CN109741113A - A kind of user's purchase intention prediction technique based on big data - Google Patents
A kind of user's purchase intention prediction technique based on big data Download PDFInfo
- Publication number
- CN109741113A CN109741113A CN201910021419.8A CN201910021419A CN109741113A CN 109741113 A CN109741113 A CN 109741113A CN 201910021419 A CN201910021419 A CN 201910021419A CN 109741113 A CN109741113 A CN 109741113A
- Authority
- CN
- China
- Prior art keywords
- user
- model
- feature
- commodity
- purchase
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Abstract
The present invention relates to machine learning, big data technical field, in particular to a kind of user's purchase intention prediction technique based on big data, including pretreatment operation is carried out to data set；Training set, verifying collection and test set operation are divided according to data set；Construction feature Engineering operation；Feature selecting operation is carried out to the Feature Engineering of building；Machine learning model is established according to the feature of selection and carries out Model Fusion operation；By the model of building, predicts that user is following and whether buy a specified commodity operation in；The present invention is by evaluating data and user behavior data progress data prediction and analysis extraction feature to user basic information data, commodity essential information data, commodity, it establishes machine learning model and carries out Model Fusion operation, to accurately predict the purchasing demand in user's future, the matching result of output high latent user and end article, the target group of high quality are provided for precision marketing, while also providing shopping experience that is simpler, quick, saving worry for electric business user.
Description
Technical field
The present invention relates to machine learning, big data technical field, in particular to a kind of user based on big data buys meaning
To prediction technique.
Background technique
In recent years, domestic each electric business platform emerges rapidly, and many electric business platforms are while keeping high speed development, precipitating
Several hundred million loyal users, has accumulated the truthful data of magnanimity.How rule is found out from historical data, go prediction user's future
Purchasing demand, allow most suitable commodity to meet the most desirable people, be that big data applies the critical issue in precision marketing,
It is all electric business platforms required core technology when doing intelligentized updating.
The present invention passes through the algorithm of big data technology and machine learning, and building user buys the prediction model of commodity, output
The matching result of height latent user and end article, provide the target group of high quality for precision marketing.Meanwhile mining data is behind
Potential meaning provides shopping experience that is simpler, quick, saving worry for electric business user.
Summary of the invention
The present invention proposes that a kind of user based on big data buys meaning for the critical issue in electric business platform precision marketing
To prediction technique, such as Fig. 1, comprising:
S1, pretreatment operation is carried out to data set；
S2, training set, verifying collection and test set operation are divided according to data set；
S3, construction feature Engineering operation；
S4, feature selecting operation is carried out to the Feature Engineering of building；
S5, machine learning model is established according to the feature of selection and carries out Model Fusion operation；
S6, the model by building, whether prediction user is following buys specified commodity operation in one day.
Further, described that the user that pretreatment operation includes deletion only purchaser record is carried out to data set, delete nothing
It collects or without shopping cart or without the user of buying behavior, deletes last 10 days commodity not interacted and user.
Further, dividing training set, verifying collection and test set operation according to data set included using sliding window method, with 7 days
For a cycle, length of window is 10 days progress sliding windows, to construct multiple trained windows, expands training set quantity；Wherein, it tests
The mode of 5 folding cross validations is taken in the building of card collection, wherein 4 parts are used as training data, 1 part is used as verify data, test set
Building chooses the data for predicting 10 days a few days ago as test data.
