagram according to the categorised decision tree training method of some embodiments of the invention.
Fig. 6 is the exemplary process diagram according to the categorised decision tree training method of other embodiments of the invention.
Fig. 7 is the exemplary process diagram according to the disaggregated model method of adjustment of some embodiments of the invention.
Fig. 8 is the exemplary block diagram according to the selection device of some embodiments of the invention.
Fig. 9 is the exemplary block diagram according to the selection device of other embodiments of the invention.
Figure 10 is the exemplary block diagram according to the selection device of yet other embodiments of the invention.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.Below
Description only actually at least one exemplary embodiment be it is illustrative, never as to the present invention and its application or make
Any restrictions.Based on the embodiments of the present invention, those of ordinary skill in the art are not making creative work premise
Under every other embodiment obtained, shall fall within the protection scope of the present invention.
Unless specifically stated otherwise, positioned opposite, the digital table of the component and step that otherwise illustrate in these embodiments
It is not limited the scope of the invention up to formula and numerical value.
Simultaneously, it should be appreciated that for ease of description, the size of various pieces shown in attached drawing is not according to reality
Proportionate relationship draw.
Technology, method and apparatus known to person of ordinary skill in the relevant may be not discussed in detail, but suitable
In the case of, the technology, method and apparatus should be considered as authorizing part of specification.
It is shown here and discuss all examples in, any occurrence should be construed as merely illustratively, without
It is as limitation.Therefore, the other examples of exemplary embodiment can have different values.
It should also be noted that similar label and letter indicate similar terms in following attached drawing, therefore, once a certain Xiang Yi
It is defined in a attached drawing, then in subsequent attached drawing does not need that it is further discussed.
Fig. 1 is the exemplary process diagram according to the selection method of some embodiments of the invention.As shown in Figure 1, the embodiment
Selection method include step S102~S106.
In step s 102, it is less than according to commodity data on the line of alternative commodity and to distance under the line in selection shop pre-
If the user behavior data of user determines the characteristic value of alternative commodity on the line of value.To selection shop under the line of commodity to be selected
Shop.
On line commodity data include the sales datas of commodity, data on flows, after-sale service data, supply chain data, on line
User's representation data of user etc..User behavior data is under line around shop, with the matched crowd's information of user on line,
The mark of these crowds can be obtained by operating position under the line of acquisition user when user operate on line.
User under line around shop can be defined as in the covering for being less than preset value with distance under the line to selection shop
In range, the number of predetermined registration operation within a preset time reaches the user of preset value.For example, in nearest 3 months, to be selected
Lower list number is more than 5 times, browses web sites or can be by using the user that number of applications is more than 20 times in the range of 3 kilometers around product
The user being defined as under line around shop.
In step S104, the characteristic value of alternative commodity is inputted into disaggregated model, wherein disaggregated model is with training sample
The characteristic value of data is as input, sales volume information as mark value training.The classification results of disaggregated model include for selection
Shop selects alternative commodity and not for the alternative commodity of selection shop selection.
The case where feature used in training stage and forecast period is consistent, and characteristic value is according to each commodity is true
It is fixed.When selecting feature, can to the feature in commodity data and user behavior data on line carry out Fusion Features with obtain into
Used feature when row training and prediction, so that feature is provided with the attribute of online and offline simultaneously, it can preferably be line
Commodity in the selection line of lower shop.
In step s 106, determined whether according to the classification results of disaggregated model to select alternative commodity to selection shop.
Method through the foregoing embodiment, can by under commodity data on line and line to the user around selection shop
Data determine the characteristic value of alternative commodity, and in advance trained disaggregated model is used to be classified to alternative commodity, with determination
It whether is that the alternative commodity are selected to selection shop, so as to from quickly determining shop needs under line on magnanimity line in commodity
The commodity for adopting pin improve selection efficiency and accuracy rate.
In some embodiments, commodity data on line, user behavior data particular content can be with reference table 1.This field
Technology can according to need one or more group data in selection following table, or also can choose other numbers except table 1
According to.
Table 1
It, can be according to commodity sign by different spies after feature on obtaining line in commodity data and user behavior data
Sign carries out fusion treatment.One example of the feature after integration can be as shown in table 2.In table 2, whether last is arranged " smooth
Pin " is mark value, and each column in addition to first row and last column indicate a feature.
Table 2
In some embodiments, feature can further be screened after primarily determining progress feature.Below with reference to
Fig. 2A and 2B describes the embodiment of feature of present invention screening technique.
Fig. 2A is the exemplary process diagram according to the Feature Selection method of some embodiments of the invention.As shown in Figure 2 A, should
The Feature Selection method of embodiment includes step S2012~S2014.
In step S2012, the correlation between different characteristic is calculated.
The method of one embodiment of correlation calculations is the covariance calculated between two variables.If two variables
Variation reaches unanimity, then the covariance of the two is positive；If the variation tendency of two variables on the contrary, if covariance be negative value；Such as
Two variable independences of fruit, then covariance is 0.Formula (1) is an example of covariance calculation method.
In formula (1), n indicates total amount of data, and i indicates the mark of data；X and Y respectively indicates two different variables,
Such as indicate two features, XiAnd YiI-th of value of two variables is respectively indicated,WithRespectively indicate the equal of two variables
Value；Cov (X, Y) indicates the covariance of two variables.
