CN104331816A - Knowledge learning and privacy protection based big-data user purchase intention predicating method - Google Patents
Knowledge learning and privacy protection based big-data user purchase intention predicating method Download PDFInfo
- Publication number
- CN104331816A CN104331816A CN201410588278.5A CN201410588278A CN104331816A CN 104331816 A CN104331816 A CN 104331816A CN 201410588278 A CN201410588278 A CN 201410588278A CN 104331816 A CN104331816 A CN 104331816A
- Authority
- CN
- China
- Prior art keywords
- lambda
- sigma
- sample
- formula
- data sample
- 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
Abstract
The invention discloses a knowledge learning and privacy protection based big-data user purchase intention predicating method which comprises following steps of: (1) performing normalization processing on a large number of historical data and a small number of current data; (2) grouping the data and establishing a training sample set; (3) counting user purchase intention probability of each group; (4) calculating group labels; (5) training the training set by using an improved support vector machine; (6) constructing a prediction function; (7) inputting to-be-predicted data into the predication function to obtain a prediction result. As the improved support vector machine is used in the method, the small number of current data set probability information and the large number of historical data set probability information are blended into a structural risk minimization learning framework, learning of knowledge in different periods is realized by virtue of constructing similar distance items among data, and accordingly, the knowledge learning and privacy protection based big-data user purchase intention predicating method which is applicable for learning problems of big samples is constructed.
Description
Technical field
The invention belongs to marketing technical field, relate to mode identification technology, is knowledge based study and large data user's purchase intention Forecasting Methodology of secret protection.
Background technology
The invention belongs to marketing technical field, relate to mode identification technology, is knowledge based study and large data user's purchase intention Forecasting Methodology of secret protection.
Consumer is the guide of the various business activities of enterprise, and the purchase intention of consumer is the basis of buying behavior, can be used for predicting the behavior of consumer.From marketing angle, after enterprise grasps the purchase intention of consumer, just can the raw-material purchase of reasonable arrangement, the structure of adjustment product, the production schedule of formulation product; After marketing personnel grasp the purchase intention of consumer, just can recommend dependent merchandise to consumer targetedly, improve sales volume; After the purchase intention of consumer is grasped in market, supermarket, just can on purpose commodities purchased, the revolution of acceleration commodity.Therefore, the purchase intention of research user is the important content of marketing research, all has important theory and realistic meaning to correctly guiding consumption of resident and instructing enterprise to formulate scientific and reasonable production, marketing strategy.
The measuring method of generally common user's purchase intention has two classes: the first kind measures the purchase intention of user, as Choice Based model, this model is provided with eight ATTRIBUTE INDEX, to each index determination weight, then user is to this eight indexs scoring, finally according to weight and score calculation user to the purchase intention of this product.Equations of The Second Kind uses infotech, as Wang Ping uses decision tree and neural net method to set up client's purchase intention disaggregated model to predict the purchase intention (" maintenance data digging technology prediction client purchase intention---method and positive research " of client, information science, in May, 2005); The people such as Wu Guohua introduce several attribute variable, build the probabilistic model (" customer purchasing behavior analysis of Influential Factors and heavily purchase the prediction of probability ", management engineering journal, in January, 2005) describing customer purchasing behavior or prediction purchase probability.But these two class methods all existing defects, first kind method is simple, but the weight of index is wayward, estimation precision is low, and the historical data used in method is not enough to the whole features reflecting current data, the prerequisite of Equations of The Second Kind method buys before and after client to obey Gamma distribution interval time, and this exists significant limitation in actual use; And under these two class methods are not all suitable for large data cases, in model use procedure, particularly disclose the purchase intention of user, do not consider the confidentiality of userspersonal information.
Summary of the invention
Technical matters to be solved by this invention is: the first, and existing user's purchase intention Forecasting Methodology is under the scene of a large amount of historical correlation data and a small amount of latest data, and the accuracy of prediction is not high; The second, existing method is not suitable for large data scene, and the time needed for training pattern is longer; 3rd, existing method can not the privacy of available protecting userspersonal information.
Technical scheme of the present invention is: large data user's purchase intention Forecasting Methodology of knowledge based study and secret protection; use the support vector machine technology improved; current a small amount of data group purchase intention probabilistic information and a large amount of history data set purchase intention probabilistic informations are incorporated in structural risk minimization learning framework; the study of knowledge is realized by structure different times data similarity distance item; user's purchase intention is predicted, comprises the following steps:
Step one: normalized is done to a large amount of historical data sample and a small amount of current data sample, obtains initial sample set (x
i, y
i)
i=1 ..., N, wherein x
ifor the proper vector of sample, y
i={+1 ,-1} is the class label of sample, and N is total number of sample;
Step 2: to historical data sample and current data sample division group, divide that often to organize data amount check roughly the same, build training sample set D={D
1..., D
n, D
n+1..., D
n+d, wherein, n group is historical data sample, and rear d group is current data sample;
Step 3: the Probability p of counting user purchase intention in each grouping
i, calculate such as formula (1):
Wherein, | S
i| be the number of samples that each grouping comprises;
Step 4: calculating group label l
i, calculate such as formula (2):
Step 5: use the support vector machine improved to train training set, the form of training pattern is such as formula (3):
Wherein,
for
for nuclear space mapping function, w
hwith
the weight vector of historical data sample and current data sample respectively, C
hand C
chistorical data sample and current data sample regularization parameter respectively, ξ
iwith
be respectively the slack variable of historical data sample, ξ
jand ξ
i *be respectively the slack variable of current data sample, λ is balance parameters is a normal number, l
ifor the group label often organized obtained in described step 4, ε
i(i=1 ..., n) with ε '
i(i=n+1 ..., n+d) and be respectively the approximation accuracy often organizing sample in historical data sample and current data sample, computing formula is as following form:
Wherein p
ifor the probability calculated in described step 3, Δ is a less normal number;
Above formula (3) can convert following quadratic programming form to:
s.t.f
Tβ=0.
Wherein β=[α
h, α
h*, α, α
*]
t,
L
h=[l
1,...,l
n],L
c=[l
n+1,...,l
n+d],
By solving formula (3) and formula (5) can obtain w, the optimum solution w of b
c *, b
c *, wherein w
c *formula (6) can be expressed as:
B
c *formula (7) can be expressed as:
Step 6: structure forecast function, the form of anticipation function is formula (8):
Wherein, w
c *and b
c *for the optimum solution that described step 5 obtains;
Step 7: be input to after user data normalization to be predicted in described anticipation function, obtain predicting the outcome of user's purchase intention, if:
Beneficial effect of the present invention shows: 1) the present invention utilizes a large amount of historical datas to obtain relevant knowledge, auxiliary only comprise a small amount of current data and carry out knowledge learning, set up the forecast model based on the support vector machine improved, inherit the advantage of support vector machine based on empirical risk minimization framework largest interval, and reach the effect that current data better predicts.2) replace sample label when carrying out historical data sample and current data sample training with group probability, therefore space complexity of the present invention is (O (N
2)), time complexity is (O (N
3)) (wherein N is the group number that historical data and current data are divided into), under being applicable to large data environment than traditional SVM, (traditional SVM space complexity is (O (M in use
2)), time complexity is (O (M
3)) (wherein M is all sample number of historical data and current data), wherein N M.3) sample label contains very important privacy information; sample label is replaced with group probability when carrying out historical data sample and current data sample training; advantage is the Privacy Safeguarding provided raw data, can not reveal in the training process the privacy information of available protecting user.
Accompanying drawing explanation
Fig. 1 is the general flow chart of large data user's purchase intention Forecasting Methodology of knowledge based of the present invention study and secret protection.
Embodiment
For making the object, technical solutions and advantages of the present invention clearly understand, below in conjunction with Concrete facts example, and with reference to accompanying drawing, the present invention is described in more detail.
The present embodiment chooses the enquiry data of certain digital product as research object, wherein historical data is the enquiry data of other model digital products of this brand before half a year, comprise 10000 samples altogether, current data is the enquiry data of the same brand new model digital product recently released, and comprises 100 samples altogether.All data comprise 40 attributes altogether, comprising: AGE, SEX, MARITAL, JOB, TRAVTIME, STATECOD, DOMESTIC, BRAND, MODEL NUMBER, AMOUNT, PRICE, RETURN etc.
Step one: normalized is done to a large amount of historical data sample and a small amount of current data sample;
In order to improve the accuracy rate of Forecasting Methodology, need data normalization in [0,1] or [-1,1] interval.The present embodiment adopts method for normalizing: according to maximal value and the minimum value of every eigenwert, all data of every eigenwert be transformed into linearly in [0,1] interval, conversion formula is as follows:
Wherein x is the eigenwert before normalization, and max (x) and min (x) represent respectively and get maximal value and minimum value to x, and x' is the eigenwert after normalization.After completing the normalization of eigenwert, by all eigenwerts composition proper vector, proper vector x is expressed as x={x
1..., x
40}
t, wherein x
1..., x
40represent the concrete numerical value after above-mentioned 40 attribute normalization respectively.
Initial training sample set comprises 10000 historical data sample and 50 current data sample samples, is expressed as (x
i, y
i)
i=1 ..., N, wherein x
ifor the proper vector of sample, y
i={+1 ,-1} is the class label of sample, N=10050, and test sample book collection is 50 remaining current data samples, is expressed as (x
i, y
i)
i=1 ..., M, M=50, the y in test set
ibe used for the accuracy rate of verification method;
Step 2: to historical data sample and current data sample according to the principle division group often organizing 10 samples, historical data sample is divided into 100 groups, and current data sample is divided into 5 groups, builds training sample set D={D
1..., D
n, D
n+1..., D
n+d, wherein, n group is historical data sample, and rear d group is current data sample, n=100, d=5;
Step 3: the purchase intention Probability p of counting user in each grouping
i, computing formula is as follows:
The purchase probability of user intention
Wherein, | S
i| be the number of samples that each grouping comprises, in the present embodiment | S
i|=10;
Step 4: calculating group label l
i, as shown in the formula calculating:
Step 5: use the support vector machine improved to train training set, the form of training pattern is such as formula (4):
Wherein,
for
for nuclear space mapping function, w
hwith
the weight vector of historical data sample and current data sample respectively, C
hand C
chistorical data sample and current data sample regularization parameter respectively, C in the present embodiment
hand C
cat grid { 2
-8, 2
-7, 2
-6, 2
-5, 2
-4, 2
-3, 2
-2, 2
-1, 2
0, 2
1, 2
2, 2
3, 2
4, 2
5, 2
6, 2
7, 2
8, 2
9, 2
10middle search optimal value, ξ
iwith
be respectively the slack variable of historical data sample, ξ
jand ξ
i *be respectively the slack variable of current data sample, λ is balance parameters, is a normal number, and in the present embodiment, balance parameters λ is interval { 2
-6, 2
-5, 2
-4, 2
-3, 2
-2, 2
-1, 2
0, 2
1, 2
2, 2
3, 2
4, 2
5, 2
6, 2
7, 2
8, 2
9, 2
10, 2
11middle search optimal value, l
ifor the group label using formula (3) to calculate, ε
iwith ε '
ibe respectively the approximation accuracy often organizing sample in historical data sample and current data sample, computing formula as shown in the formula:
Wherein p
icalculated by formula (2), Δ is a less normal number, and in the present embodiment, Δ gets 0.1;
Above formula can convert following quadratic programming form to:
s.t.f
Tβ=0
Wherein β=[α
h, α
h*, α, α
*]
t,
L
h=[l
1,...,l
n],L
c=[l
n+1,...,l
n+d],
In the present embodiment, kernel function selects gaussian kernel function K (x
i, x
j)=exp (-r||x
i-x
j||
2), wherein r is the wide parameter of core, and in the present embodiment, r is interval { 2
-15, 2
-13..., 2
3middle search optimal value, x
i, x
jbe respectively the proper vector of i-th and a jth sample, || ||
2represent Euclidean distance.
By solving formula (3) and formula (5) can obtain w, the optimum solution w of b
c *, b
c *, wherein w
c *formula (7) can be expressed as:
B
c *formula (8) can be expressed as:
Step 6: structure forecast function, the form of anticipation function is:
Wherein, w
c *, b
c *for optimum solution obtained above;
Step 7: the test set comprising 50 samples is input to anticipation function
in, if:
Accuracy rate of the present invention, working time are as shown in table 1, and using the result of the inventive method with use 10000 historical data sample, 50 current data samples, 10000 historical data sample and 50 current data samples to merge the sample set in three kinds of situations to carry out training predicting the outcome of obtaining to compare as the training set of common support vector machine (SVM) (three kinds of methods are respectively SVM
10000 historical datas, SVM
50 current datasand SVM
10050 blended datas), experiment porch is MATLAB 2009 (a).Data as can be seen from table 1, other three kinds of methods accuracy rate compared with the inventive method is lower, and its reason is SVM
10000 historical datasonly use historical data to train, have ignored the change between different times data; SVM
50 current datasonly use 50 current data samples to predict, be not enough to acquisition good forecast model because training sample is very few; SVM
10050 blended datasthen do not consider the difference between different times data, these three kinds of methods all disclose the label information of sample in training in addition, do not carry out secret protection to this important information of user's purchase intention.
Table 1: the inventive method and SVM
10000 historical datas, SVM
50 current datas, SVM
10050 blended datasdiscriminating accuracy rate (%) and compare working time (second)
Accuracy rate (%) | Working time (second) | |
The present invention | 94% | 215 |
SVM 10000 historical datas | 88% | 90681 |
SVM 50 current datas | 64% | 213 |
SVM 10050 blended datas | 90% | 91190 |
above-described example just for illustration of the present invention, and is not construed as limiting the invention.Those skilled in the art can make various other various modifications and changes not departing from essence of the present invention according to these technology enlightenment disclosed by the invention, and these modifications and changes are still in protection scope of the present invention.
Claims (1)
1. large data user's purchase intention Forecasting Methodology of knowledge based study and secret protection, its feature comprises the steps:
Step one: normalized is done to a large amount of historical data sample and a small amount of current data sample, obtains initial sample set (x
i, y
i)
i=1 ..., N, wherein x
ifor the proper vector of sample, y
i={+1 ,-1} is the class label of sample, and N is total number of sample;
Step 2: to historical data sample and current data sample division group, divide that often to organize data amount check roughly the same, build training sample set D={D
1..., D
n, D
n+1..., D
n+d, wherein, n group is historical data sample, and rear d group is current data sample;
Step 3: the Probability p of counting user purchase intention in each grouping
i, calculate such as formula (1):
Wherein, | S
i| be the number of samples that each grouping comprises;
Step 4: calculating group label l
i, calculate such as formula (2):
Step 5: train the support vector machine that training set improves, the form of training pattern is such as formula (3):
Wherein,
for
for nuclear space mapping function, w
hwith
the weight vector of historical data sample and current data sample respectively, C
hand C
chistorical data sample and current data sample regularization parameter respectively, ξ
iwith
be respectively the slack variable of historical data sample, ξ
jwith
be respectively the slack variable of current data sample, λ is balance parameters, is a normal number, l
ifor the group label that formula (2) calculates, ε
iwith ε '
ibe respectively the approximation accuracy often organizing sample in historical data sample and current data sample, computing formula is as following form:
Wherein p
icalculated by formula (1), Δ is a less normal number;
Above formula can convert following quadratic programming form to:
s.t.f
Tβ=0.
Wherein
L
h=[l
1,...,l
n],L
c=[l
n+1,...,l
n+d],
kernel function, x
i, x
jbe respectively the proper vector of i-th and a jth sample,
By solving formula (3) and formula (5) can obtain w, the optimum solution w of b
c *, b
c *, wherein w
c *formula (6) can be expressed as:
B
c *formula (7) can be expressed as:
Step 6: structure forecast function, the form of anticipation function is formula (8):
Wherein, w
c *and b
c *for the optimum solution that step 5 obtains;
Step 7: be input to after user data normalization to be predicted in described anticipation function, obtain predicting the outcome of user's purchase intention, if:
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410588278.5A CN104331816B (en) | 2014-10-28 | 2014-10-28 | Knowledge based learns big data user's purchase intention Forecasting Methodology with secret protection |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410588278.5A CN104331816B (en) | 2014-10-28 | 2014-10-28 | Knowledge based learns big data user's purchase intention Forecasting Methodology with secret protection |
Publications (2)
Publication Number | Publication Date |
---|---|
CN104331816A true CN104331816A (en) | 2015-02-04 |
CN104331816B CN104331816B (en) | 2017-10-20 |
Family
ID=52406537
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201410588278.5A Active CN104331816B (en) | 2014-10-28 | 2014-10-28 | Knowledge based learns big data user's purchase intention Forecasting Methodology with secret protection |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN104331816B (en) |
Cited By (19)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104599159A (en) * | 2015-02-06 | 2015-05-06 | 浪潮集团有限公司 | User behavior recognition method based on support vector machine principle |
CN105354725A (en) * | 2015-11-20 | 2016-02-24 | 珠海多玩信息技术有限公司 | Prediction method and system of promotion effect of application |
CN105654116A (en) * | 2015-12-25 | 2016-06-08 | 苏州晨川通信科技有限公司 | Cooperation prediction method utilizing market condition data sources |
CN108183754A (en) * | 2018-01-05 | 2018-06-19 | 湘潭大学 | A kind of GSM base stations electromagnetic radiation Forecasting Methodology |
CN108256924A (en) * | 2018-02-26 | 2018-07-06 | 上海理工大学 | A kind of product marketing forecast device |
CN108305103A (en) * | 2018-02-26 | 2018-07-20 | 上海理工大学 | A kind of product marketing forecast method of the supporting vector machine model based on parameter optimization |
CN108446771A (en) * | 2018-04-02 | 2018-08-24 | 四川长虹电器股份有限公司 | A method of preventing Sale Forecasting Model over-fitting |
CN108665329A (en) * | 2017-03-29 | 2018-10-16 | 北京京东尚科信息技术有限公司 | A kind of Method of Commodity Recommendation based on user browsing behavior |
CN109146151A (en) * | 2016-02-05 | 2019-01-04 | 第四范式(北京)技术有限公司 | There is provided or obtain the method, apparatus and forecasting system of prediction result |
CN109189825A (en) * | 2018-08-10 | 2019-01-11 | 深圳前海微众银行股份有限公司 | Lateral data cutting federation learning model building method, server and medium |
CN109325508A (en) * | 2017-07-31 | 2019-02-12 | 阿里巴巴集团控股有限公司 | The representation of knowledge, machine learning model training, prediction technique, device and electronic equipment |
CN109472623A (en) * | 2018-11-05 | 2019-03-15 | 海尔电器国际股份有限公司 | Measure of managing contract |
CN110232589A (en) * | 2019-05-16 | 2019-09-13 | 浙江华坤道威数据科技有限公司 | A kind of intention customer analysis system based on big data |
US20200043066A1 (en) * | 2018-08-02 | 2020-02-06 | T-Mobile Usa, Inc. | User-behavior-based predictive product and service provisioning |
CN111401951A (en) * | 2020-03-13 | 2020-07-10 | 重庆锐云科技有限公司 | Virtual client construction method for user information transparentization under secret storage |
CN112819495A (en) * | 2019-11-18 | 2021-05-18 | 南京财经大学 | User shopping intention prediction method based on random polynomial kernel |
TWI735932B (en) * | 2019-08-21 | 2021-08-11 | 崑山科技大學 | Real estate potential customer forecasting system and method thereof |
US20210319017A1 (en) * | 2015-05-29 | 2021-10-14 | Groupon, Inc. | Mobile search |
US11403718B1 (en) * | 2019-01-23 | 2022-08-02 | Meta Platforms, Inc. | Systems and methods for sensitive data modeling |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103309990A (en) * | 2013-06-18 | 2013-09-18 | 上海晶樵网络信息技术有限公司 | User multidimensional analysis and monitoring method based on public information of Internet user |
CN103886486A (en) * | 2014-03-21 | 2014-06-25 | 吉首大学 | Electronic commerce recommending method based on support vector machine (SVM) |
US20140244361A1 (en) * | 2013-02-25 | 2014-08-28 | Ebay Inc. | System and method of predicting purchase behaviors from social media |
US20140316853A1 (en) * | 2013-04-23 | 2014-10-23 | Philip Scott Lyren | Determine a Product from Private Information of a User |
-
2014
- 2014-10-28 CN CN201410588278.5A patent/CN104331816B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20140244361A1 (en) * | 2013-02-25 | 2014-08-28 | Ebay Inc. | System and method of predicting purchase behaviors from social media |
US20140316853A1 (en) * | 2013-04-23 | 2014-10-23 | Philip Scott Lyren | Determine a Product from Private Information of a User |
CN103309990A (en) * | 2013-06-18 | 2013-09-18 | 上海晶樵网络信息技术有限公司 | User multidimensional analysis and monitoring method based on public information of Internet user |
CN103886486A (en) * | 2014-03-21 | 2014-06-25 | 吉首大学 | Electronic commerce recommending method based on support vector machine (SVM) |
Non-Patent Citations (4)
Title |
---|
吴国华等: "顾客购买行为影响因素分析及重构概率的预测", 《管理工程学报》 * |
周毅等: "移动数据业务购买意愿的关键影响因素研究", 《管理工程学报》 * |
王萍: "运用数据挖掘技术预测客户购买倾向——方法与实证研究", 《情报科学》 * |
陈彦聪: "支持向量机在个性化推荐中的应用", 《中国优秀硕士学位论文全文数据库科技信息辑》 * |
Cited By (20)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104599159A (en) * | 2015-02-06 | 2015-05-06 | 浪潮集团有限公司 | User behavior recognition method based on support vector machine principle |
US20210319017A1 (en) * | 2015-05-29 | 2021-10-14 | Groupon, Inc. | Mobile search |
CN105354725A (en) * | 2015-11-20 | 2016-02-24 | 珠海多玩信息技术有限公司 | Prediction method and system of promotion effect of application |
CN105654116A (en) * | 2015-12-25 | 2016-06-08 | 苏州晨川通信科技有限公司 | Cooperation prediction method utilizing market condition data sources |
CN109146151A (en) * | 2016-02-05 | 2019-01-04 | 第四范式(北京)技术有限公司 | There is provided or obtain the method, apparatus and forecasting system of prediction result |
CN108665329A (en) * | 2017-03-29 | 2018-10-16 | 北京京东尚科信息技术有限公司 | A kind of Method of Commodity Recommendation based on user browsing behavior |
CN109325508A (en) * | 2017-07-31 | 2019-02-12 | 阿里巴巴集团控股有限公司 | The representation of knowledge, machine learning model training, prediction technique, device and electronic equipment |
CN108183754A (en) * | 2018-01-05 | 2018-06-19 | 湘潭大学 | A kind of GSM base stations electromagnetic radiation Forecasting Methodology |
CN108305103A (en) * | 2018-02-26 | 2018-07-20 | 上海理工大学 | A kind of product marketing forecast method of the supporting vector machine model based on parameter optimization |
CN108256924A (en) * | 2018-02-26 | 2018-07-06 | 上海理工大学 | A kind of product marketing forecast device |
CN108446771A (en) * | 2018-04-02 | 2018-08-24 | 四川长虹电器股份有限公司 | A method of preventing Sale Forecasting Model over-fitting |
US20200043066A1 (en) * | 2018-08-02 | 2020-02-06 | T-Mobile Usa, Inc. | User-behavior-based predictive product and service provisioning |
US11645688B2 (en) * | 2018-08-02 | 2023-05-09 | T-Mobile Usa, Inc. | User-behavior-based predictive product and service provisioning |
CN109189825A (en) * | 2018-08-10 | 2019-01-11 | 深圳前海微众银行股份有限公司 | Lateral data cutting federation learning model building method, server and medium |
CN109472623A (en) * | 2018-11-05 | 2019-03-15 | 海尔电器国际股份有限公司 | Measure of managing contract |
US11403718B1 (en) * | 2019-01-23 | 2022-08-02 | Meta Platforms, Inc. | Systems and methods for sensitive data modeling |
CN110232589A (en) * | 2019-05-16 | 2019-09-13 | 浙江华坤道威数据科技有限公司 | A kind of intention customer analysis system based on big data |
TWI735932B (en) * | 2019-08-21 | 2021-08-11 | 崑山科技大學 | Real estate potential customer forecasting system and method thereof |
CN112819495A (en) * | 2019-11-18 | 2021-05-18 | 南京财经大学 | User shopping intention prediction method based on random polynomial kernel |
CN111401951A (en) * | 2020-03-13 | 2020-07-10 | 重庆锐云科技有限公司 | Virtual client construction method for user information transparentization under secret storage |
Also Published As
Publication number | Publication date |
---|---|
CN104331816B (en) | 2017-10-20 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN104331816A (en) | Knowledge learning and privacy protection based big-data user purchase intention predicating method | |
Lin et al. | Empirical mode decomposition–based least squares support vector regression for foreign exchange rate forecasting | |
Wu et al. | A patent quality analysis and classification system using self-organizing maps with support vector machine | |
Niu | Walmart sales forecasting using xgboost algorithm and feature engineering | |
Bebarta et al. | Forecasting and classification of Indian stocks using different polynomial functional link artificial neural networks | |
Alem et al. | An integrated fuzzy DEA-fuzzy AHP approach: a new model for ranking decision-making units | |
Wen et al. | Daily sales forecasting for grapes by support vector machine | |
CN113111924A (en) | Electric power customer classification method and device | |
Saha et al. | Demand forecasting of a multinational retail company using deep learning frameworks | |
Zhang et al. | Multi factor stock selection model based on LSTM | |
Fan et al. | Constructing an assessment index system for strategic risk management in coal science and technology enterprises | |
Ramya et al. | Environment change prediction to adapt climate-smart agriculture using big data analytics | |
Zougagh et al. | Prediction models of demand in supply chain | |
CN101976301A (en) | Method and device for preprocessing historical data in yearly load forecasting | |
Hallman | A comparative study on Linear Regression and Neural Networks for estimating order quantities of powder blends | |
Ginevičius et al. | A quantitative evaluation of the company environment for the formation of its effective expansion strategy | |
Du et al. | Xgboost based strategic consumers classification model on e-commerce platform | |
Hai | Data mining of enterprise financial management based on AHP | |
Huang et al. | Clustering analysis on e-commerce transaction based on k-means clustering | |
Chen et al. | PLS-SVR optimized by PSO algorithm for electricity consumption forecasting | |
Hakim et al. | Implementation of Random Forest Algorithm on Palm Oil Price Data | |
Li | Sales Forecasting Model of E-commerce Activities Based on Improved Random Forest Algorithm | |
Artaye et al. | IMPLEMENTATION OF NAIVE BAYES CLASSIFIER ALGORITHM FOR DETERMINING CUSTOMERS’BUYING INTEREST ON SOFA (A CASE STUDY IN BAYAU KELUMER FURNITURE) | |
Hong | Research on the comprehensive evaluation of sports management system with analytical hierarchy process | |
Saiyin et al. | Research on Apparel Sales Forecast Based on ID3 Decision Tree Algorithm |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant | ||
TR01 | Transfer of patent right | ||
TR01 | Transfer of patent right |
Effective date of registration: 20211210 Address after: 213000 room 1505, No. 9-1, Taihu East Road, Xinbei District, Changzhou City, Jiangsu Province Patentee after: CHANGZHOU HUALONG NETWORK TECHNOLOGY CO.,LTD. Address before: Gehu Lake Road Wujin District 213164 Jiangsu city of Changzhou province No. 1 Patentee before: CHANGZHOU University |