Further, construction feature Engineering operation includes basic statistical syndrome, temporal aspect group, assemblage characteristic group, industry
It is engaged in syndrome totally four groups of syndromes, foundation characteristic group mainly portrays user's basic act and commodity essential information, including user is clear
It lookes at number, user and collects number, user purchase number, user is added to buy number, goods browse number, articles storage number, commodity and add purchase number, commodity purchase
Number, user purchase conversion ratio, commodity purchasing conversion ratio are bought, temporal aspect group mainly portrays user and the time decaying of commodity becomes
Gesture, including user at the last 2nd, 3,4,5,7,10 day to the browsing number of commodity, collection number plus purchase number and the maximum for buying number
Value, minimum value, mean value, median, variance, summation, assemblage characteristic group mainly foundation characteristic combination come carry out go deep into excavate number
According to comprising information, including user-goods browse number, user-articles storage number, user-commodity add purchase number, user-commodity purchase
Buy number, user-merchandise classification browses number, user-merchandise classification collects number, user-merchandise classification adds purchase number, user-commodity class
Not Gou Mai number, user-Brand browsing number, user-Brand collection number, user-Brand add purchase number, user-quotient
Product Brand Buying number, service feature group mainly excavate user information from service layer, including commodity behavior occurs for the first time for user
Range prediction day number of days, commodity behavior range prediction day number of days, the behavior of user's first time and last row occurs for user last time
For the number of days maximum value apart from number of days, user's Continuous behavior, behavior number maximum value in one hour user's last day.
Further, feature selecting operation is carried out to the feature of building and is included in the complete rear output spy of XGBoost model training
Importance is levied, feature importance is ranked up according to this, maximum 300 features of selected characteristic importance；Pierre is used again
The similarity that inferior related coefficient calculates between 300 features filters out if two characteristic similarities have reached 95% or more
The low feature of characteristic importance in the two features.
Further, establishing machine learning model according to the feature of selection and carrying out Model Fusion operation includes that building is multiple
The biggish model of diversity factor carries out Model Fusion, it is contemplated that often performance is preferably, special to select in prediction for integrated model
The integrated model of catboost, GBDT and XGBoost totally three tree-shaped, in addition to increasing model difference degree, also selection is linear
Disaggregated model logistic regression is merged, using the feature of selection, construct logistic regression,
Catboost, GBDT and XGBoost totally four models, then the score F1 value of four models is calculated, then calculate by score F1 value
The weight of the linear weighted function fusion of each model.
Further, F1 fractional value indicates are as follows:
Wherein, P is accurate rate, and R is recall rate,
Further, the weight w of the linear weighted function fusion of each model is indicated are as follows:
Wherein, wiThe weight for indicating the linear weighted function fusion of i-th of model, indicates logistic as i=1
Regression model indicates catboost model as i=2, as i=3 indicates GBDT model, indicating as i=4
XGBoost model；F1iIndicate the F1 fractional value of i-th of model.
Further, carrying out Model Fusion operation includes respectively by logistic regression model, catboost mould
Type, four models of GBDT model and XGBoost model F1 fractional value respectively multiplied by the corresponding weight of model, then sum and be added
The value arrived is indicated as final prediction result are as follows:
Invention is by evaluating data and use to user basic information data, commodity essential information data, commodity
Family behavioral data carries out data prediction and feature is extracted in analysis, establishes machine learning model and carries out Model Fusion operation, from
And accurately predict the purchasing demand in user's future, the matching result of high latent user and end article are exported, is mentioned for precision marketing
Shopping experience that is simpler, quick, saving worry is provided for the target group of high quality, while also for electric business user.
Detailed description of the invention
Fig. 1 is algorithm flow chart provided in an embodiment of the present invention；
Fig. 2 is data division figure provided in an embodiment of the present invention；
Fig. 3 is Model Fusion flow chart provided in an embodiment of the present 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.It is based on
Embodiment in the present invention, it is obtained by those of ordinary skill in the art without making creative efforts every other
Embodiment shall fall within the protection scope of the present invention.
The present invention proposes a kind of user's purchase intention prediction technique based on big data, comprising:
S1, pretreatment operation is carried out to data set；
S2, training set, verifying collection and test set operation are divided according to data set；
S3, construction feature Engineering operation；
S4, feature selecting operation is carried out to the Feature Engineering of building；
S5, machine learning model is established according to the feature of selection and carries out Model Fusion operation；
S6, the model by building, whether prediction user is following buys specified commodity operation in one day.
