CN112613953A - Commodity selection method, system and computer readable storage medium - Google Patents
Commodity selection method, system and computer readable storage medium Download PDFInfo
- Publication number
- CN112613953A CN112613953A CN202011588745.6A CN202011588745A CN112613953A CN 112613953 A CN112613953 A CN 112613953A CN 202011588745 A CN202011588745 A CN 202011588745A CN 112613953 A CN112613953 A CN 112613953A
- Authority
- CN
- China
- Prior art keywords
- commodity
- data
- feature
- dimensional
- model
- 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
- 238000010187 selection method Methods 0.000 title claims abstract description 18
- 238000003860 storage Methods 0.000 title claims abstract description 11
- 238000012549 training Methods 0.000 claims abstract description 35
- 238000004422 calculation algorithm Methods 0.000 claims description 24
- 238000000034 method Methods 0.000 claims description 24
- 238000010801 machine learning Methods 0.000 claims description 18
- 238000007405 data analysis Methods 0.000 claims description 8
- 238000004590 computer program Methods 0.000 claims description 6
- 238000012545 processing Methods 0.000 claims description 6
- 238000007418 data mining Methods 0.000 claims description 5
- 238000004140 cleaning Methods 0.000 claims description 4
- 230000000694 effects Effects 0.000 description 7
- 238000011156 evaluation Methods 0.000 description 7
- 238000012360 testing method Methods 0.000 description 5
- 238000004519 manufacturing process Methods 0.000 description 4
- 238000012706 support-vector machine Methods 0.000 description 4
- 238000011478 gradient descent method Methods 0.000 description 3
- 238000012986 modification Methods 0.000 description 3
- 230000004048 modification Effects 0.000 description 3
- 238000013473 artificial intelligence Methods 0.000 description 2
- 238000005034 decoration Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000007477 logistic regression Methods 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 238000007637 random forest analysis Methods 0.000 description 2
- 230000009467 reduction Effects 0.000 description 2
- RYGMFSIKBFXOCR-UHFFFAOYSA-N Copper Chemical compound [Cu] RYGMFSIKBFXOCR-UHFFFAOYSA-N 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000013145 classification model Methods 0.000 description 1
- 229910052802 copper Inorganic materials 0.000 description 1
- 239000010949 copper Substances 0.000 description 1
- 230000009193 crawling Effects 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000007429 general method Methods 0.000 description 1
- 238000000265 homogenisation Methods 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 238000005065 mining Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000012552 review Methods 0.000 description 1
- 238000010845 search algorithm Methods 0.000 description 1
- 239000013049 sediment Substances 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
- 230000002194 synthesizing effect Effects 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
- G06Q30/06—Buying, selling or leasing transactions
- G06Q30/0601—Electronic shopping [e-shopping]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/02—Marketing; Price estimation or determination; Fundraising
- G06Q30/0201—Market modelling; Market analysis; Collecting market data
- G06Q30/0202—Market predictions or forecasting for commercial activities
-
- 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
- G06Q30/06—Buying, selling or leasing transactions
- G06Q30/0601—Electronic shopping [e-shopping]
- G06Q30/0623—Item investigation
- G06Q30/0625—Directed, with specific intent or strategy
-
- 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
- G06Q30/06—Buying, selling or leasing transactions
- G06Q30/0601—Electronic shopping [e-shopping]
- G06Q30/0633—Lists, e.g. purchase orders, compilation or processing
- G06Q30/0635—Processing of requisition or of purchase orders
Landscapes
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Accounting & Taxation (AREA)
- Finance (AREA)
- Strategic Management (AREA)
- Theoretical Computer Science (AREA)
- Development Economics (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Marketing (AREA)
- General Business, Economics & Management (AREA)
- Economics (AREA)
- Software Systems (AREA)
- Entrepreneurship & Innovation (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- Medical Informatics (AREA)
- Evolutionary Computation (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Game Theory and Decision Science (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a commodity selection method, a commodity selection system and a computer readable storage medium, wherein the commodity selection method comprises the following steps: determining characteristic attribute data influencing commodity sales according to pre-obtained commodity characteristic historical data and user group preference historical data which are classified according to commodity categories; dividing the feature attribute data into a plurality of dimensional features; training and modeling feature attribute data of the plurality of dimensional features according to a preset score rule of each dimensional feature to obtain a commodity score model of the commodity category; extracting a plurality of dimensional feature data of the target commodity in the commodity category according to the plurality of dimensional features; and determining the recommendation score of the target commodity on a target platform according to the plurality of dimensional feature data and the commodity scoring model. The invention effectively solves the problems that the prior commodity selection mainly depends on human experience and subjective scoring, which causes the lack of data support of selection decision, human intervention and larger subjective influence.
Description
Technical Field
The invention relates to the field of artificial intelligence, in particular to a commodity selection method, a commodity selection system and a computer readable storage medium.