In step S2014, correlation is screened out greater than a feature in the different characteristic of preset value.For example, can root
Retain main feature according to business demand.
It is thus possible to only retain a feature among similar multiple features, computational efficiency is improved.
Fig. 2 B is the exemplary process diagram according to the Feature Selection method of other embodiments of the invention.As shown in Figure 2 B,
The Feature Selection method of the embodiment includes step S2022~S2024.
In step S2022, the correlation between feature and sales volume information is calculated.The calculation method of correlation still can be with
Reference formula (1).At this point, one of them is characterized in two variable Xs and Y of formula (1), another is sales volume information, marks
Note value.
In step S2024, the feature for being less than preset value with the correlation of sales volume information is screened out.
It is thus possible to which removing influences lesser feature to result, the accuracy of classification is improved.
Below with reference to the embodiment of Fig. 3 interpretive classification model training method.
Fig. 3 is the exemplary process diagram according to the disaggregated model training method of some embodiments of the invention.As shown in figure 3,
The disaggregated model training method of the embodiment includes step S302~S306.
In step s 302, commodity data and user behavior data on the line for training are obtained.
In step s 304, feature is carried out to commodity data on the line of acquisition and user behavior data according to commodity sign to melt
It closes, generates training sample data, wherein training sample data include the corresponding characteristic value of each commodity sign and sales volume information.
Sales volume information can be the data such as sales volume, sales volume；Can also be " situation of selling well ", " non-situation of selling well " etc. two classification or
Polytypic label, for example, SKU mark value with " situation of selling well " of day sales volume at 1000 or more can be preset.
Sales volume information can be that data under pure line, data or both have both at the same time on pure line, and those skilled in the art can be with
It is selected as needed.It is thus possible to according to commodity, whether fast sale decides whether to select the commodity for shop under line.
In step S306, using the characteristic value of training sample data as input, corresponding sales volume information as label
Value is trained disaggregated model.Disaggregated model can be the models such as decision tree, neural network, logistic regression.
By taking decision tree as an example, the classification node in categorised decision tree that training is completed can be the feature of commodity, leaf knot
Point can be with presentation class result.Fig. 4 is the categorised decision tree schematic diagram in some embodiments of the invention.As shown in figure 4, if one
The amount of collection of a commodity is greater than preset value x and is import, then is classified as " selection commodity "；If a commodity
Amount of collection be greater than preset value x, be not import, low to crowd's preference under the line around selection shop, then classified
For " not selecting commodity ".
The embodiment of categorised decision tree training method of the present invention is described below with reference to Fig. 5 A.
Fig. 5 A is the exemplary process diagram according to the categorised decision tree training method of some embodiments of the invention.Such as Fig. 5 A institute
Show, the categorised decision tree training method of the embodiment includes step S502~S506.
In step S502, according to training sample data corresponding to site position to be added, calculating is not added into classification
The information gain of feature in decision tree.Training sample data corresponding to site position to be added refer to using categorised decision tree
Remaining data when be classified to current location for all training sample data.
When node to be added is root node, due to not yet carrying out any classification before root node is classified,
The corresponding training sample data of root node are whole training sample data used by training book categorised decision tree.
When node to be added is not root node, training sample data corresponding to site position to be added are knot to be added
Sorted training sample data are carried out using father node feature in the corresponding training sample data of father node of point.For example, father
Node corresponding data A, B, C, D, father node feature be " whether quantity in stock is greater than preset value y ", then quantity in stock greater than y data A,
B can be the corresponding data in first of father node site position, and data C, D of the quantity in stock less than or equal to y can be father's knot
The corresponding data in the sub- site position of second of point.
In step S504, the maximum feature of information gain is added to site position to be added.
It is then possible to be classified according to newly added feature to the corresponding training sample data in new addition site position.
To obtain training sample data corresponding to site position to be added in next circulation.
In step S506, judgement is current, and whether there are also features to be not added in decision tree.If so, returning to step
S502；If it is not, terminating training process, and the leaf node of presentation class result is added for current class decision tree.Leaf
Classification results represented by child node are determined according to the mark value of the corresponding training sample data in site position to be added.That is, such as
Fruit leaf node is compared with fraternal leaf node, should if the commodity data that mark value is A is more or accounting is more
The classification results of leaf node are set as A.
It in some embodiments, can not be that node addition to be added is special when the information gain of feature is lesser
Sign.The embodiment of categorised decision tree training method of the present invention is described below with reference to Fig. 5 B.
Fig. 5 B is the exemplary process diagram according to the categorised decision tree training method of other embodiments of the invention.Such as Fig. 5 B
Shown, the categorised decision tree training method of the embodiment includes step S512~S520.
In step S512, according to training sample data corresponding to site position to be added, calculating is not added into classification
The information gain of feature in decision tree.
In step S514, judge whether that the information gain of each feature being not added into categorised decision tree is respectively less than pre-
If information gain threshold.If so, executing step S516；If it is not, executing step S518.
In step S516, to the leaf node of site position to be added addition presentation class result, wherein classification results
It is determined according to the corresponding mark value in site position to be added.