Such as Fig. 3, the present invention carries out data prediction to training set, then extracts feature to the data after pretreatment, later
To selecting for the feature of extraction, the feature of selection is put into building logistic regression, catboost, GBDT
Among this four models of XGBoost, then the value that this four models are obtained carries out linear weighted function fusion, can be predicted.
Carrying out pretreatment operation to data set includes deleting the user for there was only purchaser record, is deleted without collection or without shopping cart
Or the user without buying behavior, delete last 10 days commodity not interacted and user.
Further, dividing training set, verifying collection and test set operation according to data set included using sliding window method, with 7 days
For a cycle, length of window is 10 days progress sliding windows, to construct multiple trained windows, dilated data set quantity；Wherein, it tests
The mode of 5 folding cross validations is taken in the building of card collection, wherein 4 parts are used as training set data, 1 part as verifying collection data, test
The building of collection chooses the data for predicting 10 days a few days ago as test data；The present embodiment provides a kind of specific embodiment, such as schemes
2, user's purchase intention when if desired predicting 20180601, then training set includes T1, T2, T3, T4, T5, and wherein T1 includes
20180423~20180502 this 10 days characteristic, T2 include 20180430~20180509 this 10 days characteristic,
T3 includes 20180507~20180516 this 10 days characteristic, and T4 includes 20180514~20180523 this 10 days feature
Data, T5 include 20180521~20180530 this 10 days characteristic, the data that test set is before 20,180,601 10, i.e.,
Data of 20180522~20180601 data as test set, T1~T5 select a data as verifying collection, other
As the data of training set, such as Fig. 2, if the data of T1 can be trained to come using the data of T5 as the data of verifying collection
Prediction 20180503 user intent, i.e., as shown in Fig. 2, using characteristic interval data prediction label section, by training set
Data input the hyper parameter for the data point reuse model for carrying out models fitting to model, and verifying being utilized to collect and for the energy to model
Power carries out entry evaluation, in the network reference services of model and then by the data input model of test set, carries out purchase intention
Prediction.
Further, construction feature Engineering operation includes basic statistical syndrome, temporal aspect group, assemblage characteristic group, industry
It is engaged in syndrome totally four groups of syndromes, foundation characteristic group mainly portrays user's basic act and commodity essential information, including user is clear
It lookes at number, user and collects number, user purchase number, user is added to buy number, goods browse number, articles storage number, commodity and add purchase number, commodity purchase
Number, user purchase conversion ratio, commodity purchasing conversion ratio are bought, temporal aspect group mainly portrays user and the time decaying of commodity becomes
Gesture, including user at the last 2nd, 3,4,5,7,10 day to the browsing number of commodity, collection number plus purchase number and the maximum for buying number
Value, minimum value, mean value, median, variance, summation, assemblage characteristic group mainly foundation characteristic combination come carry out go deep into excavate number
According to comprising information, including user-goods browse number, user-articles storage number, user-commodity add purchase number, user-commodity purchase
Buy number, user-merchandise classification browses number, user-merchandise classification collects number, user-merchandise classification adds purchase number, user-commodity class
Not Gou Mai number, user-Brand browsing number, user-Brand collection number, user-Brand add purchase number, user-quotient
Product Brand Buying number, service feature group mainly excavate user information from service layer, including commodity behavior occurs for the first time for user
Range prediction day number of days, commodity behavior range prediction day number of days, the behavior of user's first time and last row occurs for user last time
For the number of days maximum value apart from number of days, user's Continuous behavior, behavior number maximum value in one hour user's last day.
Further, feature selecting operation is carried out to the feature of building and is included in the complete rear output spy of XGBoost model training
Importance is levied, feature importance is ranked up according to this, maximum 300 features of selected characteristic importance；Pierre is used again
The similarity that inferior related coefficient calculates between 300 features filters out if two characteristic similarities have reached 95% or more
The low feature of characteristic importance in the two features；Final remaining feature is exactly the feature after feature selecting, can directly be put
The training into model.
Further, establishing machine learning model according to the feature of selection and carrying out Model Fusion operation includes that building is multiple
The biggish model of diversity factor carries out Model Fusion, it is contemplated that often performance is preferably, special to select in prediction for integrated model
The integrated model of catboost, GBDT and XGBoost totally three tree-shaped, in addition to increasing model difference degree, also selection is linear
Disaggregated model logistic regression is merged, using the feature of selection, construct logistic regression,
Catboost, GBDT and XGBoost totally four models, then the score F1 value of four models is calculated, then calculate by score F1 value
The weight w of the linear weighted function fusion of each model, finally the prediction result of four models respectively multiplied by the corresponding weight of model,
The value for being added and obtaining is summed again as final prediction result.
F1 fractional value indicates are as follows:
Wherein, P is accurate rate, and R is recall rate,
The weight w of the linear weighted function fusion of each model is indicated are as follows:
Wherein, wiThe weight for indicating the linear weighted function fusion of i-th of model, indicates logistic as i=1
Regression model indicates catboost model as i=2, as i=3 indicates GBDT model, indicating as i=4
XGBoost model；F1iIndicate the F1 fractional value of i-th of model.
Finally the prediction result of four models respectively multiplied by the corresponding weight of model, then the value conduct for being added and obtaining of summing
Final prediction result, prediction result F are indicated are as follows:
The present invention is mainly by evaluating data and user to user basic information data, commodity essential information data, commodity
Behavioral data carries out data prediction and feature is extracted in analysis, establishes machine learning model and carries out Model Fusion operation, thus
The accurately purchasing demand in prediction user's future exports the matching result of high latent user and end article, provides for precision marketing
The target group of high quality, while also shopping experience that is simpler, quick, saving worry is provided for electric business user.
It although an embodiment of the present invention has been shown and described, for the ordinary skill in the art, can be with
A variety of variations, modification, replacement can be carried out to these embodiments without departing from the principles and spirit of the present invention by understanding
And modification, the scope of the present invention is defined by the appended.
Claims (9)
1. a kind of user's purchase intention prediction technique based on big data, which comprises the following steps:
S1, pretreatment operation is carried out to data set；
S2, training set, verifying collection and test set operation are divided according to data set；
S3, construction feature Engineering operation；
S4, feature selecting operation is carried out to the Feature Engineering of building；
S5, machine learning model is established according to the feature of selection and carries out Model Fusion operation；
S6, the model by building, whether prediction user is following buys specified commodity operation in one day.
2. a kind of user's purchase intention prediction technique based on big data according to claim 1, which is characterized in that described
Carrying out pretreatment operation to data set includes deleting the user for there was only purchaser record, is deleted without collection or without shopping cart or without purchase
The user of behavior deletes last 10 days commodity not interacted and user.
3. a kind of user's purchase intention prediction technique based on big data according to claim 1, which is characterized in that according to
It includes using sliding window method that data set, which divides training set, verifying collection and test set operation, and with 7 days for a cycle, length of window is
10 days progress sliding windows, to construct multiple trained windows, dilated data set quantity；Wherein, the building for verifying collection takes 5 foldings to intersect
The mode of verifying, wherein 4 parts are used as training data, 1 part is used as verify data, and the building of test set, which is chosen, predicts 10 days a few days ago
Data are as test data.
4. a kind of user's purchase intention prediction technique based on big data according to claim 1, which is characterized in that building
Feature Engineering operation is including including basic statistical syndrome, temporal aspect group, assemblage characteristic group, service feature group totally four groups of features
Group, foundation characteristic group include that user browses number, user collects number, user adds purchase number, user to buy number, goods browse number, commodity
Collection number, commodity add purchase number, commodity purchasing number, user to buy conversion ratio, commodity purchasing conversion ratio, and temporal aspect group includes user
The the last 2nd, 3,4,5,7,10 day to goods browse number, collection number plus purchase number and buy the maximum value of number, minimum value, mean value,
Median, variance, summation, assemblage characteristic group include that user-goods browse number, user-articles storage number, user-commodity add purchase
Number, user-commodity purchasing number, user-merchandise classification browsing number, user-merchandise classification collection number, user-merchandise classification add purchase
Number, user-merchandise classification purchase number, user-Brand browsing number, user-Brand collect number, user-Brand
Add purchase number, user-Brand purchase number, service feature group includes that commodity behavior range prediction day day occurs for the first time for user
Number, user finally time generation commodity behavior range prediction day number of days, the behavior of user's first time and last behavior are apart from number of days, use
Number of days maximum value, the behavior number maximum value in one hour user's last day of family Continuous behavior.
5. a kind of user's purchase intention prediction technique based on big data according to claim 1, which is characterized in that structure
The feature built carries out feature selecting operation and is included in the complete rear output feature importance of XGBoost model training, important according to feature
Property is ranked up feature, maximum 300 features of selected characteristic importance；300 spies are calculated with Pearson correlation coefficient again
It is important to filter out feature in the two features if two characteristic similarities have reached 95% or more for similarity between sign
Spend low feature.
6. a kind of user's purchase intention prediction technique based on big data according to claim 1, which is characterized in that according to
The feature of selection establishes machine learning model including the use of the feature of selection, construct logistic regression model,
Catboost model, GBDT model and XGBoost model totally four models, then the F1 fractional value of four models is calculated, then pass through
F1 fractional value calculates the weight of the linear weighted function fusion of each model.
7. a kind of user's purchase intention prediction technique based on big data according to claim 6, which is characterized in that F1 points
Numerical value is expressed as:
Wherein, P indicates that accurate rate, R indicate recall rate.
8. a kind of user's purchase intention prediction technique based on big data according to claim 7, which is characterized in that each
The weight of the linear weighted function fusion of model indicates are as follows:
Wherein, wiThe weight for indicating the linear weighted function fusion of i-th of model, logistic regression mould is indicated as i=1
The weight of type indicates the weight of catboost model, the weight for indicating GBDT model as i=3, the table as i=4 as i=2
Show the weight of XGBoost model；F1iIndicate the F1 fractional value of i-th of model.
9. a kind of user's purchase intention prediction technique based on big data according to claim 8, which is characterized in that carry out
Model Fusion operation includes respectively by logistic regression model, catboost model, GBDT model and XGBoost
The F1 fractional value of four models of model is respectively multiplied by the corresponding weight of model, then the value for being added and obtaining of summing is as final prediction
As a result, indicating are as follows:
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910021419.8A CN109741113A (en) | 2019-01-10 | 2019-01-10 | A kind of user's purchase intention prediction technique based on big data |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910021419.8A CN109741113A (en) | 2019-01-10 | 2019-01-10 | A kind of user's purchase intention prediction technique based on big data |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109741113A true CN109741113A (en) | 2019-05-10 |
Family
ID=66364269
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910021419.8A Pending CN109741113A (en) | 2019-01-10 | 2019-01-10 | A kind of user's purchase intention prediction technique based on big data |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109741113A (en) |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110264245A (en) * | 2019-05-27 | 2019-09-20 | 浙江华坤道威数据科技有限公司 | A kind of on-line marketing auxiliary system based on network communication |
| CN110544131A (en) * | 2019-09-06 | 2019-12-06 | 创新奇智(重庆)科技有限公司 | Data-driven E-commerce user purchasing behavior prediction method |
| CN110570251A (en) * | 2019-09-11 | 2019-12-13 | 瞿秋昱 | Big data-based user purchase intention prediction system and method |
| CN110941963A (en) * | 2019-11-29 | 2020-03-31 | 福州大学 | Text attribute viewpoint abstract generation method and system based on sentence emotion attributes |
| CN111352976A (en) * | 2020-03-05 | 2020-06-30 | 广东工业大学 | Search advertisement conversion rate prediction method and device for shopping nodes |
| CN112085525A (en) * | 2020-09-04 | 2020-12-15 | 长沙理工大学 | User network purchasing behavior prediction research method based on hybrid model |
| CN112464094A (en) * | 2020-11-30 | 2021-03-09 | 泰康保险集团股份有限公司 | Information recommendation method and device, electronic equipment and storage medium |
| JP2021121922A (en) * | 2020-09-25 | 2021-08-26 | ベイジン バイドゥ ネットコム サイエンス アンド テクノロジー カンパニー リミテッド | Multi-model training method and apparatus based on feature extraction, electronic device, and medium |
| CN113505309A (en) * | 2021-05-21 | 2021-10-15 | 深圳市蘑菇财富技术有限公司 | Marketing recommendation method, device, equipment and storage medium based on distributed data |
| CN114004425A (en) * | 2021-12-29 | 2022-02-01 | 北京京东振世信息技术有限公司 | Article circulation information prediction model generation method, information generation method and device |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107766580A (en) * | 2017-11-20 | 2018-03-06 | 北京奇虎科技有限公司 | The method for pushing and device of message |
| CN107909433A (en) * | 2017-11-14 | 2018-04-13 | 重庆邮电大学 | A kind of Method of Commodity Recommendation based on big data mobile e-business |
| CN108090788A (en) * | 2017-12-22 | 2018-05-29 | 苏州大学 | Ad conversion rates predictor method based on temporal information integrated model |
| US20180276689A1 (en) * | 2017-03-24 | 2018-09-27 | Verizon Patent And Licensing Inc. | Analyzing big data to determine a data plan |
-
2019
- 2019-01-10 CN CN201910021419.8A patent/CN109741113A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20180276689A1 (en) * | 2017-03-24 | 2018-09-27 | Verizon Patent And Licensing Inc. | Analyzing big data to determine a data plan |
| CN107909433A (en) * | 2017-11-14 | 2018-04-13 | 重庆邮电大学 | A kind of Method of Commodity Recommendation based on big data mobile e-business |
| CN107766580A (en) * | 2017-11-20 | 2018-03-06 | 北京奇虎科技有限公司 | The method for pushing and device of message |
| CN108090788A (en) * | 2017-12-22 | 2018-05-29 | 苏州大学 | Ad conversion rates predictor method based on temporal information integrated model |
Cited By (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110264245A (en) * | 2019-05-27 | 2019-09-20 | 浙江华坤道威数据科技有限公司 | A kind of on-line marketing auxiliary system based on network communication |
| CN110544131A (en) * | 2019-09-06 | 2019-12-06 | 创新奇智(重庆)科技有限公司 | Data-driven E-commerce user purchasing behavior prediction method |
| CN110570251A (en) * | 2019-09-11 | 2019-12-13 | 瞿秋昱 | Big data-based user purchase intention prediction system and method |
| CN110941963A (en) * | 2019-11-29 | 2020-03-31 | 福州大学 | Text attribute viewpoint abstract generation method and system based on sentence emotion attributes |
| CN111352976A (en) * | 2020-03-05 | 2020-06-30 | 广东工业大学 | Search advertisement conversion rate prediction method and device for shopping nodes |
| CN111352976B (en) * | 2020-03-05 | 2023-05-09 | 广东工业大学 | Search advertisement conversion rate prediction method and device for shopping node |
| CN112085525A (en) * | 2020-09-04 | 2020-12-15 | 长沙理工大学 | User network purchasing behavior prediction research method based on hybrid model |
| JP7095140B2 (en) | 2020-09-25 | 2022-07-04 | ベイジン バイドゥ ネットコム サイエンス テクノロジー カンパニー リミテッド | Multi-model training methods and equipment based on feature extraction, electronic devices and media |
| JP2021121922A (en) * | 2020-09-25 | 2021-08-26 | ベイジン バイドゥ ネットコム サイエンス アンド テクノロジー カンパニー リミテッド | Multi-model training method and apparatus based on feature extraction, electronic device, and medium |
| CN112464094A (en) * | 2020-11-30 | 2021-03-09 | 泰康保险集团股份有限公司 | Information recommendation method and device, electronic equipment and storage medium |
| CN112464094B (en) * | 2020-11-30 | 2024-03-15 | 泰康保险集团股份有限公司 | Information recommendation method and device, electronic equipment and storage medium |
| CN113505309A (en) * | 2021-05-21 | 2021-10-15 | 深圳市蘑菇财富技术有限公司 | Marketing recommendation method, device, equipment and storage medium based on distributed data |
| CN113505309B (en) * | 2021-05-21 | 2024-02-09 | 深圳市蘑菇财富技术有限公司 | Marketing recommendation method, device, equipment and storage medium based on distributed data |
| CN114004425B (en) * | 2021-12-29 | 2022-06-07 | 北京京东振世信息技术有限公司 | Article circulation information prediction model generation method, information generation method and device |
| CN114004425A (en) * | 2021-12-29 | 2022-02-01 | 北京京东振世信息技术有限公司 | Article circulation information prediction model generation method, information generation method and device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109741113A (en) | A kind of user's purchase intention prediction technique based on big data | |
| CN102419779B (en) | Method and device for personalized searching of commodities sequenced based on attributes | |
| CN107909433A (en) | A kind of Method of Commodity Recommendation based on big data mobile e-business | |
| CN103700005B (en) | A kind of correlation rule based on the many minimum supports of self adaptation recommends method | |
| Dobson et al. | Heuristics for pricing and positioning a product-line using conjoint and cost data | |
| CN106327227A (en) | Information recommendation system and information recommendation method | |
| CN107316234A (en) | Personalized commercial Forecasting Methodology and device | |
| CN104281956A (en) | Dynamic recommendation method capable of adapting to user interest changes based on time information | |
| CN105528374A (en) | A commodity recommendation method in electronic commerce and a system using the same | |
| CN105740430A (en) | Personalized recommendation method with socialization information fused | |
| CN111222332A (en) | Commodity recommendation method combining attention network and user emotion | |
| CN104935963A (en) | Video recommendation method based on timing sequence data mining | |
| CN107909427A (en) | A kind of Recognition with Recurrent Neural Network method for lifting recommended models timing driving ability | |
| CN102163211A (en) | Information processing device, importance calculation method, and program | |
| CN109784979A (en) | A kind of supply chain needing forecasting method of big data driving | |
| CN104239496B (en) | A kind of method of combination fuzzy weighted values similarity measurement and cluster collaborative filtering | |
| CN111914895B (en) | Water conservancy portal information recommendation method based on multilayer attention mechanism and fusion map | |
| CN102841929A (en) | Recommending method integrating user and project rating and characteristic factors | |
| CN108573408B (en) | Popular commodity list making method for maximizing benefits | |
| CN102156747B (en) | Method and device for forecasting collaborative filtering mark by introduction of social tag | |
| CN108765042A (en) | A kind of retail shop's proposed algorithm based on user Yu retail shop's network-physical space interbehavior | |
| CN104063429A (en) | Predicting method for user behavior in e-commerce | |
| CN104239335B (en) | User-specific information acquisition methods and device | |
| CN104615721B (en) | For the method and system based on return of goods related information Recommendations | |
| CN107481093A (en) | Personalized shop Forecasting Methodology and device |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190510 |
|
| RJ01 | Rejection of invention patent application after publication |