Background
With the advent of the big data era, intelligent prediction becomes a great development trend in the field of science and technology. How to predict the future possibility and development by combining the methods of machine learning and deep learning according to the existing data becomes one of the problems that many enterprises need to solve urgently.
The commodity selection is one of the key points in the E-commerce field and the E-purchase field, good sales can be achieved only when the commodity is selected, more users are harvested, and a good image is established for an enterprise.
Disclosure of Invention
The embodiment of the invention provides a commodity selection method, a commodity selection system and a computer readable storage medium, which are used for solving the problems of lack of support of selection decision data, human intervention and large subjective influence caused by the fact that the existing commodity selection mainly depends on human experience and subjective scoring.
In a first aspect, the present invention provides a commodity selection method, including:
determining characteristic attribute data influencing commodity sales according to pre-obtained commodity characteristic historical data and user group preference historical data which are classified according to commodity categories;
dividing the feature attribute data into a plurality of dimensional features;
training and modeling feature attribute data of the plurality of dimensional features according to a preset score rule of each dimensional feature to obtain a commodity score model of the commodity category;
extracting a plurality of dimensional feature data of the target commodity in the commodity category according to the plurality of dimensional features;
and determining the recommendation score of the target commodity on a target platform according to the plurality of dimensional feature data and the commodity scoring model.
Optionally, the training and modeling the feature attribute data of the multiple dimensional features according to the preset score rule of each dimensional feature to obtain the commodity score model of the commodity category includes:
according to the scoring rule of each dimension characteristic, respectively training and modeling the characteristic attribute data of the dimension characteristics by adopting a plurality of machine learning algorithms to obtain a plurality of scoring models to be evaluated;
and evaluating the scoring model with the highest accuracy and AUC value from the plurality of scoring models to be evaluated as the commodity scoring model.
Optionally, the training and modeling, according to the score rule of each dimension feature, the feature attribute data of the plurality of dimension features by using a plurality of machine learning algorithms to obtain a plurality of score models to be evaluated includes:
and respectively training and modeling the feature attribute data of the plurality of dimensional features by adopting a plurality of machine learning algorithms according to the scoring rule of each dimensional feature and the weight of each dimensional feature to obtain a plurality of models to be assessed.
Optionally, the commodity selection method further includes:
carrying out back propagation on the commodity scoring model by using a gradient descent mode, and updating the weight of each dimension characteristic; the counter-propagating comprises:
calling the relationship between the predicted value of the input layer and the real value of the output layer of the commodity scoring model through iterative processing;
the weight of each dimensional feature is updated with minimized error from the reverse direction of the output, hidden and input layers.
Optionally, the plurality of dimensional features includes a brand attribute, an e-commerce dimension attribute, a user group preference attribute, a popularity attribute, and a profitability attribute.
Optionally, the e-commerce dimension attributes include popular and unpopular;
before training and modeling the feature attribute data of the multiple dimension features according to the preset score rule of each dimension feature to obtain the commodity score model of the commodity category, the method comprises the following steps:
setting the predicted labels as popular or unpopular, and respectively modeling and training the commodity characteristic historical data by adopting the plurality of machine learning algorithms to obtain a plurality of commodity popularity models to be evaluated;
and evaluating a commodity popularity model with the highest accuracy and AUC value from the plurality of commodity popularity models to be evaluated as a commodity popularity prediction model.
Optionally, the commodity selection method further includes:
acquiring commodity characteristic historical data from commodity historical information data acquired from an e-commerce website and/or an e-shopping website in advance according to commodity categories through big data mining and analysis;
and acquiring user group preference history data from a pre-acquired commodity sales order of the target platform according to the commodity category.
Optionally, the training and modeling the feature attribute data of the multiple dimensional features according to the preset score rule of each dimensional feature to obtain the commodity score model of the commodity category includes:
cleaning the characteristic attribute data, and performing thermal coding treatment on the cleaned characteristic attribute data;
splitting each feature attribute data subjected to thermal coding into a plurality of attributes to obtain thermal coding data of a plurality of dimensional features;
normalizing the thermally encoded data for the plurality of dimensional features;
and training and modeling the normalized thermal coding data of the multiple dimension characteristics according to a preset score rule of each dimension characteristic to obtain a commodity score model of the commodity category.
In a second aspect, the present invention provides a commodity selection system, including:
a memory, a processor, and a computer program stored on the memory and executable on the processor;
the computer program when being executed by the processor realizes the steps of the commodity selection method as defined in any one of the above
In a third aspect, the invention provides a computer readable storage medium having stored thereon an item selection program, which when executed by a processor, implements the steps of the item selection method as defined in any one of the above.