In step S518, the maximum feature of information gain is added to site position to be added, as classification node.
In step S520, judgement is current, and whether there are also features to be not added in decision tree.If so, returning to step
S512；If it is not, terminating training process, and the leaf of presentation class result is added for the classification node of current class decision tree
Child node.
Illustratively a kind of calculation method of recommended information gain below.If for training sample data collection S and feature T, S
In include X positive sample and Y negative sample, positive sample and negative sample respectively indicate different final classifications as a result, for example distinguishing table
Show best-selling product and non-best-selling product.Classification marker for feature T is carried out to data set S, such as sets T and represents " amount of collection ",
The positive sample X1 item of amount of collection "high", negative sample Y1 item；The positive sample X2 item of amount of collection " low ", negative sample Y2 item.Then based on spy
T is levied, the mode for calculating the information gain-ratio of feature T is as follows.
Firstly, calculating the comentropy Info (S) of data set S using formula (2).
Then, the comentropy Info (T) of feature T is calculated using formula (3).
Finally, calculating the information gain-ratio GainRatio (T) of feature T using formula (4) and (5), wherein SplitInfo
It (T) is intermediate variable.
Method through the foregoing embodiment can preferentially be added to categorised decision tree for big node is influenced on classification results
In, so that the disaggregated model trained is more accurate, and then improve the accuracy of selection.
In some embodiments, multiple models can be trained in advance.Categorised decision tree instruction of the present invention is described below with reference to Fig. 6
Practice the embodiment of method.
Fig. 6 is the exemplary process diagram according to the categorised decision tree training method of other embodiments of the invention.Such as Fig. 6 institute
Show, the categorised decision tree training method of the embodiment includes step S602~S606.
In step S602, multiple training sample data subsets are obtained, wherein each training sample data subset is corresponding
Commodity belong to same category.For example, the data in the same training sample data subset may belong to same category, same region
Classification etc..
In step s 604, each training sample data trained disaggregated model is respectively adopted, obtains multiple subclassifications
Model.Specific training method can refer to previous embodiment, and which is not described herein again.
In step S606, according to the predictablity rate of each subclassification model, the weight of each subclassification model is determined,
To be determined whether according to the weighing computation results of the classification results of each subclassification model for the alternative quotient of selection shop selection
Product.The weight of subclassification model can be with its predictablity rate correlation.
It is thus possible to according to multiple models are trained the characteristics of different classes of commodity in advance, and according to the prediction of multiple models
As a result it synthetically determines prediction result, improves the accuracy of selection.
In some embodiments, marked data gathered in advance can be divided into training set and test set.It is adopting
After training set training pattern, it can be tested using test set to disaggregated model.Present invention classification is described below with reference to Fig. 7
The embodiment of model method of adjustment.
Fig. 7 is the exemplary process diagram according to the disaggregated model method of adjustment of some embodiments of the invention.As shown in fig. 7,
The disaggregated model method of adjustment of the embodiment includes step S702~S706.
In step S702, the characteristic value for testing commodity is input in categorised decision tree.
In step S704, categorised decision tree is determined according to the mark value of the classification results of categorised decision tree and test commodity
Predictablity rate.
In step S706, it is lower than preset accuracy rate threshold value in response to predictablity rate, executes following at least one behaviour
Re -training categorised decision tree after work: modification information gain threshold；More new feature；Update training sample data.
Since information gain threshold determines that current location is the node for adding feature node or presentation class result, because
This can be adjusted the structure of decision tree by modification information gain threshold；More new feature, which can be, to be increased feature, reduces
Feature or replacement feature, so as to be predicted according to different data characteristics；Updating training sample data can be increasing
Addend evidence reduces data or replacement data, so as to reject Outliers, store the user that can more represent around shop
The data of hobby.
It is tested and is adjusted by the disaggregated model completed to training, the predictablity rate of model can be improved, into one
Step improves the accuracy and selection efficiency of selection.
The embodiment of selection device of the present invention is described below with reference to Fig. 8.
Fig. 8 is the exemplary block diagram according to the selection device of some embodiments of the invention.As shown in figure 8, the embodiment
Selection device 80 include: alternative product features value determining module 810, be configured as commodity number on the line according to alternative commodity
According to, with the user behavior data for being less than user on the line of preset value to distance under the line in selection shop determine the features of alternative commodity
Value；Alternative product features value input module 820, is configured as the characteristic value of alternative commodity inputting disaggregated model, wherein classification
Model is to be trained using the characteristic value of training sample data as input, sales volume information as mark value；Selection determining module
830, it is configured as being determined whether according to the classification results of disaggregated model to select alternative commodity to selection shop.
In some embodiments, selection device 80 further include: disaggregated model training module 840 is configured as acquisition and is used for
Commodity data and user behavior data on trained line；According to commodity sign to commodity data on the line of acquisition and user behavior number
According to Fusion Features are carried out, training sample data are generated, wherein training sample data include the corresponding characteristic value of each commodity sign
With sales volume information；It is used as input, corresponding sales volume information as mark value to classification mould using the characteristic value of training sample data
Type is trained.
In some embodiments, disaggregated model is categorised decision tree, and the classification node in categorised decision tree is the spy of commodity
Sign, leaf node presentation class result.
In some embodiments, selection device 80 further include: categorised decision tree training module 850 is configured as obtaining and use
In commodity data and user behavior data on trained line；According to commodity sign to commodity data and user behavior on the line of acquisition
Data carry out Fusion Features, generate training sample data, wherein training sample data include the corresponding feature of each commodity sign
Value and sales volume information；Disaggregated model is trained, comprising: according to training sample data corresponding to site position to be added,
The information gain for calculating the feature being not added into categorised decision tree, is added to node to be added for the maximum feature of information gain
Position, as classification node, repeat above the step of calculating information gain, addition feature until all features are all added to classification
In decision tree.
In some embodiments, selection device 80 further include: categorised decision tree training module 850 is configured as obtaining and use
In commodity data and user behavior data on trained line；According to commodity sign to commodity data and user behavior on the line of acquisition
Data carry out Fusion Features, generate training sample data, wherein training sample data include the corresponding feature of each commodity sign
Value and sales volume information；Disaggregated model is trained, comprising: according to training sample data corresponding to site position to be added,
The information gain for calculating the feature being not added into categorised decision tree, in the letter of each feature being not added into categorised decision tree
Breath gain be respectively less than preset information gain threshold in the case where, to site position to be added addition presentation class result leaf
Node, the case where the information gain for the feature being not added into categorised decision tree is not respectively less than preset information gain threshold
Under, the maximum feature of information gain is added to site position to be added as classification node, repeat it is above calculate information gain,
The step of adding feature is all added in categorised decision tree until all features；Wherein, classification results are according to node position to be added
Corresponding mark value is set to determine.
In some embodiments, selection device 80 further include: disaggregated model testing and debugging module 860 is configured as to survey
The characteristic value of examination commodity is input in categorised decision tree；It is true according to the classification results of categorised decision tree and the mark value of test commodity
Determine the predictablity rate of categorised decision tree；It is lower than preset accuracy rate threshold value in response to predictablity rate, executes following at least one
Re -training categorised decision tree after kind operation: modification information gain threshold；More new feature；Update training sample data.
In some embodiments, selection device 80 further include: multi-model training module 870 is configured as obtaining multiple instructions
Practice sample data subset, wherein the corresponding commodity of each training sample data subset belong to same category；Each instruction is respectively adopted
Practice sample data trained disaggregated model, obtains multiple subclassification models；According to the predictablity rate of each subclassification model,
The weight of each subclassification model is determined, to be according to the determination of the weighing computation results of the classification results of each subclassification model
No is to select alternative commodity to selection shop.
In some embodiments, selection device 80 further include: it is following at least to be configured as execution for Feature Selection module 880
A kind of Feature Selection method: correlation is screened out greater than a feature in the different characteristic of preset value；It screens out and sales volume information
Correlation is less than the feature of preset value.
In some embodiments, commodity data comprises at least one of the following on line: bought on line feature, browse on line it is special
Sign, search characteristics on line add shopping cart feature, brand sales volume feature on line, evaluating characteristic on line, primary attribute on line on line
Feature is bought on the line of user in preset range under feature, line；User behavior data comprises at least one of the following: user location is special
It is special that sign, user buy feature, user's browsing feature, user's search characteristics, user's addition shopping cart feature, user's purchase brand
Sign, user base attributive character.
Fig. 9 is the exemplary block diagram according to the selection device of other embodiments of the invention.As shown in figure 9, the implementation
The selection device 900 of example includes: memory 910 and the processor 920 for being coupled to the memory 910, and processor 920 is configured
To execute the selection method in any one aforementioned embodiment based on the instruction being stored in memory 910.
Wherein, memory 910 is such as may include system storage, fixed non-volatile memory medium.System storage
Device is for example stored with operating system, application program, Boot loader (Boot Loader) and other programs etc..
Figure 10 is the exemplary block diagram according to the selection device of yet other embodiments of the invention.As shown in Figure 10, the reality
The selection device 1000 for applying example includes: memory 1010 and processor 1020, can also include input/output interface 1030, net
Network interface 1040, memory interface 1050 etc..These interfaces 1030,1040,1050 and memory 1010 and processor 1020 it
Between can for example be connected by bus 1060.Wherein, input/output interface 1030 is that display, mouse, keyboard, touch screen etc. are defeated
Enter output equipment and connecting interface is provided.Network interface 1040 provides connecting interface for various networked devices.Memory interface 1050 is
The external storages such as SD card, USB flash disk provide connecting interface.
The embodiment of the present invention also provides a kind of computer readable storage medium, is stored thereon with computer program, special
Sign is that the program realizes any one aforementioned selection method when being executed by processor.
Those skilled in the art should be understood that the embodiment of the present invention can provide as method, system or computer journey
Sequence product.Therefore, complete hardware embodiment, complete software embodiment or combining software and hardware aspects can be used in the present invention
The form of embodiment.Moreover, it wherein includes the calculating of computer usable program code that the present invention, which can be used in one or more,
Machine can use the meter implemented in non-transient storage medium (including but not limited to magnetic disk storage, CD-ROM, optical memory etc.)
The form of calculation machine program product.
The present invention be referring to according to the method for the embodiment of the present invention, the process of equipment (system) and computer program product
Figure and/or block diagram describe.It is interpreted as to be realized by computer program instructions each in flowchart and/or the block diagram
The combination of process and/or box in process and/or box and flowchart and/or the block diagram.It can provide these computer journeys
Sequence instruct to general purpose computer, special purpose computer, Embedded Processor or other programmable data processing devices processor with
A machine is generated, so that the instruction generation executed by computer or the processor of other programmable data processing devices is used for
Realize the dress for the function of specifying in one or more flows of the flowchart and/or one or more blocks of the block diagram
It sets.
These computer program instructions, which may also be stored in, is able to guide computer or other programmable data processing devices with spy
Determine in the computer-readable memory that mode works, so that it includes referring to that instruction stored in the computer readable memory, which generates,
Enable the manufacture of device, the command device realize in one box of one or more flows of the flowchart and/or block diagram or
The function of being specified in multiple boxes.
These computer program instructions also can be loaded onto a computer or other programmable data processing device, so that counting
Series of operation steps are executed on calculation machine or other programmable devices to generate computer implemented processing, thus in computer or
The instruction executed on other programmable devices is provided for realizing in one or more flows of the flowchart and/or block diagram one
The step of function of being specified in a box or multiple boxes.
The foregoing is merely presently preferred embodiments of the present invention, is not intended to limit the invention, it is all in spirit of the invention and
Within principle, any modification, equivalent replacement, improvement and so on be should all be included in the protection scope of the present invention.
Claims (20)
1. a kind of selection method, comprising:
It is less than the use of user on the line of preset value according to commodity data on the line of alternative commodity and to distance under the line in selection shop
Family behavioral data determines the characteristic value of alternative commodity；
The characteristic value of the alternative commodity is inputted into disaggregated model, wherein the disaggregated model is with the spy of training sample data
Value indicative is as input, sales volume information as mark value training；
Determined whether to select the alternative commodity to selection shop to be described according to the classification results of the disaggregated model.
2. selection method according to claim 1, further includes:
Obtain commodity data and user behavior data on the line for training；
Fusion Features are carried out to commodity data on the line of acquisition and user behavior data according to commodity sign, generate number of training
According to, wherein the training sample data include the corresponding characteristic value of each commodity sign and sales volume information；
Disaggregated model is instructed as mark value as input, corresponding sales volume information using the characteristic value of training sample data
Practice.
3. selection method according to claim 1, wherein the disaggregated model is categorised decision tree, the categorised decision
Classification node in tree is the feature of commodity, leaf node presentation class result.
4. selection method according to claim 3, further includes:
Obtain commodity data and user behavior data on the line for training；
Fusion Features are carried out to commodity data on the line of acquisition and user behavior data according to commodity sign, generate number of training
According to, wherein the training sample data include the corresponding characteristic value of each commodity sign and sales volume information；
Disaggregated model is trained, comprising:
According to training sample data corresponding to site position to be added, the letter for the feature being not added into categorised decision tree is calculated
Cease gain；
The maximum feature of information gain is added to the site position to be added, as classification node；
Above the step of calculating information gain, addition feature is repeated, until all features are all added in categorised decision tree.
5. selection method according to claim 3, further includes:
Obtain commodity data and user behavior data on the line for training；
Fusion Features are carried out to commodity data on the line of acquisition and user behavior data according to commodity sign, generate number of training
According to, wherein the training sample data include the corresponding characteristic value of each commodity sign and sales volume information；
Disaggregated model is trained, comprising:
According to training sample data corresponding to site position to be added, the letter for the feature being not added into categorised decision tree is calculated
Cease gain；
The case where the information gain of each feature being not added into categorised decision tree is respectively less than preset information gain threshold
Under, to the leaf node of site position to be added addition presentation class result, wherein classification results are according to the node to be added
The corresponding mark value in position determines；
The case where the information gain for the feature being not added into categorised decision tree is not respectively less than preset information gain threshold
Under, the maximum feature of information gain is added to the site position to be added, as classification node；
Above the step of calculating information gain, addition feature is repeated, until all features are all added in categorised decision tree.
6. selection method according to claim 4, further includes:
The characteristic value for testing commodity is input in categorised decision tree；
The predictablity rate of categorised decision tree is determined according to the mark value of the classification results of categorised decision tree and test commodity；
It is lower than preset accuracy rate threshold value in response to the predictablity rate, executes re -training point after following at least one operation
Class decision tree: modification information gain threshold, more new feature update training sample data.
7. selection method according to claim 1, further includes:
Obtain multiple training sample data subsets, wherein the corresponding commodity of each training sample data subset belong to same category；
Each training sample data trained disaggregated model is respectively adopted, obtains multiple subclassification models；
According to the predictablity rate of each subclassification model, the weight of each subclassification model is determined, so as to according to every height point
The weighing computation results of the classification results of class model determine whether to select alternative commodity to selection shop to be described.
8. selection method according to any one of claims 1 to 7 further includes executing following at least one Feature Selection side
Method:
Correlation is screened out greater than a feature in the different characteristic of preset value；
Screen out the feature for being less than preset value with the correlation of sales volume information.
9. selection method according to any one of claims 1 to 7, wherein
Commodity data comprises at least one of the following on line: buying feature on line, browses feature, search characteristics on line on line, on line
Add shopping cart feature, brand sales volume feature on line, evaluating characteristic on line, on line under primary attribute feature, line in preset range
Feature is bought on the line of user；
User behavior data comprises at least one of the following: user location feature, user buy feature, user browses feature, user
Search characteristics, user add shopping cart feature, user buys brand identity, user base attributive character.
10. a kind of selection device, comprising:
Alternative product features value determining module is configured as commodity data on the line according to alternative commodity and to selection shop
The user behavior data that distance is less than user on the line of preset value under line determines the characteristic value of alternative commodity；
Alternative product features value input module is configured as the characteristic value of the alternative commodity inputting disaggregated model, wherein institute
Stating disaggregated model is to be trained using the characteristic value of training sample data as input, sales volume information as mark value；
Selection determining module is configured as being determined whether to select to selection shop to be described according to the classification results of the disaggregated model
Select the alternative commodity.
11. selection device according to claim 10, further includes:
Disaggregated model training module is configured as obtaining commodity data and user behavior data on the line for training；According to quotient
Product mark carries out Fusion Features to commodity data on the line of acquisition and user behavior data, generates training sample data, wherein institute
Stating training sample data includes the corresponding characteristic value of each commodity sign and sales volume information；Using the characteristic value of training sample data
Disaggregated model is trained as mark value as input, corresponding sales volume information.
12. selection device according to claim 10, wherein the disaggregated model is categorised decision tree, and the classification is determined
Classification node in plan tree is the feature of commodity, leaf node presentation class result.
13. selection device according to claim 12, further includes:
Categorised decision tree training module is configured as obtaining commodity data and user behavior data on the line for training；According to
Commodity sign carries out Fusion Features to commodity data on the line of acquisition and user behavior data, generates training sample data, wherein
The training sample data include the corresponding characteristic value of each commodity sign and sales volume information；Disaggregated model is trained, is wrapped
It includes: according to the letter for the feature that training sample data corresponding to site position to be added, calculating are not added into categorised decision tree
Gain is ceased, the maximum feature of information gain is added to the site position to be added, as classification node, repeats above calculate
The step of information gain, addition feature, is all added in categorised decision tree until all features.
14. selection device according to claim 12, further includes:
Categorised decision tree training module is configured as obtaining commodity data and user behavior data on the line for training；According to
Commodity sign carries out Fusion Features to commodity data on the line of acquisition and user behavior data, generates training sample data, wherein
The training sample data include the corresponding characteristic value of each commodity sign and sales volume information；Disaggregated model is trained, is wrapped
It includes: according to the letter for the feature that training sample data corresponding to site position to be added, calculating are not added into categorised decision tree
Gain is ceased, is respectively less than the feelings of preset information gain threshold in the information gain of each feature being not added into categorised decision tree
The leaf node that presentation class result is added under condition, to site position to be added, in the feature being not added into categorised decision tree
Information gain be respectively less than preset information gain threshold in the case where, the maximum feature of information gain is added to it is described
Site position to be added as classification node, repeat above the step of calculating information gain, addition feature until all features all
It is added in categorised decision tree；
Wherein, classification results are determined according to the corresponding mark value in the site position to be added.
15. selection device according to claim 13, further includes:
Disaggregated model testing and debugging module is configured as the characteristic value for testing commodity being input in categorised decision tree；According to point
The classification results of class decision tree and the mark value of test commodity determine the predictablity rate of categorised decision tree；In response to the prediction
Accuracy rate is lower than preset accuracy rate threshold value, executes re -training categorised decision tree after following at least one operation: modification information
Gain threshold；More new feature；Update training sample data.
16. selection device according to claim 10, further includes:
Multi-model training module is configured as obtaining multiple training sample data subsets, wherein each training sample data subset
Corresponding commodity belong to same category；Each training sample data trained disaggregated model is respectively adopted, obtains multiple sons point
Class model；According to the predictablity rate of each subclassification model, the weight of each subclassification model is determined, so as to according to every height
The weighing computation results of the classification results of disaggregated model determine whether to select alternative commodity to selection shop to be described.
17. selection device described in any one of 0~16 according to claim 1, further includes:
Feature Selection module is configured as executing following at least one Feature Selection method: screening out correlation greater than preset value
A feature in different characteristic；Screen out the feature for being less than preset value with the correlation of sales volume information.
18. selection device described in any one of 0~16 according to claim 1, wherein
Commodity data comprises at least one of the following on line: buying feature on line, browses feature, search characteristics on line on line, on line
Add shopping cart feature, brand sales volume feature on line, evaluating characteristic on line, on line under primary attribute feature, line in preset range
Feature is bought on the line of user；
User behavior data comprises at least one of the following: user location feature, user buy feature, user browses feature, user
Search characteristics, user add shopping cart feature, user buys brand identity, user base attributive character.
19. a kind of selection device, comprising:
Memory；And
It is coupled to the processor of the memory, the processor is configured to the instruction based on storage in the memory,
Execute such as selection method according to any one of claims 1 to 9.
20. a kind of computer readable storage medium, is stored thereon with computer program, power is realized when which is executed by processor
Benefit require any one of 1~9 described in selection method.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810693419.8A CN108960719B (en) | 2018-06-29 | 2018-06-29 | Method and device for selecting products and computer readable storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810693419.8A CN108960719B (en) | 2018-06-29 | 2018-06-29 | Method and device for selecting products and computer readable storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108960719A true CN108960719A (en) | 2018-12-07 |
| CN108960719B CN108960719B (en) | 2021-10-15 |
Family
ID=64487825
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810693419.8A Active CN108960719B (en) | 2018-06-29 | 2018-06-29 | Method and device for selecting products and computer readable storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108960719B (en) |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111340520A (en) * | 2018-12-19 | 2020-06-26 | 北京京东尚科信息技术有限公司 | Method and apparatus for generating information |
| CN111445134A (en) * | 2020-03-26 | 2020-07-24 | 珠海随变科技有限公司 | Commodity sales prediction method, commodity sales prediction apparatus, computer device, and storage medium |
| CN111768139A (en) * | 2019-06-27 | 2020-10-13 | 北京沃东天骏信息技术有限公司 | Stock processing method, apparatus, device and storage medium |
| CN111860299A (en) * | 2020-07-17 | 2020-10-30 | 北京奇艺世纪科技有限公司 | Target object grade determining method and device, electronic equipment and storage medium |
| CN112016949A (en) * | 2019-05-31 | 2020-12-01 | 口碑(上海)信息技术有限公司 | Matching method and device for verification and cancellation area, storage medium and electronic device |
| CN112465533A (en) * | 2019-09-09 | 2021-03-09 | 中国移动通信集团河北有限公司 | Intelligent product selection method and device and computing equipment |
| CN112613953A (en) * | 2020-12-29 | 2021-04-06 | 北京环球国广媒体科技有限公司 | Commodity selection method, system and computer readable storage medium |
| CN112906723A (en) * | 2019-11-19 | 2021-06-04 | 北京京邦达贸易有限公司 | Feature selection method and device |
| CN112967109A (en) * | 2021-03-15 | 2021-06-15 | 杭州邻汇网络科技有限公司 | Intelligent flash shop product selection method |
| CN113269607A (en) * | 2021-05-21 | 2021-08-17 | 深圳市方圆展示制品有限公司 | Online and offline experience smart display system |
| CN113434611A (en) * | 2021-07-08 | 2021-09-24 | 宁波权智科技有限公司 | Sales classification method and device based on artificial intelligence |
| CN113507623A (en) * | 2021-07-09 | 2021-10-15 | 广东蕾特恩科技发展有限公司 | Intelligent commodity selection system and method based on data analysis |
| CN114117245A (en) * | 2022-01-26 | 2022-03-01 | 深圳市云中鹤科技股份有限公司 | Product screening method and device based on big data |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2016060307A1 (en) * | 2014-10-17 | 2016-04-21 | 주식회사 스타트업팩토리 | Purchase service system using product unique code, and method therefor |
| CN106485562A (en) * | 2015-09-01 | 2017-03-08 | 苏宁云商集团股份有限公司 | A kind of commodity information recommendation method based on user's history behavior and system |
| CN107292666A (en) * | 2017-06-20 | 2017-10-24 | 北京京东尚科信息技术有限公司 | Sales potential determination methods and device |
| CN107403345A (en) * | 2017-09-22 | 2017-11-28 | 北京京东尚科信息技术有限公司 | Best-selling product Forecasting Methodology and system, storage medium and electric terminal |
| CN107578332A (en) * | 2017-09-22 | 2018-01-12 | 深圳乐信软件技术有限公司 | A kind of method, apparatus, equipment and storage medium for recommending cash commodity |
| CN107767103A (en) * | 2017-11-03 | 2018-03-06 | 李良浩 | The method and purchase system of on-line off-line quick shopping are realized based on mobile Internet |
| CN107944913A (en) * | 2017-11-21 | 2018-04-20 | 重庆邮电大学 | High potential user's purchase intention Forecasting Methodology based on big data user behavior analysis |
-
2018
- 2018-06-29 CN CN201810693419.8A patent/CN108960719B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2016060307A1 (en) * | 2014-10-17 | 2016-04-21 | 주식회사 스타트업팩토리 | Purchase service system using product unique code, and method therefor |
| CN106485562A (en) * | 2015-09-01 | 2017-03-08 | 苏宁云商集团股份有限公司 | A kind of commodity information recommendation method based on user's history behavior and system |
| CN107292666A (en) * | 2017-06-20 | 2017-10-24 | 北京京东尚科信息技术有限公司 | Sales potential determination methods and device |
| CN107403345A (en) * | 2017-09-22 | 2017-11-28 | 北京京东尚科信息技术有限公司 | Best-selling product Forecasting Methodology and system, storage medium and electric terminal |
| CN107578332A (en) * | 2017-09-22 | 2018-01-12 | 深圳乐信软件技术有限公司 | A kind of method, apparatus, equipment and storage medium for recommending cash commodity |
| CN107767103A (en) * | 2017-11-03 | 2018-03-06 | 李良浩 | The method and purchase system of on-line off-line quick shopping are realized based on mobile Internet |
| CN107944913A (en) * | 2017-11-21 | 2018-04-20 | 重庆邮电大学 | High potential user's purchase intention Forecasting Methodology based on big data user behavior analysis |
Non-Patent Citations (1)
| Title |
|---|
| 费斐 等: "决策树算法在团购商品销售预测中的应用", 《计算机系统应用》 * |
Cited By (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111340520A (en) * | 2018-12-19 | 2020-06-26 | 北京京东尚科信息技术有限公司 | Method and apparatus for generating information |
| CN112016949A (en) * | 2019-05-31 | 2020-12-01 | 口碑(上海)信息技术有限公司 | Matching method and device for verification and cancellation area, storage medium and electronic device |
| CN111768139B (en) * | 2019-06-27 | 2023-04-07 | 北京沃东天骏信息技术有限公司 | Stock processing method, apparatus, device and storage medium |
| CN111768139A (en) * | 2019-06-27 | 2020-10-13 | 北京沃东天骏信息技术有限公司 | Stock processing method, apparatus, device and storage medium |
| CN112465533A (en) * | 2019-09-09 | 2021-03-09 | 中国移动通信集团河北有限公司 | Intelligent product selection method and device and computing equipment |
| CN112906723A (en) * | 2019-11-19 | 2021-06-04 | 北京京邦达贸易有限公司 | Feature selection method and device |
| CN112906723B (en) * | 2019-11-19 | 2024-01-16 | 北京京邦达贸易有限公司 | Feature selection method and device |
| CN111445134B (en) * | 2020-03-26 | 2021-05-28 | 珠海必要工业科技股份有限公司 | Commodity sales prediction method, commodity sales prediction apparatus, computer device, and storage medium |
| CN111445134A (en) * | 2020-03-26 | 2020-07-24 | 珠海随变科技有限公司 | Commodity sales prediction method, commodity sales prediction apparatus, computer device, and storage medium |
| CN111860299A (en) * | 2020-07-17 | 2020-10-30 | 北京奇艺世纪科技有限公司 | Target object grade determining method and device, electronic equipment and storage medium |
| CN111860299B (en) * | 2020-07-17 | 2023-09-08 | 北京奇艺世纪科技有限公司 | Method and device for determining grade of target object, electronic equipment and storage medium |
| CN112613953A (en) * | 2020-12-29 | 2021-04-06 | 北京环球国广媒体科技有限公司 | Commodity selection method, system and computer readable storage medium |
| CN112967109A (en) * | 2021-03-15 | 2021-06-15 | 杭州邻汇网络科技有限公司 | Intelligent flash shop product selection method |
| CN113269607A (en) * | 2021-05-21 | 2021-08-17 | 深圳市方圆展示制品有限公司 | Online and offline experience smart display system |
| CN113434611A (en) * | 2021-07-08 | 2021-09-24 | 宁波权智科技有限公司 | Sales classification method and device based on artificial intelligence |
| CN113507623A (en) * | 2021-07-09 | 2021-10-15 | 广东蕾特恩科技发展有限公司 | Intelligent commodity selection system and method based on data analysis |
| CN114117245A (en) * | 2022-01-26 | 2022-03-01 | 深圳市云中鹤科技股份有限公司 | Product screening method and device based on big data |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108960719B (en) | 2021-10-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108960719A (en) | Selection method and apparatus and computer readable storage medium | |
| CN111444334B (en) | Data processing method, text recognition device and computer equipment | |
| CN106485562B (en) | Commodity information recommendation method and system based on user historical behaviors | |
| CN109902708A (en) | A kind of recommended models training method and relevant apparatus | |
| CN104077306B (en) | The result ordering method and system of a kind of search engine | |
| CN108665355A (en) | Financial product recommends method, apparatus, equipment and computer storage media | |
| CN109800885A (en) | It is determined for the rule of black box machine learning model | |
| CN108205768A (en) | Database building method and data recommendation method and device, equipment and storage medium | |
| CN104798043B (en) | A kind of data processing method and computer system | |
| CN109583468A (en) | Training sample acquisition methods, sample predictions method and corresponding intrument | |
| CN110084374A (en) | Construct method, apparatus and prediction technique, device based on the PU model learnt | |
| CN110428298A (en) | A kind of shop recommended method, device and equipment | |
| CN106651519A (en) | Method and system for personalized recommendation based on label information | |
| CN110163647A (en) | A kind of data processing method and device | |
| CN109903111A (en) | For the sort method of personalized recommendation, order models training method and ordering system | |
| CN109582876A (en) | Tourism industry user portrait building method, device and computer equipment | |
| CN107424007A (en) | A kind of method and apparatus for building electronic ticket susceptibility identification model | |
| CN110069699B (en) | Ranking model training method and device | |
| KR102412461B1 (en) | Method for predicting demand using visual schema of product and system thereof | |
| CN109345050A (en) | A kind of quantization transaction prediction technique, device and equipment | |
| CN110458643B (en) | Repetitive commodity information recommendation method based on feature fusion and electronic equipment | |
| CN107729915A (en) | For the method and system for the key character for determining machine learning sample | |
| CN105468628B (en) | A kind of sort method and device | |
| CN110580489B (en) | Data object classification system, method and equipment | |
| Ray et al. | Short-term load forecasting using genetic algorithm |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |