CN112381568A - Target crowd circle selection method, target crowd circle selection model construction method and device - Google Patents
Target crowd circle selection method, target crowd circle selection model construction method and device Download PDFInfo
- Publication number
- CN112381568A CN112381568A CN202011256903.8A CN202011256903A CN112381568A CN 112381568 A CN112381568 A CN 112381568A CN 202011256903 A CN202011256903 A CN 202011256903A CN 112381568 A CN112381568 A CN 112381568A
- Authority
- CN
- China
- Prior art keywords
- target
- sample set
- release
- tested
- crowd
- 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
- 238000010276 construction Methods 0.000 title claims abstract description 21
- 238000010187 selection method Methods 0.000 title claims description 8
- 230000000694 effects Effects 0.000 claims abstract description 73
- 238000000034 method Methods 0.000 claims abstract description 48
- 238000012549 training Methods 0.000 claims description 24
- 238000004422 calculation algorithm Methods 0.000 claims description 22
- 238000013101 initial test Methods 0.000 claims description 13
- 230000006399 behavior Effects 0.000 claims description 9
- 238000012360 testing method Methods 0.000 claims description 8
- 238000000605 extraction Methods 0.000 claims description 7
- 238000007637 random forest analysis Methods 0.000 claims description 6
- 238000007477 logistic regression Methods 0.000 claims description 5
- 238000005070 sampling Methods 0.000 claims description 3
- 238000010200 validation analysis Methods 0.000 claims description 2
- 238000004891 communication Methods 0.000 description 6
- 238000012545 processing Methods 0.000 description 6
- 238000011156 evaluation Methods 0.000 description 5
- 230000006870 function Effects 0.000 description 5
- 238000012986 modification Methods 0.000 description 5
- 230000004048 modification Effects 0.000 description 5
- 238000006243 chemical reaction Methods 0.000 description 4
- 238000004140 cleaning Methods 0.000 description 4
- 230000002860 competitive effect Effects 0.000 description 4
- 230000008569 process Effects 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 230000002159 abnormal effect Effects 0.000 description 2
- 238000012795 verification Methods 0.000 description 2
- 241000283690 Bos taurus Species 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000007796 conventional method Methods 0.000 description 1
- 238000002790 cross-validation Methods 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 238000005034 decoration Methods 0.000 description 1
- 230000005611 electricity Effects 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 230000010365 information processing Effects 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
- 238000005406 washing Methods 0.000 description 1
Images
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
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0241—Advertisements
- G06Q30/0242—Determining effectiveness of advertisements
- G06Q30/0245—Surveys
-
- 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
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Strategic Management (AREA)
- Finance (AREA)
- Development Economics (AREA)
- Accounting & Taxation (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Entrepreneurship & Innovation (AREA)
- Physics & Mathematics (AREA)
- Game Theory and Decision Science (AREA)
- Economics (AREA)
- Marketing (AREA)
- General Business, Economics & Management (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Evolutionary Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a target crowd circling method, a target crowd circling model construction method and a target crowd circling model construction device, wherein the target crowd circling method comprises the following steps: acquiring target brand categories to be tested, release information, release channels and release effect indexes; acquiring prior release information of a target brand category to be tested on a target platform in a first preset time period, and selecting the prior release information as a prediction sample set through a label custom circle; the target crowd package is obtained by inputting a prediction sample set, target brand categories to be tested, putting information, putting channels and putting effect indexes into a target crowd selection model which is constructed in advance, wherein the target crowd package comprises a plurality of user identifications which are sorted according to the scores of the putting effect indexes.
Description
Technical Field
The invention relates to the technical field of computer information processing, in particular to a target crowd circling method and a target crowd circling model construction method and device.
Background
At present, the crowd application (product recommendation through advertisement, short message, push and the like) of brand merchants in the e-commerce environment is generally carried out in an industry hidden passenger mode, and the recommendation is mainly carried out by directly selecting the industry category to which the brand belongs. However, by adopting the industry hidden customer method, the returned data cannot reach the brand, and the returned data is not fine enough, so that the requirement of fine operation of the brand businessman cannot be met.
In addition, as disclosed in chinese patent with publication number CN110517070A, the patent name is a method and apparatus for circle-selecting consumer groups, which obtains a first group of people and a second group of people by searching keywords, obtains associated information according to attribute information, and then obtains consumer groups related to circle-selecting consumer groups after determining circle-selecting logic. Thus, the patent defines consumer demographics based on the relevance of attributes of the demographics set, however, attributes (occupation, hobbies, etc.) of the demographics are prone to be biased, resulting in inaccurate later-stage selection.
Therefore, a potential consumer selection method meeting brand refinement operation and having high accuracy needs to be found.
Disclosure of Invention
In order to solve the technical problems, the invention provides a target crowd circling method, a target crowd circling model construction method and a target crowd circling model construction device, which can carry out circling selection on potential purchasing crowds according to brands and have high precision of circling selection results. The technical scheme provided by the invention is as follows:
in a first aspect, a target population circle selection method is provided, the method at least comprising the following steps:
acquiring target brand categories to be tested, release information, release channels and release effect indexes;
acquiring prior release information of the target brand category to be tested on a target platform in a first preset time period, and selecting the prior release information as a prediction sample set through a label custom circle;
and inputting the prediction sample set, the target brand category to be tested, the release information, the release channel and the release effect index into a pre-constructed target crowd selection model to obtain a target crowd packet, wherein the target crowd packet comprises a plurality of user identifications which are sorted according to the release effect index scores.
In a preferred embodiment, the obtaining of the prior release information of the target brand category to be tested on the target platform within the first preset time period and the selecting as the prediction sample set through the tag custom circle includes the following sub-steps:
acquiring prior release information of the target brand category to be tested on a target platform in a first preset time period, and extracting a user identifier and a release effect in the prior release information;
selecting a user set with the putting effect within a preset threshold value as an initial test sample set;
and on the basis of the initial test sample set, combining the received label custom selection information to obtain a prediction sample set.
In a preferred embodiment, the method further comprises: the method comprises the following steps of constructing a target crowd circle selection model in advance, wherein the target crowd circle selection model comprises the following substeps:
extracting a first positive sample set and a first negative sample set on the target platform according to a target brand category to be tested, wherein the first positive sample set comprises all sample users who purchase products related to the target brand category on the target platform in a second preset time period and corresponding sample user characteristics, and the first negative sample set comprises sample users and corresponding sample user characteristics which are randomly selected from dates when the first positive sample set users do not purchase the products related to the target brand category in the second preset time period;
taking the sample user characteristics in the first positive sample set and the first negative sample set as input, taking the input effect index score as output, and respectively training n models through n preset algorithms, wherein n is more than or equal to 2;
and determining the model with the highest accuracy rate in the n models as a target crowd selection model.
In a preferred embodiment, before inputting the prediction sample set into the pre-constructed target population selection model, the method further includes real-time model verification, including:
extracting a second positive sample set and a second negative sample set on the target platform according to a target brand category to be tested, wherein the second positive sample set comprises all sample users who purchase products related to the target brand category on the target platform within a third preset time period and corresponding sample user characteristics, and the second negative sample set comprises sample users and corresponding sample user characteristics which are randomly selected from dates when the users of the second positive sample set do not purchase the products related to the target brand category within the third preset time period;
taking the sample user characteristics in the second positive sample set and the second negative sample set as input, taking the input effect index score as output, and respectively training m models through m preset algorithms;
and determining the model with the highest accuracy rate in the n + m models as a target crowd selection model, wherein m is more than or equal to 2.
In a preferred embodiment, the sample user characteristics include at least one dimension data of a user tag characteristic of the user on the target platform, a user corresponding brand descending characteristic, a user shopping on the target platform, and a user competition descending characteristic on the target platform.
In a preferred embodiment, the algorithm includes, but is not limited to, at least one of logistic regression, random forest, or xgboost.
In a second aspect, a method for constructing a target crowd circling model is provided, which comprises the following steps:
extracting a first positive sample set and a first negative sample set on the target platform according to a target brand category to be tested, wherein the first positive sample set comprises all sample users who purchase products related to the target brand category on the target platform in a second preset time period and corresponding sample user characteristics, and the first negative sample set comprises sample users and corresponding sample user characteristics which are randomly selected from dates when the first positive sample set users do not purchase the products related to the target brand category in the second preset time period;
taking the sample user characteristics in the first positive sample set and the first negative sample set as input, taking the input effect index score as output, and respectively training n models through n preset algorithms, wherein n is more than or equal to 2;
and determining the model with the highest accuracy rate in the n models as a target crowd selection model.
In a third aspect, there is provided a target crowd selection device, the device comprising:
the system comprises a first acquisition module, a second acquisition module and a third acquisition module, wherein the first acquisition module is used for acquiring target brand categories to be tested, release information, release channels and release effect indexes;
the second acquisition module is used for acquiring prior release information of the target brand category to be tested on the target platform in a first preset time period and selecting the prior release information as a prediction sample set through a label custom circle;
and the testing module is used for inputting the prediction sample set, the target brand category to be tested, the release information, the release channel and the release effect index into a pre-constructed target crowd selection model to obtain a target crowd packet, and the target crowd packet comprises a plurality of user identifications which are sorted according to the release effect index scores.
In a fourth aspect, a target population selection model building device is provided, the device at least comprising:
a first extraction module: the system comprises a target platform, a first positive sample set and a first negative sample set, wherein the target platform is used for extracting the first positive sample set and the first negative sample set according to a target brand category to be tested, the first positive sample set comprises all sample users and corresponding sample user characteristics for purchasing products related to the target brand category on the target platform in a second preset time period, and the first negative sample set comprises sample users and corresponding sample user characteristics which are randomly selected from dates on which the first positive sample set users do not purchase the products related to the target brand category in the second preset time period;
a training module: the system comprises a first positive sample set and a first negative sample set, a second positive sample set and a third negative sample set, wherein the first positive sample set and the second negative sample set are used for sampling user characteristics, the second positive sample set and the second negative sample set are used as input, the input effect index score is used as output, n models are respectively trained through n preset algorithms, and n is more than or equal to 2;
a determination module: and the model with the highest accuracy in the n models is determined as the target crowd selection model.
In a fifth aspect, there is provided a computer system comprising:
one or more processors; and
a memory associated with the one or more processors for storing program instructions that, when read and executed by the one or more processors, perform operations comprising:
acquiring target brand categories to be tested, release information, release channels and release effect indexes;
acquiring prior release information of the target brand category to be tested on a target platform in a first preset time period, and selecting the prior release information as a prediction sample set through a label custom circle;
and inputting the prediction sample set, the target brand category to be tested, the release information, the release channel and the release effect index into a pre-constructed target crowd selection model to obtain a target crowd packet, wherein the target crowd packet comprises a plurality of user identifications which are sorted according to the release effect index scores.
Compared with the prior art, the invention has the beneficial effects that:
the invention provides a target crowd circling method, a target crowd circling model construction method and a target crowd circling model construction device, wherein the target crowd circling method comprises the following steps: acquiring target brand categories to be tested, release information, release channels and release effect indexes; acquiring prior release information of a target brand category to be tested on a target platform in a first preset time period, and selecting the prior release information as a prediction sample set through a label custom circle; inputting a prediction sample set, a target brand category to be tested, putting information, a putting channel and a putting effect index into a pre-constructed target crowd selection model to obtain a target crowd package, wherein the target crowd package comprises a plurality of user identifications sorted according to the scores of the putting effect index;
in addition, when the target crowd selection model is constructed, the model is built according to specific brands, the adjustability of different brands in different industries is met, and the crowd accuracy is improved; training a corresponding number of models in parallel through multiple algorithms, and determining the model with the highest accuracy as a target crowd selection model so as to select an optimal model; further, before prediction is carried out each time, the positive sample and the negative sample can be extracted again to construct a plurality of models again, the models constructed before and after comparison are comprehensively compared, the worst rate is automatically compared, and the model with the highest accuracy is determined as a target crowd selection model until the model is solidified;
the embodiments of the present application only need to achieve any technical effect.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings needed to be used in the description of the embodiments will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts.
Fig. 1 is a flowchart of a target crowd circling method according to an embodiment of the present invention;
fig. 2 is a flowchart of a target crowd circling model construction method provided by the second embodiment of the present invention;
fig. 3 is a schematic structural diagram of a target group circle selection device according to a third embodiment of the present invention;
fig. 4 is a schematic structural diagram of a target crowd circling model construction device according to a fourth embodiment of the present invention;
fig. 5 is a computer system architecture diagram according to a fifth embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
In view of the problem that the brand cannot be accurately achieved when the current e-commerce platform circles potential consumers for product recommendation, the embodiment provides a target group circle selection method, a target group circle selection model construction method and a target group circle selection model construction device, so that targeted recommendation can be performed for brand categories, and accuracy of potential target group circle selection is improved.
The following further describes a target crowd circling method, a target crowd circling model construction device and a target crowd circling model construction system with reference to specific embodiments.
Example one
Referring to fig. 1, the present embodiment provides a target group selection method, which at least includes the following steps:
s1, obtaining target brand categories to be tested, release information, release channels and release effect indexes.
The target crowd circling method in the embodiment is performed on products of specified categories under brands, and is not performed on the crowd circling method on products of all brands of specified categories in the industry as in the conventional method, so that the target crowd circling method in the embodiment can be refined to the brands, and during specific execution, people can circle the crowds only by determining information such as the brands and the categories by a service end without the participation of brand merchants and operators in label circling, and finally, a result of circling on the products of the specified categories is obtained.
The releasing information comprises releasing magnitude and releasing period, and the releasing period is different and the required releasing magnitude is different according to the budget of the activity. The delivery channels comprise advertisement channels and short message/PUSH channels, the delivery channels are different, the user identifications are also different, the advertisement channels generally use a unique device to represent the user identifications, and the short message and PUSH channels generally use member codes to represent the user identifications. The delivery effect index means that evaluation effect needs to be obtained after the method is executed, when the delivery channel is an advertisement, the delivery effect index is click rate, and when the delivery channel is short message/PUSH, the delivery effect index is conversion rate.
Illustratively, the current items are: the brand category is Daisen-vacuum cleaner, the putting magnitude is 10w people, the putting period is 3 days, the putting channel is advertisement, and the putting effect index is click rate.
S2, acquiring prior release information of a target brand category to be tested on a target platform in a first preset time period, and selecting the prior release information as a prediction sample set through a label custom circle; the method specifically comprises the following steps:
s21, acquiring prior release information of the target brand category to be tested on the target platform in a first preset time period, and extracting the user identification and the release effect in the prior release information.
S22, selecting a user set with the putting effect within a preset threshold value as an initial test sample set;
and S23, obtaining a prediction sample set by combining the received label self-defined selection information on the basis of the initial test sample set.
The target platform may be any platform such as suning, kyoton, and the like, which is not limited in this application. The prior delivery information refers to delivery process information in any preset first preset time period before the current delivery. Therefore, on the premise of determining the brand category on a specified target platform, the release information in the first preset time period is determined and known, but new prior release information can be obtained by adjusting the first preset time period, so that a prior release user set with a better release effect can be selected as an initial test sample set according to experience or multiple adjustments when the method is executed, namely, the user set with the release effect within a preset threshold value is used as the initial test sample set. Compared with the method that a target platform or a user set of a certain product is used as a test sample set at will, the target crowd pack has a better putting effect.
After the initial test sample set is determined, the prediction sample set is obtained by combining the received label self-defined selection information on the basis of the initial test sample set. And (4) performing custom label selection on the basis of the initial test sample set according to the project requirement of the release by service personnel according to experience to finally obtain a prediction sample set.
S3, inputting the prediction sample set, the target brand category to be tested, the release information, the release channel and the release effect index into a pre-constructed target crowd selection model to obtain a target crowd package, wherein the target crowd package comprises a plurality of user identifications sorted according to the release effect index scores. Specifically, step S3 specifically includes:
s31, inputting the prediction sample set, the target brand category to be tested, the release information, the release channel and the release effect index into a pre-constructed target crowd selection model to obtain a target crowd pack, wherein the target crowd pack comprises a user identifier and a corresponding release effect index score. Wherein, when the conversion is predicted to be successful, the input effect index is divided into 100, and when the conversion is not successful, the input effect index is divided into 0.
In a preferred embodiment, the prediction sample set, the target brand category to be tested, the release information, the release channel, and the release effect index are obtained and stored in a data table in advance, and the information in the data table is input into a pre-training model during testing.
And S32, sorting the user identifications in the target crowd package according to the putting effect index scores, and preferably sorting the user identifications in the target crowd package in a descending order.
After the head groups are arranged in a descending order, the head groups according with the putting amount are selected as the group of the circle-selection people, so that the flexibility of the group circle-selection can be improved.
Illustratively, the crowd pack outputting the dyson-cleaner in this embodiment includes the top 10w appliance models in descending order of click rate.
Further, the method further includes step S0: the method comprises the following steps of constructing a target crowd circle selection model in advance, wherein the target crowd circle selection model comprises the following substeps:
s01, extracting a first positive sample set and a first negative sample set on the target platform according to the target brand category to be tested, wherein the first positive sample set comprises all sample users who purchase the products related to the target brand category on the target platform in a second preset time period and corresponding sample user characteristics, and the first negative sample set comprises sample users and corresponding sample user characteristics which are randomly selected from dates when the first positive sample set users do not purchase the products related to the target brand category in the second preset time period.
The sample user characteristics comprise at least one dimension data of user label characteristics of a user on the target platform, descending characteristics of a brand corresponding to the user, shopping behaviors of the user on the target platform in a whole station and behaviors of the user under competitive products on the target platform. Exemplarily, the following steps are carried out:
the user label features are user labels printed for users according to user behaviors, basic information and the like, and the specific feature dimensions comprise: age, gender, member level, purchasing power, credit level, loyalty, member home city level, member registration days, member value rating (conversion angle, traffic angle).
The downward characteristic of the brand corresponding to the user is the behavior of the user under the brand needing to be predicted, and the characteristic dimension is purchase dimension, shopping cart adding dimension, collection dimension, browsing dimension, evaluation dimension, complaint dimension and the like. Specifically, the purchasing dimension includes shopping interval, purchasing day, purchasing amount, purchasing frequency, customer unit price, returning frequency and the like; adding the dimension of the shopping cart comprises adding times of shopping and the like; the collection dimension comprises collection times and the like; the browsing dimensions comprise browsing times, browsing days, browsing commodity number and the like; the evaluation dimension comprises the average evaluation star grade of the commodities, the evaluation number of the commodities and the like; the complaint dimension includes the number of complaints, etc.
The shopping behavior characteristics of the whole station on the user target platform are the shopping behaviors of various categories of the user on the target platform, the characteristic dimensions comprise shopping interval, purchase day, purchase amount and purchase frequency dimensions, and the related categories comprise air conditioners, ice washing, black electricity, digital codes, computers, communication, small household appliances, kitchens and bathrooms, department of personal care, mothers and babies, makeup, imported health and freshness keeping, brewed wine, division of medical institutions, grain and oil rest food, wine, fresh food, home furnishing, home decoration, combined purchase, room-owned and the like.
The user competition product descending on the target platform is characterized in that the user acts under a competitive brand needing to predict the brand, and the characteristic dimensionality comprises the following steps: purchase dimensions, add shopping cart dimensions, collection dimensions, browse dimensions, and the like. The specific dimension content is similar to the user tag characteristics of the user on the target platform, and is not repeated.
Preferably, the first positive sample set is commensurate in magnitude with the first negative sample set to improve model confidence.
And S02, taking the sample user characteristics in the first positive sample set and the first negative sample set as input, taking the putting effect index score as output, and respectively training n models through n preset algorithms, wherein n is more than or equal to 2. The algorithm includes, but is not limited to, at least one of logistic regression, random forest, or xgboost.
Specifically, a first sample set and a second sample set are combined into a sample set, 70% of data in the sample set is randomly extracted as a training set, and the rest 30% of data is randomly extracted as a testing set, wherein the training set data is used for training the model, and the testing set is used for evaluating the model.
In a preferred embodiment, when the sample size is small, cross-validation can be used to improve the accuracy and recall of the model. Meanwhile, the random forest algorithm can also be combined with manual debugging of key parameters (the number of trees and the depth of the trees) to output an optimal model.
And S03, determining the model with the highest accuracy in the n models as the target crowd selection model.
Based on the training, the training of the target crowd selection model is completed.
In a preferred embodiment, when the model is not solidified and the testing is performed, before inputting the prediction sample set into the pre-constructed target population selection model, the method further comprises real-time model verification, including:
sa1, extracting a second positive sample set and a second negative sample set on the target platform according to a target brand category to be tested, where the second positive sample set includes all sample users who purchase a product related to the target brand category on the target platform within a third preset time period and corresponding sample user characteristics, and the second negative sample set includes sample users and corresponding sample user characteristics randomly selected from dates on which the second positive sample set users do not purchase the product related to the target brand category within the third preset time period;
sa2, inputting the sample user characteristics in the second positive sample set and the second negative sample set, outputting the putting effect index scores, and respectively training m models through m preset algorithms, wherein m is more than or equal to 2;
sa3, determining the model with the highest accuracy in the n + m models as a target crowd selection model.
Of course, after the model is solidified, the solidified model may be directly adopted without selecting the optimal model in the manner of retraining or verifying by using the second positive sample set and the second negative sample set as described in Sa1-Sa 3.
In a preferred embodiment, before acquiring the prior-release information and the sample user characteristics, performing underlying data cleaning on the acquired source data, wherein the underlying data cleaning rules comprise processing of underlying abnormal data. The cleaning of the business layer during processing comprises the following steps: rejecting refund orders, rejecting swiped orders (cattle users), rejecting public cards (one user identification corresponds to a large-magnitude order, and the service system judges that the public card is a public membership card), and rejecting abnormal browsing data (one member code or equipment corresponds to a large-magnitude pv number, and the service system judges that the crawler user is suspected). The cleaning of the data layer comprises: for discrete features of string format, converting into int and double types, and for classification features, one-hot coding is required.
The method obtains the prediction sample set on the basis of the prior delivery information on the target platform in the preset time period, so that the prediction sample set can be quickly adjusted in a mode of adjusting the preset time period, and the delivery effect of the target crowd bag is improved;
in addition, when the target crowd selection model is constructed, a corresponding number of models are trained in parallel through a plurality of algorithms, and the model with the highest accuracy is determined as the target crowd selection model, so that the optimal model is selected; furthermore, before prediction is carried out each time, the positive sample and the negative sample can be extracted again to construct a plurality of models again, the models constructed before and after comparison are comprehensively compared, the worst rate is automatically compared, and the model with the highest accuracy rate is determined as the target crowd selection model until the model is solidified.
Example two
Referring to fig. 2, the present embodiment provides a method for constructing a target crowd circling model, which includes the following steps:
s101, extracting a first positive sample set and a first negative sample set on a target platform according to a target brand category to be tested, wherein the first positive sample set comprises all sample users who purchase products related to the target brand category on the target platform within a second preset time period and corresponding sample user characteristics, and the first negative sample set comprises sample users and corresponding sample user characteristics which are randomly selected from dates when the first positive sample set users do not purchase the products related to the target brand category within the second preset time period;
s102, inputting the sample user characteristics in the first positive sample set and the first negative sample set, outputting the putting effect index scores, and respectively training n models through n preset algorithms, wherein n is more than or equal to 2;
s103, determining the model with the highest accuracy rate in the n models as a target crowd selection model.
The sample user characteristics comprise at least one dimension data of user tag characteristics of the user on the target platform, descending characteristics of a brand corresponding to the user, shopping behaviors of the user on the target platform and descending characteristics of competitive products of the user on the target platform.
The algorithm includes at least one of logistic regression, random forest, or xgboost.
For the specific implementation process and the effects of the method for constructing the target crowd circling model in this embodiment, please refer to the description in embodiment 1, which is not described herein again.
EXAMPLE III
In order to implement the target crowd circling method in the first embodiment, this embodiment provides a target crowd circling device corresponding to the first embodiment, as shown in fig. 3, the device at least includes:
the system comprises a first acquisition module, a second acquisition module and a third acquisition module, wherein the first acquisition module is used for acquiring target brand categories to be tested, release information, release channels and release effect indexes;
the second acquisition module is used for acquiring prior release information of the target brand category to be tested on the target platform in a first preset time period and selecting the prior release information as a prediction sample set through a label custom circle;
and the testing module is used for inputting the prediction sample set, the target brand category to be tested, the release information, the release channel and the release effect index into a pre-constructed target crowd selection model to obtain a target crowd packet, and the target crowd packet comprises a plurality of user identifications which are sorted according to the release effect index scores.
Further, the second obtaining module includes:
the system comprises a first acquisition unit, a second acquisition unit and a third acquisition unit, wherein the first acquisition unit is used for acquiring prior release information of a target brand category to be tested on a target platform in a first preset time period and extracting a user identifier and a release effect in the prior release information;
the second acquisition unit is used for selecting a user set with the putting effect within a preset threshold value as an initial test sample set;
and the first processing unit is used for obtaining a prediction sample set by combining the received label self-defined selection information on the basis of the initial test sample set.
The device further comprises: the model training module is used for constructing a target crowd selection model in advance, and specifically comprises the following steps:
a first extraction unit: the system comprises a target platform, a first positive sample set and a first negative sample set, wherein the target platform is used for extracting the first positive sample set and the first negative sample set according to a target brand category to be tested, the first positive sample set comprises all sample users and corresponding sample user characteristics for purchasing products related to the target brand category on the target platform in a second preset time period, and the first negative sample set comprises sample users and corresponding sample user characteristics which are randomly selected from dates on which the first positive sample set users do not purchase the products related to the target brand category in the second preset time period;
the first training unit is used for taking the sample user characteristics in the first positive sample set and the first negative sample set as input, putting the effect index score as output, and respectively training n models through n preset algorithms, wherein n is more than or equal to 2;
and the first determining unit is used for determining the model with the highest accuracy rate in the n models as the target crowd selection model.
The apparatus also includes a real-time model validation module comprising:
a second extraction unit: the system comprises a target platform, a first positive sample set and a first negative sample set, wherein the target platform is used for extracting the first positive sample set and the first negative sample set according to a target brand category to be tested, the first positive sample set comprises all sample users and corresponding sample user characteristics for purchasing products related to the target brand category on the target platform in a first preset time period, and the first negative sample set comprises sample users and corresponding sample user characteristics which are randomly selected from dates for purchasing the products related to the target brand category by the first positive sample set users in the first preset time period;
the second training unit is used for taking the sample user characteristics in the second positive sample set and the second negative sample set as input, putting the effect index score as output, and respectively training m models through m preset algorithms, wherein m is more than or equal to 2;
and the second determining unit is used for determining the model with the highest accuracy rate in the n + m models as the target crowd selection model.
The sample user characteristics comprise at least one dimension data of user tag characteristics of the user on the target platform, descending characteristics of a brand corresponding to the user, shopping behaviors of the user on the target platform and descending characteristics of competitive products of the user on the target platform. The algorithm includes, but is not limited to, at least one of logistic regression, random forest, or xgboost.
It should be noted that: the target group circle selection device provided in the above embodiment is exemplified by only the division of the above functional modules when triggering the target group circle selection service, and in practical application, the function distribution may be completed by different functional modules as needed, that is, the internal structure of the device is divided into different functional modules to complete all or part of the above described functions. In addition, the target group circle selection device provided in the above embodiment and the target group circle selection method provided in the first embodiment belong to the same concept, that is, the device is based on the method, and the specific implementation process thereof is described in the method embodiment in detail, and is not described herein again.
Example four
In order to execute the method for constructing the target crowd rounding model in the second embodiment, this embodiment provides a corresponding device for constructing the target crowd rounding model, as shown in fig. 4, the device at least includes:
a first extraction module: the system comprises a target platform, a first positive sample set and a first negative sample set, wherein the target platform is used for extracting the first positive sample set and the first negative sample set according to a target brand category to be tested, the first positive sample set comprises all sample users and corresponding sample user characteristics for purchasing products related to the target brand category on the target platform in a second preset time period, and the first negative sample set comprises sample users and corresponding sample user characteristics which are randomly selected from dates on which the first positive sample set users do not purchase the products related to the target brand category in the second preset time period;
a training module: the system comprises a first positive sample set and a first negative sample set, a second positive sample set and a third negative sample set, wherein the first positive sample set and the second negative sample set are used for sampling user characteristics, the second positive sample set and the second negative sample set are used as input, the input effect index score is used as output, n models are respectively trained through n preset algorithms, and n is more than or equal to 2;
a determination module: and the model with the highest accuracy in the n models is determined as the target crowd selection model.
It should be noted that: the target crowd circling model construction device provided in the above embodiment is exemplified by only the division of the above functional modules when triggering the target crowd circling model construction service, and in practical application, the function distribution may be completed by different functional modules as needed, that is, the internal structure of the device is divided into different functional modules to complete all or part of the above described functions. In addition, the target crowd circling model construction device provided in the above embodiment and the embodiment of the target crowd circling model construction method provided in the second embodiment belong to the same concept, that is, the device is based on the method, and the specific implementation process thereof is described in detail in the method embodiment, and is not described herein again.
EXAMPLE five
Corresponding to the above method and apparatus, the present embodiment provides a computer system, including:
one or more processors; and
a memory associated with the one or more processors for storing program instructions that, when read and executed by the one or more processors, perform operations comprising:
acquiring target brand categories to be tested, release information, release channels and release effect indexes;
acquiring prior release information of the target brand category to be tested on a target platform in a first preset time period, and selecting the prior release information as a prediction sample set through a label custom circle;
and inputting the prediction sample set, the target brand category to be tested, the release information, the release channel and the release effect index into a pre-constructed target crowd selection model to obtain a target crowd packet, wherein the target crowd packet comprises a plurality of user identifications which are sorted according to the release effect index scores.
Fig. 5 illustrates an architecture of a computer system, which may include, in particular, a processor 1510, a video display adapter 1511, a disk drive 1512, an input/output interface 1513, a network interface 1514, and a memory 1520. The processor 1510, video display adapter 1511, disk drive 1512, input/output interface 1513, network interface 1514, and memory 1520 may be communicatively coupled via a communication bus 1530.
The processor 1510 may be implemented by using a general CXU (Central processing Unit), a microprocessor, an Application Specific Integrated Circuit (ASIC), or one or more Integrated circuits, and is configured to execute a relevant program to implement the technical solution provided by the present application.
The Memory 1520 may be implemented in the form of a ROM (Read Only Memory), a RAM (Random Access Memory), a static storage device, a dynamic storage device, or the like. The memory 1520 may store an operating system 1521 for controlling the operation of the computer system 1500, a Basic Input Output System (BIOS) for controlling low-level operations of the computer system 1500. In addition, a web browser 1523, a data storage management system 1524, an icon font processing system 1525, and the like can also be stored. The icon font processing system 1525 may be an application program that implements the operations of the foregoing steps in this embodiment of the application. In summary, when the technical solution provided by the present application is implemented by software or firmware, the relevant program codes are stored in the memory 1520 and called for execution by the processor 1510.
The input/output interface 1513 is used for connecting an input/output module to realize information input and output. The i/o module may be configured as a component in a device (not shown) or may be external to the device to provide a corresponding function. The input devices may include a keyboard, a mouse, a touch screen, a microphone, various sensors, etc., and the output devices may include a display, a speaker, a vibrator, an indicator light, etc.
The network interface 1514 is used to connect a communication module (not shown) to enable the device to communicatively interact with other devices. The communication module can realize communication in a wired mode (such as USB, network cable and the like) and also can realize communication in a wireless mode (such as mobile network, WIFI, Bluetooth and the like).
The bus 1530 includes a path to transfer information between the various components of the device, such as the processor 1510, the video display adapter 1511, the disk drive 1512, the input/output interface 1513, the network interface 1514, and the memory 1520.
In addition, the computer system 1500 may also obtain information of specific extraction conditions from the virtual resource object extraction condition information database 1541 for performing condition judgment, and the like.
It should be noted that although the above devices only show the processor 1510, the video display adapter 1511, the disk drive 1512, the input/output interface 1513, the network interface 1514, the memory 1520, the bus 1530, etc., in a specific implementation, the devices may also include other components necessary for proper operation. Furthermore, it will be understood by those skilled in the art that the apparatus described above may also include only the components necessary to implement the solution of the present application, and not necessarily all of the components shown in the figures.
From the above description of the embodiments, it is clear to those skilled in the art that the present application can be implemented by software plus necessary general hardware platform. Based on such understanding, the technical solutions of the present application may be embodied in the form of a software product, which may be stored in a storage medium, such as a ROM/RAM, a magnetic disk, an optical disk, or the like, and includes several instructions for enabling a computer device (which may be a personal computer, a cloud server, or a network device) to execute the method according to the embodiments or some parts of the embodiments of the present application.
The embodiments in the present specification are described in a progressive manner, and the same and similar parts among the embodiments are referred to each other, and each embodiment focuses on the differences from the other embodiments. In particular, the system or system embodiments are substantially similar to the method embodiments and therefore are described in a relatively simple manner, and reference may be made to some of the descriptions of the method embodiments for related points. The above-described system and system embodiments are only illustrative, wherein the units described as separate parts may or may not be physically separate, and the parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of the present embodiment. One of ordinary skill in the art can understand and implement the data without inventive effort.
While preferred embodiments of the present invention have been described, additional variations and modifications in those embodiments may occur to those skilled in the art once they learn of the basic inventive concepts. Therefore, it is intended that the appended claims be interpreted as including preferred embodiments and all such alterations and modifications as fall within the scope of the embodiments of the invention.
It will be apparent to those skilled in the art that various changes and modifications may be made in the present invention without departing from the spirit and scope of the invention. Thus, if such modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include such modifications and variations.
Claims (10)
1. A target population selection method is characterized by at least comprising the following steps:
acquiring target brand categories to be tested, release information, release channels and release effect indexes;
acquiring prior release information of the target brand category to be tested on a target platform in a first preset time period, and selecting the prior release information as a prediction sample set through a label custom circle;
and inputting the prediction sample set, the target brand category to be tested, the release information, the release channel and the release effect index into a pre-constructed target crowd selection model to obtain a target crowd packet, wherein the target crowd packet comprises a plurality of user identifications which are sorted according to the release effect index scores.
2. The method according to claim 1, wherein the step of obtaining the prior release information of the target brand category to be tested on the target platform in the first preset time period and selecting the prior release information as the prediction sample set through the tag custom circle comprises the following sub-steps:
acquiring prior release information of the target brand category to be tested on a target platform in a first preset time period, and extracting a user identifier and a release effect in the prior release information;
selecting a user set with the putting effect within a preset threshold value as an initial test sample set;
and on the basis of the initial test sample set, combining the received label custom selection information to obtain a prediction sample set.
3. The method of claim 1, further comprising: the method comprises the following steps of constructing a target crowd circle selection model in advance, wherein the target crowd circle selection model comprises the following substeps:
extracting a first positive sample set and a first negative sample set on the target platform according to a target brand category to be tested, wherein the first positive sample set comprises all sample users who purchase products related to the target brand category on the target platform in a second preset time period and corresponding sample user characteristics, and the first negative sample set comprises sample users and corresponding sample user characteristics which are randomly selected from dates when the first positive sample set users do not purchase the products related to the target brand category in the second preset time period;
taking the sample user characteristics in the first positive sample set and the first negative sample set as input, taking the input effect index score as output, and respectively training n models through n preset algorithms, wherein n is more than or equal to 2;
and determining the model with the highest accuracy rate in the n models as a target crowd selection model.
4. The method of claim 3, wherein prior to entering the set of prediction samples into the pre-constructed target population circle selection model, further comprising real-time model validation, comprising:
extracting a second positive sample set and a second negative sample set on the target platform according to a target brand category to be tested, wherein the second positive sample set comprises all sample users who purchase products related to the target brand category on the target platform within a third preset time period and corresponding sample user characteristics, and the second negative sample set comprises sample users and corresponding sample user characteristics which are randomly selected from dates when the users of the second positive sample set do not purchase the products related to the target brand category within the third preset time period;
taking the sample user characteristics in the second positive sample set and the second negative sample set as input, taking the input effect index score as output, and respectively training m models through m preset algorithms;
and determining the model with the highest accuracy rate in the n + m models as a target crowd selection model, wherein m is more than or equal to 2.
5. The method of claim 3 or 4, wherein the sample user characteristics comprise at least one dimension data of a user tag characteristic of a user on the target platform, a user corresponding brand descending characteristic, a user shopping behavior on the target platform, and a user competition descending characteristic on the target platform.
6. The method of claim 5, wherein the algorithm includes, but is not limited to, at least one of logistic regression, random forest, or xgboost.
7. A construction method of a target crowd selection model is characterized by comprising the following steps:
extracting a first positive sample set and a first negative sample set on the target platform according to a target brand category to be tested, wherein the first positive sample set comprises all sample users who purchase products related to the target brand category on the target platform in a second preset time period and corresponding sample user characteristics, and the first negative sample set comprises sample users and corresponding sample user characteristics which are randomly selected from dates when the first positive sample set users do not purchase the products related to the target brand category in the second preset time period;
taking the sample user characteristics in the first positive sample set and the first negative sample set as input, taking the input effect index score as output, and respectively training n models through n preset algorithms, wherein n is more than or equal to 2;
and determining the model with the highest accuracy rate in the n models as a target crowd selection model.
8. A target crowd circling device is characterized by comprising
The system comprises a first acquisition module, a second acquisition module and a third acquisition module, wherein the first acquisition module is used for acquiring target brand categories to be tested, release information, release channels and release effect indexes;
the second acquisition module is used for acquiring prior release information of the target brand category to be tested on the target platform in a first preset time period and selecting the prior release information as a prediction sample set through a label custom circle;
and the testing module is used for inputting the prediction sample set, the target brand category to be tested, the release information, the release channel and the release effect index into a pre-constructed target crowd selection model to obtain a target crowd packet, and the target crowd packet comprises a plurality of user identifications which are sorted according to the release effect index scores.
9. The utility model provides a target crowd selects model construction device in circles which characterized in that: the apparatus at least comprises:
a first extraction module: the system comprises a target platform, a first positive sample set and a first negative sample set, wherein the target platform is used for extracting the first positive sample set and the first negative sample set according to a target brand category to be tested, the first positive sample set comprises all sample users and corresponding sample user characteristics for purchasing products related to the target brand category on the target platform in a second preset time period, and the first negative sample set comprises sample users and corresponding sample user characteristics which are randomly selected from dates on which the first positive sample set users do not purchase the products related to the target brand category in the second preset time period;
a training module: the system comprises a first positive sample set and a first negative sample set, a second positive sample set and a third negative sample set, wherein the first positive sample set and the second negative sample set are used for sampling user characteristics, the second positive sample set and the second negative sample set are used as input, the input effect index score is used as output, n models are respectively trained through n preset algorithms, and n is more than or equal to 2;
a determination module: and the model with the highest accuracy in the n models is determined as the target crowd selection model.
10. A computer system, comprising:
one or more processors; and
a memory associated with the one or more processors for storing program instructions that, when read and executed by the one or more processors, perform operations comprising:
acquiring target brand categories to be tested, release information, release channels and release effect indexes;
acquiring prior release information of the target brand category to be tested on a target platform in a first preset time period, and selecting the prior release information as a prediction sample set through a label custom circle;
and inputting the prediction sample set, the target brand category to be tested, the release information, the release channel and the release effect index into a pre-constructed target crowd selection model to obtain a target crowd packet, wherein the target crowd packet comprises a plurality of user identifications which are sorted according to the release effect index scores.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011256903.8A CN112381568A (en) | 2020-11-11 | 2020-11-11 | Target crowd circle selection method, target crowd circle selection model construction method and device |
| CA3138745A CA3138745A1 (en) | 2020-11-11 | 2021-11-12 | Method and device for circling and creating target population |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011256903.8A CN112381568A (en) | 2020-11-11 | 2020-11-11 | Target crowd circle selection method, target crowd circle selection model construction method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112381568A true CN112381568A (en) | 2021-02-19 |
Family
ID=74582947
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011256903.8A Pending CN112381568A (en) | 2020-11-11 | 2020-11-11 | Target crowd circle selection method, target crowd circle selection model construction method and device |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN112381568A (en) |
| CA (1) | CA3138745A1 (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114663149A (en) * | 2022-03-24 | 2022-06-24 | 润联软件系统(深圳)有限公司 | Product delivery method based on privacy protection and related equipment thereof |
| CN118096241A (en) * | 2024-04-24 | 2024-05-28 | 泉州好物有约品牌管理有限公司 | Brand marketing system based on big data technology |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160225014A1 (en) * | 2015-01-30 | 2016-08-04 | Wal-Mart Stores, Inc. | System and method for building a targeted audience for an online advertising campaign |
| CN108205766A (en) * | 2016-12-19 | 2018-06-26 | 阿里巴巴集团控股有限公司 | Information-pushing method, apparatus and system |
-
2020
- 2020-11-11 CN CN202011256903.8A patent/CN112381568A/en active Pending
-
2021
- 2021-11-12 CA CA3138745A patent/CA3138745A1/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160225014A1 (en) * | 2015-01-30 | 2016-08-04 | Wal-Mart Stores, Inc. | System and method for building a targeted audience for an online advertising campaign |
| CN108205766A (en) * | 2016-12-19 | 2018-06-26 | 阿里巴巴集团控股有限公司 | Information-pushing method, apparatus and system |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114663149A (en) * | 2022-03-24 | 2022-06-24 | 润联软件系统(深圳)有限公司 | Product delivery method based on privacy protection and related equipment thereof |
| CN118096241A (en) * | 2024-04-24 | 2024-05-28 | 泉州好物有约品牌管理有限公司 | Brand marketing system based on big data technology |
Also Published As
| Publication number | Publication date |
|---|---|
| CA3138745A1 (en) | 2022-05-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110222272B (en) | Potential customer mining and recommending method | |
| Kirby-Hawkins et al. | An investigation into the geography of corporate e-commerce sales in the UK grocery market | |
| CN108876526B (en) | Commodity recommendation method and device and computer-readable storage medium | |
| Ryans | Estimating consumer preferences for a new durable brand in an established product class | |
| US11403690B2 (en) | Determining brand affinity of users | |
| CN104239338A (en) | Information recommendation method and information recommendation device | |
| JP2013239160A (en) | Information providing system, information providing method and information providing program | |
| KR101646312B1 (en) | Personal Action-Based Interest and Preference Analysis Method and System | |
| CN110019699A (en) | Pass through the classification of grammer slot between domain | |
| Suh | The role of consideration sets in brand choice: The moderating role of product characteristics | |
| CN110750697B (en) | Merchant classification method, device, equipment and storage medium | |
| Hemalatha | Market basket analysis–a data mining application in Indian retailing | |
| CN109711917A (en) | Information-pushing method and device | |
| CN112381568A (en) | Target crowd circle selection method, target crowd circle selection model construction method and device | |
| CN110807691B (en) | Cross-commodity-class commodity recommendation method and device | |
| CN113781149A (en) | Information recommendation method and device, computer-readable storage medium and electronic equipment | |
| CN115564531A (en) | Commodity combination recommendation method and device, equipment, medium and product thereof | |
| Alam | An investigation on the use of digital marketing towards the customer satisfaction and brand loyalty of resturants in Saudi Arabia | |
| Ismail | Understanding the factors that affect the adoption of innovative high-technology brands: The case of apple iPhone in Malaysia | |
| JP2020160710A (en) | Data processing device and data processing method | |
| CN110968670B (en) | Method, device, equipment and storage medium for acquiring attributes of popular commodities | |
| JP5603678B2 (en) | Demand forecasting apparatus and demand forecasting method | |
| CN110781399A (en) | Cross-platform information pushing method and device | |
| CN112015970A (en) | Product recommendation method, related equipment and computer storage medium | |
| CN115797014A (en) | Home decoration industry online and offline commodity recommendation method and device and storage medium |
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: 20210219 |
|
| RJ01 | Rejection of invention patent application after publication |