By applying the technical scheme of the invention, the problems of lack of support of selection decision data, human intervention and large subjective influence caused by the fact that the existing commodity selection mainly depends on human experience and subjective scoring can be effectively solved.
Drawings
Fig. 1 is a flowchart of a commodity selection method according to an embodiment of the present invention.
FIG. 2 is a schematic diagram of model evaluation according to an embodiment of the invention.
FIG. 3 is a graph of model effects according to an embodiment of the invention.
FIG. 4 is a graph of model evaluation accuracy and AUC values according to an embodiment of the present invention.
Detailed Description
The invention discloses a commodity selection method and system based on big data and a computer readable storage medium, and a person skilled in the art can appropriately improve process parameters by referring to the contents in the text for realization. It is expressly intended that all such alterations and modifications which are obvious to those skilled in the art are deemed to be incorporated herein by reference, and that the techniques of the invention may be practiced and applied by those skilled in the art without departing from the spirit, scope and range of equivalents of the invention.
In the present invention, unless otherwise specified, scientific and technical terms used herein have the meanings that are commonly understood by those skilled in the art.
In order to make those skilled in the art better understand the technical solution of the present invention, the following detailed description of the present invention is provided with reference to specific embodiments.
Example one
An embodiment of the present invention provides a method for selecting a commodity, as shown in fig. 1, including:
s101, determining characteristic attribute data influencing commodity sales according to commodity characteristic historical data and user group preference historical data which are obtained in advance and classified according to commodity categories;
s102, dividing the feature attribute data into a plurality of dimensional features;
s103, training and modeling feature attribute data of the plurality of dimensional features according to a preset score rule of each dimensional feature to obtain a commodity score model of the commodity category; for example, through a traditional machine learning algorithm and a data analysis method, the characteristics of sales influencing the commodity classification are divided into a plurality of dimensions, a score rule is formulated for each dimension, and a commodity scoring model is established;
s104, extracting a plurality of dimensional feature data of the target commodity in the commodity category according to the plurality of dimensional features;
and S105, determining the recommendation score of the target commodity on a target platform according to the plurality of dimensional feature data and the commodity grading model.
According to the embodiment of the invention, the characteristic attribute data influencing commodity sales are determined according to the commodity characteristic historical data and the user group preference historical data which are obtained in advance and classified according to commodity categories; dividing the feature attribute data into a plurality of dimensional features; training and modeling feature attribute data of the plurality of dimensional features according to a preset score rule of each dimensional feature to obtain a commodity score model of the commodity category; extracting a plurality of dimensional feature data of the target commodity in the commodity category according to the plurality of dimensional features; and determining the recommendation score of the target commodity on a target platform according to the plurality of dimensional characteristic data and the commodity scoring model, so that the problems of lack of support of commodity decision data, human intervention and large subjective influence caused by the fact that the existing commodity selection mainly depends on human experience and subjective scoring are solved.
In some embodiments, the method of merchandise selection further comprises:
acquiring commodity characteristic historical data from commodity historical information data acquired from an e-commerce website and/or an e-shopping website in advance according to commodity categories through big data mining and analysis;
and acquiring user group preference history data from a pre-acquired commodity sales order of the target platform according to the commodity category. The e-commerce website and the e-shopping website comprise a target platform and a shopping platform except the target platform.
For example, embodiments of the present invention analyze the categories of products, each product category (or simply category) being specific to a particular type of product, such as a brand of wine in the wine field.
Obtaining commodity information data of the e-commerce website and the e-purchase website according to classification, wherein the commodity information data comprises commodity price, commodity characteristics, commodity sales volume, commodity evaluation, brand public praise and popularity information, and performing data noise reduction to obtain available commodity characteristic historical data;
the commodity sales orders of N years (N is more than 3) of the existing sediment of the target platform can be obtained according to the classification, the sales volume of each classification of the commodity, the brand sales volume under the classification and the hot sales price interval of the classification are obtained through data analysis and collection, and the preference conditions of the user group as a whole are collected to obtain the preference historical data of the user group.
That is, the step of acquiring the commodity characteristic history data from the commodity history information data acquired from the e-commerce website and/or the e-purchase website acquired in advance according to the commodity category includes:
acquiring commodity information data of a commodity website, wherein the commodity information data comprises commodity price, characteristic attributes (such as brand, specification, quantity and the like) of the commodity, commodity sales volume, commodity pictures, commodity good evaluation number, commodity evaluation number and the like;
acquiring brand data including brand popularity, brand public praise, brand value, brand ranking and the like;
and combining the commodity information data and the brand data to form usable commodity characteristic historical data.
The method for acquiring user group preference history data from a pre-acquired commodity sales order of a target platform according to commodity categories comprises the following steps:
acquiring classification information, brand information and price information of the existing commodities;
counting the sales volume and brand price distribution condition of the commodity by taking the small classification as a unit;
counting classification preference, brand preference under classification and price preference condition under classification of a user;
and recording the statistical result into a database and updating at regular time.
In some embodiments, the training and modeling the feature attribute data of multiple dimensional features according to a preset score rule of each dimensional feature to obtain a product scoring model of the product category includes:
cleaning the characteristic attribute data, and performing thermal coding treatment on the cleaned characteristic attribute data;
splitting each feature attribute data subjected to thermal coding into a plurality of attributes to obtain thermal coding data of a plurality of dimensional features;
normalizing the thermally encoded data for the plurality of dimensional features;
and training and modeling the normalized thermal coding data of the multiple dimension characteristics according to a preset score rule of each dimension characteristic to obtain a commodity score model of the commodity category.
For example, the pulled e-commerce data (feature attribute data) is cleaned, and the policy is as follows: if the null value exceeds 60%, deleting the specific attribute, otherwise, giving a value of-1;
performing thermal coding treatment on all classification data (such as the acidity of wine, which is an attribute of high acidity, medium acidity and low acidity), wherein the effect is that the acidity is divided into a plurality of attributes, namely acidity _ high acidity, acidity _ medium acidity and acidity _ low acidity, if the acidity of a piece of data is high acidity, the attribute is treated into 0 and 1, 1 represents high acidity, and 0 represents non-high acidity;
normalizing the feature data, reducing each row of attributes to be between 0 and 1, selecting features by adopting a single variable method, selecting most of the features to be 85 percent, and performing later training;
the data set is divided into a training set (80% of data) and a testing set (20% of data), and a plurality of models are built through various machine learning algorithms to fit the characteristic attribute data of the training set with the historical comment number or the historical sales volume.
In some embodiments, the training and modeling the feature attribute data of multiple dimensional features according to a preset score rule of each dimensional feature to obtain a product scoring model of the product category includes:
according to the scoring rule of each dimension characteristic, respectively training and modeling the characteristic attribute data of the dimension characteristics by adopting a plurality of machine learning algorithms to obtain a plurality of scoring models to be evaluated;
and evaluating the scoring model with the highest accuracy and AUC value from the plurality of scoring models to be evaluated as the commodity scoring model.
Optionally, the training and modeling, according to the score rule of each dimension feature, the feature attribute data of the plurality of dimension features by using a plurality of machine learning algorithms to obtain a plurality of score models to be evaluated includes:
and respectively training and modeling the feature attribute data of the plurality of dimensional features by adopting a plurality of machine learning algorithms according to the scoring rule of each dimensional feature and the weight of each dimensional feature to obtain a plurality of models to be assessed.
The plurality of dimensional features includes a brand attribute, an e-commerce dimension attribute, a user group preference attribute, a popularity attribute, and a profitability attribute. The weight of each dimension feature may be set as follows:
the brand attribute accounts for 25%, the e-commerce dimension attribute accounts for 10%, the user group preference attribute accounts for 25%, the popularity attribute accounts for 25%, and the profit margin attribute accounts for 15%.
The brand attributes include: brand awareness, brand index, brand attention, brand public praise, brand popularity; the brand popularity, the brand attention and the data on the brand public praise reference brand network are obtained, the brand popularity is the overall score of the brand, and the score is higher; the public praise mainly comprises three types of general, returning and not-good, wherein the score of returning is highest, then general and then not-good; the brand attention and brand popularity are higher according to the larger actual data; because data such as the grade of the brand and the like are changed along with time, the commodity grading system disclosed by the invention can be used for grading the data acquired in real time, so that the grading is more accurate;
the brand index refers to the Baidu index of the brand name, the higher the Baidu index is, the higher the score is, and the score is 0 without the Baidu index;
the supplier score is the sales volume of the supplier of the commodity in the company in the last year, and the better the sales volume is, the higher the score is; the new supplier judges whether the new supplier has scores such as risk, registered fund, personnel scale and the like according to the information of the supplier;
the popularity attributes are scored according to the popularity of the goods, including attributes of origin, mouthfeel, type, acidity, sweetness, and the like. Wherein the content of the first and second substances,
the property of the place of origin: grading according to the production place arrangement of hot selling of the wine on the E-business platform, wherein the most popular grading is higher; the production place attributes such as taste, type, acidity and sweetness are graded according to the popularity of the commodities;
the profit margin attribute: the higher the profit margin, the higher the score;
the user group preference attributes include member classification preferences, member brand preferences, and member price preferences.
Member classification preference: grading according to the sales volume ratio of wine classification in the company, wherein the higher the ratio is, the higher the score is;
member brand preference: grading according to the sales volume proportion of each brand under the classification of the wine, wherein the better the sales volume is, the higher the score is;
member price preference: grading according to the price interval of the hot sales under the wine classification;
e-commerce dimension attribute: and (4) scoring according to the result obtained by calling the prediction model for the commodity, wherein the scoring is divided into two grades, the first grade is 10 percent popular, and the second grade is unpopular and not scoring.
Before training and modeling the feature attribute data of the multiple dimension features according to the preset score rule of each dimension feature to obtain the commodity score model of the commodity category, the method comprises the following steps:
setting the predicted labels as popular or unpopular, and respectively modeling and training the commodity characteristic historical data by adopting the plurality of machine learning algorithms to obtain a plurality of commodity sales prediction models to be evaluated;
and evaluating the commodity sales forecasting model with the highest accuracy and AUC value from the plurality of to-be-evaluated commodity sales forecasting models to serve as a commodity popularity forecasting model. The e-commerce data acquired by the S1 can be analyzed, trained and modeled through a traditional machine learning algorithm KNN, logistic regression, SVM, random forest and xgboost, the predicted label is popular or unpopular, and the predicted result is used as an option dimension; the rating scale of whether popular is the number of historical reviews or whether the historical sales is greater than the median of the attribute.
After the commodities are classified, a commodity grading model is constructed according to the sales condition of an e-commerce website and the user group characteristics of an enterprise and the preference condition of the whole user group on the basis of the characteristic attribute of each classification, and the commodities are graded; firstly, crawling a large amount of commodity data information on the Internet, performing data noise reduction processing and analysis, and deeply mining the relationship between the characteristic attribute and sales volume of a commodity; then combining with sales data of enterprises, and primarily establishing a commodity scoring model; establishing a commodity sales forecasting model based on crawled commodity data information through multiple algorithms such as logistic regression, random forests, SVM (support vector machine), KNN (K nearest neighbor), xgboost and the like, fusing multiple algorithms to obtain a more accurate model, and forecasting whether the commodity is popular in the E-commerce website; and finally, combining the model prediction result of the E-commerce with the actual sales data of the enterprise, establishing a commodity grading model suitable for the enterprise, grading and recommending commodities, and providing a selection idea and a decision reference for the online retail enterprise.
Taking wine classification as an example, the method comprises the steps of obtaining information such as commodity price, taste, production place, type, acidity and sweetness, the number of bottles contained in each commodity, capacity and brand name and the like from an e-commerce website, obtaining data such as popularity, public praise and score of the brand according to the brand name, cleaning and normalizing the obtained data, dividing the data set into 80% of training sets and 20% of testing sets by using a traditional machine learning algorithm, training the data, and finally obtaining a commodity popularity prediction model.
The commodity popularity prediction model is used as a commodity rating model, the commodity sales prediction model with the highest accuracy and the highest AUC value is evaluated in the commodity sales prediction models, and the model can be persisted by adopting the accuracy and the AUC area of the confusion matrix, so that the persisted commodity rating model and the commodity popularity prediction model can be called for prediction. In the test and evaluation of the two models, the judgment is mainly made by the following formula, namely, the judgment of the correct result in the test set according to the classification model accounts for the data volume of the test set, as shown in fig. 2.
The evaluation is carried out according to the AUC area, both the ROC curve and the AUC area are tools for measuring the accuracy of the classified models, the ROC is a line, if the ROC curve is selected to judge the accuracy of the models, the closer to the ROC curve at the upper left corner, the higher the accuracy of the models is, and the more ideal the models are; AUC is the area under the ROC line, if we choose to judge the accuracy of the model by the area of AUC, the greater the AUC area value of the model, the higher the accuracy of the model, and the more ideal the model, as shown in FIG. 3, which is a ROC/RUC graph of this small classification of wine.
Criterion for judging the quality of the classifier (prediction model) from AUC:
AUC is [0.85,0.95], the effect is good;
AUC ═ 0.7,0.85], general effect;
AUC ═ 0.5,0.7], less effective;
AUC is 0.5, the follower guesses the same (for example: missing copper plate), and the model has no prediction value;
AUC <0.5, worse than random guess; but is better than random guessing as long as it always works against prediction;
as shown in fig. 4, five algorithms are used to perform modeling and prediction respectively, the SVM algorithm with the lowest accuracy is excluded by synthesizing the algorithms, and the other 4 models are fused to obtain a more accurate model, wherein the accuracy of the model is 0.84, and the AUC is 0.93, i.e., the effect is good.
In some embodiments, the method of merchandise selection further comprises:
carrying out back propagation on the commodity scoring model by using a gradient descent mode, and updating the weight of each dimension characteristic; the counter-propagating comprises:
calling the relationship between the predicted value of the input layer and the real value of the output layer of the commodity scoring model through iterative processing;
the weight of each dimensional feature is updated with minimized error from the reverse direction of the output, hidden and input layers.
Namely, a gradient descent method is used for carrying out back propagation on the commodity scoring model, and the weight parameters of the model are updated;
the gradient descent method is an optimization algorithm, which is commonly used in machine learning and artificial intelligence to recursively approximate a minimum deviation model, and the iterative formula is that, the gradient descent method represents a negative direction of a gradient and represents a search step length in the direction of the gradient. The gradient direction can be obtained by deriving the function, the step length is relatively troublesome to determine, if the step length is too large, the step length may diverge, and if the step length is too small, the convergence speed is too slow. The general method for determining the step length is determined by a linear search algorithm, namely, the coordinate of the next point is regarded as a function of ak +1, and then ak +1 meeting the minimum value of f (ak +1) is solved, because in general, if the gradient vector is 0, an extreme point is reached, the amplitude of the gradient is also 0, and when the gradient descent algorithm is adopted for optimal solution, the termination condition of the iteration of the algorithm is that the amplitude of the gradient vector is close to 0, and a very small constant threshold value can be set;
the back propagation method is mainly a method for updating weight and bias in a reverse direction by using errors, and is mainly characterized in that:
processing the instances in the training set by iteration;
comparing the relationship between the predicted value of the input layer and the real value of the output layer after calling the model;
in the reverse direction (from the output layer)
Hidden layer
Input layer) to update the weights for each dimension with minimized error.
The embodiment of the invention collects the preference and interest of users by analyzing and sorting the existing historical data and establishes the preference historical data of a user group;
the method comprises the steps of collecting commodity characteristic information through big data mining and analysis, establishing a model based on characteristics of commodities and actual sales conditions, analyzing main factors influencing commodity sales, and establishing a commodity grading model. For example, selecting small classes as wine for data mining and analysis, evaluating the production area, the capacity, the brand and multiple dimensions of the wine, determining the weight and the score, and establishing a grading model. When a new commodity is introduced, the commodity is scored according to the commodity scoring model, and decision making is assisted. In addition, historical data is collected and collated, and data analysis and processing are carried out on commodities which are tried to be sold and have poor final results, so that data support is provided during decision making. And the grading weight of the existing commodities is adjusted according to historical data, so that the homogenization of the commodities is reduced. By evaluating a plurality of small categories, a better effect is achieved. The system can assist decision-making to a certain extent and provide technical support and data support for commodity selection.
Example two
The embodiment of the invention provides a commodity selection system, which comprises:
a memory, a processor, and a computer program stored on the memory and executable on the processor;
the computer program, when executed by the processor, implements the steps of the method of merchandise selection according to any one of the embodiments.
EXAMPLE III
An embodiment of the present invention provides a computer-readable storage medium, where an item selection program is stored on the computer-readable storage medium, and when the item selection program is executed by a processor, the steps of the item selection method according to any one of embodiments are implemented.
In the specific implementation process of the second embodiment and the third embodiment, reference may be made to the first embodiment, and corresponding technical effects are achieved.
The foregoing is only a preferred embodiment of the present invention, and it should be noted that, for those skilled in the art, various modifications and decorations can be made without departing from the principle of the present invention, and these modifications and decorations should also be regarded as the protection scope of the present invention.
Claims (10)
1. A commodity selection method is characterized by comprising the following steps:
determining characteristic attribute data influencing commodity sales according to pre-obtained commodity characteristic historical data and user group preference historical data which are classified according to commodity categories;
dividing the feature attribute data into a plurality of dimensional features;
training and modeling feature attribute data of the plurality of dimensional features according to a preset score rule of each dimensional feature to obtain a commodity score model of the commodity category;
extracting a plurality of dimensional feature data of the target commodity in the commodity category according to the plurality of dimensional features;
and determining the recommendation score of the target commodity on a target platform according to the plurality of dimensional feature data and the commodity scoring model.
2. The method for selecting commodities according to claim 1, wherein the training and modeling of feature attribute data of a plurality of dimensional features according to a preset score rule of each dimensional feature to obtain a commodity score model of the commodity category comprises:
according to the scoring rule of each dimension characteristic, respectively training and modeling the characteristic attribute data of the dimension characteristics by adopting a plurality of machine learning algorithms to obtain a plurality of scoring models to be evaluated;
and evaluating the scoring model with the highest accuracy and AUC value from the plurality of scoring models to be evaluated as the commodity scoring model.
3. The method for selecting a commodity according to claim 2, wherein the step of respectively training and modeling feature attribute data of a plurality of dimensional features by adopting a plurality of machine learning algorithms according to the score rule of each dimensional feature to obtain a plurality of scoring models to be evaluated comprises the steps of:
and respectively training and modeling the feature attribute data of the plurality of dimensional features by adopting a plurality of machine learning algorithms according to the scoring rule of each dimensional feature and the weight of each dimensional feature to obtain a plurality of models to be assessed.
4. The merchandise selection method of claim 3, further comprising:
carrying out back propagation on the commodity scoring model by using a gradient descent mode, and updating the weight of each dimension characteristic; the counter-propagating comprises:
calling the relationship between the predicted value of the input layer and the real value of the output layer of the commodity scoring model through iterative processing;
the weight of each dimensional feature is updated with minimized error from the reverse direction of the output, hidden and input layers.
5. The method of selecting merchandise of claim 2, wherein the plurality of dimensional characteristics includes a brand attribute, an e-commerce dimensional attribute, a user group preference attribute, a popularity attribute, and a profit margin attribute.
6. The method of merchandise selection according to claim 5, wherein the e-commerce dimensional attributes include popular and unpopular;
before training and modeling the feature attribute data of the multiple dimension features according to the preset score rule of each dimension feature to obtain the commodity score model of the commodity category, the method comprises the following steps:
setting the predicted labels as popular or unpopular, and respectively modeling and training the commodity characteristic historical data by adopting the plurality of machine learning algorithms to obtain a plurality of commodity sales prediction models to be evaluated;
and evaluating the commodity sales forecasting model with the highest accuracy and AUC value from the plurality of to-be-evaluated commodity sales forecasting models to serve as a commodity popularity forecasting model.
7. The merchandise selection method of any one of claims 1-6, further comprising:
acquiring commodity characteristic historical data from commodity historical information data acquired from an e-commerce website and/or an e-shopping website in advance according to commodity categories through big data mining and analysis;
and acquiring user group preference history data from a pre-acquired commodity sales order of the target platform according to the commodity category.
8. The method for selecting commodities according to any one of claims 1 to 6, wherein the training and modeling of feature attribute data of a plurality of dimensional features according to a preset score rule of each dimensional feature to obtain a commodity score model of the commodity category comprises:
cleaning the characteristic attribute data, and performing thermal coding treatment on the cleaned characteristic attribute data;
splitting each feature attribute data subjected to thermal coding into a plurality of attributes to obtain thermal coding data of a plurality of dimensional features;
normalizing the thermally encoded data for the plurality of dimensional features;
and training and modeling the normalized thermal coding data of the multiple dimension characteristics according to a preset score rule of each dimension characteristic to obtain a commodity score model of the commodity category.
9. An article selection system, comprising:
a memory, a processor, and a computer program stored on the memory and executable on the processor;
the computer program, when being executed by the processor, realizes the steps of the method of merchandise selection according to any one of claims 1-8.
10. A computer-readable storage medium, characterized in that the computer-readable storage medium has stored thereon an item selection program, which when executed by a processor implements the steps of the item selection method according to any one of claims 1-8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011588745.6A CN112613953A (en) | 2020-12-29 | 2020-12-29 | Commodity selection method, system and computer readable storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011588745.6A CN112613953A (en) | 2020-12-29 | 2020-12-29 | Commodity selection method, system and computer readable storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112613953A true CN112613953A (en) | 2021-04-06 |
Family
ID=75248650
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011588745.6A Pending CN112613953A (en) | 2020-12-29 | 2020-12-29 | Commodity selection method, system and computer readable storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112613953A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113781107A (en) * | 2021-08-27 | 2021-12-10 | 湖州市吴兴区数字经济技术研究院 | E-commerce promotion pricing decision-making auxiliary method and system based on big data |
| CN114912968A (en) * | 2021-12-10 | 2022-08-16 | 杭州喵智科技有限公司 | Method and system for personalized product selection of automobile accessory supply chain |
| CN116911962A (en) * | 2023-09-14 | 2023-10-20 | 长威信息科技发展股份有限公司 | Article selecting device and method based on data model |
| CN117391811A (en) * | 2023-11-01 | 2024-01-12 | 深圳市沃客非凡科技有限公司 | Self-defined weight commodity selection system based on commodity feature engineering |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107563841A (en) * | 2017-08-03 | 2018-01-09 | 电子科技大学 | A kind of commending system decomposed that scored based on user |
| CN108960719A (en) * | 2018-06-29 | 2018-12-07 | 北京京东尚科信息技术有限公司 | Selection method and apparatus and computer readable storage medium |
| CN109961307A (en) * | 2017-12-25 | 2019-07-02 | 北京京东尚科信息技术有限公司 | The appraisal procedure and device of object-oriented |
| CN110020905A (en) * | 2018-01-09 | 2019-07-16 | 阿里巴巴集团控股有限公司 | A kind of method, apparatus and system of digitization selection |
| CN110246007A (en) * | 2019-05-28 | 2019-09-17 | 中国联合网络通信集团有限公司 | A kind of Method of Commodity Recommendation and device |
| CN111028062A (en) * | 2019-12-12 | 2020-04-17 | 名创优品（横琴）企业管理有限公司 | Potential commodity mining method and device and computer readable storage medium |
| CN111738785A (en) * | 2019-03-28 | 2020-10-02 | 北京京东尚科信息技术有限公司 | Product selection method, system and storage medium |
| KR20200132763A (en) * | 2019-05-16 | 2020-11-25 | (주)제이코스 | Goods recommending machine learning server for walfare internet shopping mall |
-
2020
- 2020-12-29 CN CN202011588745.6A patent/CN112613953A/en active Pending
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107563841A (en) * | 2017-08-03 | 2018-01-09 | 电子科技大学 | A kind of commending system decomposed that scored based on user |
| CN109961307A (en) * | 2017-12-25 | 2019-07-02 | 北京京东尚科信息技术有限公司 | The appraisal procedure and device of object-oriented |
| CN110020905A (en) * | 2018-01-09 | 2019-07-16 | 阿里巴巴集团控股有限公司 | A kind of method, apparatus and system of digitization selection |
| CN108960719A (en) * | 2018-06-29 | 2018-12-07 | 北京京东尚科信息技术有限公司 | Selection method and apparatus and computer readable storage medium |
| CN111738785A (en) * | 2019-03-28 | 2020-10-02 | 北京京东尚科信息技术有限公司 | Product selection method, system and storage medium |
| KR20200132763A (en) * | 2019-05-16 | 2020-11-25 | (주)제이코스 | Goods recommending machine learning server for walfare internet shopping mall |
| CN110246007A (en) * | 2019-05-28 | 2019-09-17 | 中国联合网络通信集团有限公司 | A kind of Method of Commodity Recommendation and device |
| CN111028062A (en) * | 2019-12-12 | 2020-04-17 | 名创优品（横琴）企业管理有限公司 | Potential commodity mining method and device and computer readable storage medium |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113781107A (en) * | 2021-08-27 | 2021-12-10 | 湖州市吴兴区数字经济技术研究院 | E-commerce promotion pricing decision-making auxiliary method and system based on big data |
| CN114912968A (en) * | 2021-12-10 | 2022-08-16 | 杭州喵智科技有限公司 | Method and system for personalized product selection of automobile accessory supply chain |
| CN116911962A (en) * | 2023-09-14 | 2023-10-20 | 长威信息科技发展股份有限公司 | Article selecting device and method based on data model |
| CN116911962B (en) * | 2023-09-14 | 2024-03-29 | 长威信息科技发展股份有限公司 | Article selecting device and method based on data model |
| CN117391811A (en) * | 2023-11-01 | 2024-01-12 | 深圳市沃客非凡科技有限公司 | Self-defined weight commodity selection system based on commodity feature engineering |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110222272B (en) | Potential customer mining and recommending method | |
| CN107562818B (en) | Information recommendation system and method | |
| CN112613953A (en) | Commodity selection method, system and computer readable storage medium | |
| CN109711955B (en) | Poor evaluation early warning method and system based on current order and blacklist base establishment method | |
| US20090125382A1 (en) | Quantifying a Data Source's Reputation | |
| CN110288484B (en) | Insurance classification user recommendation method and system based on big data platform | |
| CN110717654B (en) | Product quality evaluation method and system based on user comments | |
| CN111949887A (en) | Item recommendation method and device and computer-readable storage medium | |
| CN116304299A (en) | Personalized recommendation method integrating user interest evolution and gradient promotion algorithm | |
| CN110910207A (en) | Method and system for improving commodity recommendation diversity | |
| CN111861679A (en) | Commodity recommendation method based on artificial intelligence | |
| CN114861050A (en) | Feature fusion recommendation method and system based on neural network | |
| CN115310722A (en) | Agricultural product price prediction method based on data statistics | |
| CN111898860A (en) | Site selection and operation strategy generation method for digital audio-visual place and storage medium | |
| CN111080396A (en) | Commodity pricing method and device based on competitive commodities and storage medium | |
| CN108960954B (en) | Content recommendation method and system based on user group behavior feedback | |
| CN115344767A (en) | Supplier evaluation method based on network data | |
| CN109960777B (en) | Personalized recommendation method and system for article comment, electronic equipment and storage medium | |
| CN110956528A (en) | Recommendation method and system for e-commerce platform | |
| Chebil et al. | Clustering social media data for marketing strategies: Literature review using topic modelling techniques | |
| Nune | Characterizing and predicting early reviewers for effective product marketing on ecommerce websites | |
| Haidn | Flight recommender system using implicit feedback | |
| KARMAGATRI et al. | Naive Bayes Sentiment Analysis On Perceptions Of Halal Certification: A Case Study On Mixue Indonesia | |
| CN117035947B (en) | Agricultural product data analysis method and cloud platform based on big data processing | |
| Madhumita et al. | Health Insurance Premium Prediction using Blockchain Technology and Random Forest Regression Algorithm |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |