CN117236996B - User behavior prediction method and system based on big data analysis - Google Patents
User behavior prediction method and system based on big data analysis Download PDFInfo
- Publication number
- CN117236996B CN117236996B CN202311161402.5A CN202311161402A CN117236996B CN 117236996 B CN117236996 B CN 117236996B CN 202311161402 A CN202311161402 A CN 202311161402A CN 117236996 B CN117236996 B CN 117236996B
- Authority
- CN
- China
- Prior art keywords
- buyer
- user
- seller
- time period
- evaluation value
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 67
- 238000007405 data analysis Methods 0.000 title claims abstract description 39
- 230000006399 behavior Effects 0.000 claims abstract description 156
- 238000011156 evaluation Methods 0.000 claims abstract description 146
- 238000010801 machine learning Methods 0.000 claims abstract description 22
- 238000012545 processing Methods 0.000 claims description 25
- 238000005457 optimization Methods 0.000 claims description 19
- 230000001932 seasonal effect Effects 0.000 claims description 19
- 230000001737 promoting effect Effects 0.000 claims description 17
- 238000012549 training Methods 0.000 claims description 6
- 230000001680 brushing effect Effects 0.000 abstract description 4
- 238000005516 engineering process Methods 0.000 abstract description 4
- 230000000694 effects Effects 0.000 description 13
- 230000009471 action Effects 0.000 description 7
- 230000008901 benefit Effects 0.000 description 7
- 230000009286 beneficial effect Effects 0.000 description 5
- 230000002159 abnormal effect Effects 0.000 description 4
- 238000004458 analytical method Methods 0.000 description 4
- 230000003542 behavioural effect Effects 0.000 description 4
- 238000011161 development Methods 0.000 description 4
- 230000006872 improvement Effects 0.000 description 4
- 230000001965 increasing effect Effects 0.000 description 4
- 230000008569 process Effects 0.000 description 4
- 206010000117 Abnormal behaviour Diseases 0.000 description 3
- 230000002349 favourable effect Effects 0.000 description 3
- 230000008014 freezing Effects 0.000 description 3
- 238000007710 freezing Methods 0.000 description 3
- 230000002265 prevention Effects 0.000 description 3
- 238000012216 screening Methods 0.000 description 3
- 239000002131 composite material Substances 0.000 description 2
- 230000007246 mechanism Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000009467 reduction Effects 0.000 description 2
- 238000012954 risk control Methods 0.000 description 2
- 238000011282 treatment Methods 0.000 description 2
- 230000004075 alteration Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 238000011217 control strategy Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 230000003014 reinforcing effect Effects 0.000 description 1
- 238000012552 review Methods 0.000 description 1
- 238000000638 solvent extraction Methods 0.000 description 1
- 238000010257 thawing Methods 0.000 description 1
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention provides a user behavior prediction method and a system based on big data analysis, wherein the method comprises the following steps: collecting buyer and corresponding seller data; obtaining a first user representation of a buyer; dividing different time periods according to season attributes and promotion attributes; obtaining a second user representation of different time periods; obtaining a comprehensive evaluation value of the seller; judging the potential bill probability of the seller according to the comprehensive evaluation value of the seller; obtaining a first comprehensive evaluation value of the buyer according to the second user portraits in different time periods; collecting equipment data of the buyer on an e-commerce platform, and obtaining second behavior data of the buyer according to the equipment data; obtaining potential bill probability of the buyer according to the first comprehensive evaluation value, the second behavior data and the seller comprehensive evaluation value of the buyer; by the method and the system, the behavior of brushing is found and prevented in time, the threshold judgment model is trained by a machine learning technology, the intelligent level of the system is improved, and the prediction and judgment are more automatic and efficient.
Description
Technical Field
The invention relates to the technical field of online shopping, in particular to a user behavior prediction method and system based on big data analysis.
Background
With the popularization of the internet, the advantage of online shopping is more prominent, and the online shopping becomes an important shopping form increasingly, and online shopping refers to a fraudulent act of conducting transactions by utilizing an internet platform through false means. The bill reader uses multiple false account numbers or robots to simulate the operations of purchasing goods, submitting orders, evaluating goods, etc., thereby increasing sales and reputation of the merchant to obtain improper benefits. The swiping party is used for disturbing the market order, misleading the purchase decisions of other users and damaging the normal transaction environment and the benefits of merchants; thus requiring the platform to predict and supervise network brush behavior.
Disclosure of Invention
The invention provides a user behavior prediction method based on big data analysis, which is beneficial to timely and accurately predicting and finding the bill-refreshing behavior and maintaining the fairness and the integrity of an e-commerce platform:
the invention provides a user behavior prediction method based on big data analysis, which comprises the following steps:
S1, collecting data of buyers and corresponding sellers on an e-commerce platform; obtaining a first user representation of a buyer;
s2, dividing the first user portrait of the buyer and the corresponding seller sales information into different time periods according to the seasonal attribute and the promotion attribute; obtaining a second user representation of different time periods;
s3, obtaining seller comprehensive evaluation values through seller sales information in different time periods; judging the potential bill probability of the seller according to the comprehensive evaluation value of the seller;
s4, obtaining a first comprehensive evaluation value of the buyer according to the second user portraits in different time periods;
S5, collecting equipment data of the buyer on an e-commerce platform, wherein the equipment data comprises an account IP address; obtaining second behavior data of the buyer according to the equipment data;
s6, obtaining potential bill probability of the buyer according to the first comprehensive evaluation value, the second behavior data and the seller comprehensive evaluation value of the buyer;
And S7, training a threshold judgment model through machine learning, and carrying out user processing and/or threshold optimization according to judgment results.
Further, a user behavior prediction method based on big data analysis, the S1 includes:
Collecting behavior data of a buyer user on an e-commerce platform and extracting characteristics; the behavior data includes first behavior data; the first row of data comprises account registration time, search records, purchase frequency and transaction data;
Collecting information of sellers in transaction data to obtain seller characteristics; the seller characteristics comprise seller name, seller credit and seller sales information;
a first user representation of the buyer is obtained based on the first row of data features and the seller features in the transaction record.
Further, a user behavior prediction method based on big data analysis, the S2 includes:
Dividing the first user portrait of the buyer and the corresponding seller sales information into different time periods according to seasonal attributes and promotional attributes; obtaining second user portraits of different time periods, wherein the different time periods are first statistical time periods; the seller sales information includes seller sales and transaction amounts.
Further, a user behavior prediction method based on big data analysis, the S3 includes:
setting seller thresholds according to sales information of sellers in different time periods, wherein the seller thresholds are sales quantity and good score quantity thresholds in a preset third time period;
Wherein the sales number threshold is
S is rounded upwards, C is the score of the current merchant, and C a is the average score of the similar merchants; h a is the average score of the same class of commodity; h is the current commodity score, S avg is the average sales volume in a third preset time in the first statistical time period; standard deviation of sales in all third preset time in sigma first statistical time period; wherein sales volume and standard deviation only count the time period of the success; t g is the current store registration time, and N is the average sales amount of similar commodities in the same third preset time period; t gavg is the average registration time length of the similar stores;
Commodity good score threshold
Wherein q is rounded upwards, F a is the median score of the similar commodity, and F max is the highest score;
The seller comprehensive evaluation value is:
Wherein W1 and W2 are weight coefficients; s a is the sales quantity of merchants corresponding to the commodities purchased by the user at present at the fourth preset time; q a is the number of good scores obtained by merchants corresponding to the commodities purchased currently by the user in the current third preset time period;
and judging the potential bill probability of the seller according to the comprehensive evaluation value of the seller.
Further, a user behavior prediction method based on big data analysis, the S4 includes:
setting a buyer threshold according to second user portraits in different time periods, wherein the buyer threshold comprises a continuous buying frequency threshold in a first preset time period; repeating the purchase times threshold value within a first preset time period of the same commodity; the threshold value of the number of times of evaluation of the same commodity in a second preset time period;
the continuous buying times threshold value in the first preset time period is as follows:
Wherein, L is rounded upwards, deltaL avg is the average continuous number of times of the user in the first preset time period in the last statistical period and the maximum continuous number of times of the user in the first preset time period in the last same first time statistical period; l max is the largest continuous number of times of ordering in a first preset time period in a first statistical period on the first statistical time period; d s is the number of times the user was judged to be in the last statistical period in the first statistical period; d z is the total number of times of the user in the last statistical period in the first statistical period; f is the frequency of purchases of the user in the last year, alpha is an adjustment coefficient, and the range is (0, 2); f=g/t, g being the total number of purchases by the user in the last year; t is the total number of active days of the user in one year;
threshold value of number of repeated purchases of same commodity in first preset time period R max is the maximum repetition of the user in one statistical period over the first statistical time period; k is a coefficient, and is related to commodity attributes, quick-release products and durable goods; the fast-eliminating product K is larger than the durable goods;
The threshold value P=v a×L;va of the evaluation times in the second preset time period of the same commodity is the annual average evaluation probability of the user and the annual evaluation times divided by the total purchase times;
The first comprehensive evaluation value of the buyer is as follows:
T a is the current commodity browsing time, and T p is the average browsing time when the user purchases the commodity; w3, w4 and w5 are weight coefficients; l a is the number of continuous buying times in the current first preset time period; r a is the number of repeated purchases of the same product in a first preset time period currently performed by the user; p a is the number of times of evaluation of the user on the same product in a second preset time period currently; s a is the sales quantity of merchants corresponding to the commodities purchased by the user at present at the fourth preset time; q a is the number of evaluations obtained by the merchant corresponding to the commodity currently purchased by the user in the current third preset time period.
Further, a user behavior prediction method based on big data analysis, the S5 includes:
Collecting equipment data of a buyer on an e-commerce platform, wherein the equipment data comprises an account IP address; obtaining second behavior data of the buyer according to the equipment data; the second behavior data comprise the number of registered accounts of the buyers, the time of registering the accounts, and the same IP address presets the time and the times of ordering in a fourth time period.
Further, a user behavior prediction method based on big data analysis, the S6 includes:
Obtaining a final evaluation value of the buyer according to the first comprehensive evaluation value of the buyer, the second behavior data and the comprehensive criticizing value of the seller:
m is the number of registered and/or ordered accounts in a fourth preset time period of the same IP address; b i is the first comprehensive evaluation value of the same IP ith account number of the buyer;
and obtaining the potential bill probability of the buyer according to the final evaluation value of the buyer.
Further, a user behavior prediction method based on big data analysis, the S7 includes:
training a threshold judgment model of the comprehensive evaluation value of the buyers and the buyers through machine learning, and checking the terminal larger than the threshold according to threshold judgment;
and carrying out user processing and/or threshold optimization according to the obtained terminal rechecking result.
The invention provides a user behavior prediction system based on big data analysis, which comprises:
A buyer first user portrayal acquisition module: collecting data of buyers and corresponding sellers on an e-commerce platform; obtaining a first user representation of a buyer;
the time period dividing module: dividing the first user portrait of the buyer and the corresponding seller sales information into different time periods according to the seasonal attribute and the promotion attribute; obtaining a second user representation of different time periods;
The seller judging module: obtaining seller comprehensive evaluation values through seller sales information in different time periods; judging the potential bill probability of the seller according to the comprehensive evaluation value of the seller;
The first comprehensive evaluation acquisition module: obtaining a first comprehensive evaluation value of the buyer according to the second user portraits in different time periods;
The second behavior data acquisition module: collecting equipment data of a buyer on an e-commerce platform, wherein the equipment data comprises an account IP address; obtaining second behavior data of the buyer according to the equipment data;
the buyer comprehensive judgment module: obtaining potential bill probability of the buyer according to the first comprehensive evaluation value, the second behavior data and the seller comprehensive evaluation value of the buyer;
and the optimization processing module is used for: through machine learning, a threshold judgment model is trained, and user processing and/or threshold optimization are performed according to judgment results.
The invention has the beneficial effects that: according to the user behavior prediction method and system based on big data analysis, the user behavior is divided into different time periods by collecting the data of the buyers and sellers on the e-commerce platform and combining the seasonal attribute and the promotion attribute, so that the behavior and shopping tendency of the buyers can be predicted more accurately. Meanwhile, the potential bill probability of the seller is judged according to the comprehensive evaluation value of the seller, so that more reliable evaluation and screening of the seller are provided; through the establishment of the first user portrait and the second user portrait of the buyer, the characteristics of purchase preference, consumption capability, search record and the like of the buyer can be more comprehensively known, and a basis is provided for judging whether to brush a line or not. The potential billing probability of the buyer is calculated by collecting the equipment data of the buyer on the e-commerce platform, such as the account IP address, and combining the first comprehensive evaluation value, the second behavior data and the seller comprehensive evaluation value of the buyer. The method is favorable for timely finding and preventing the behavior of the brushing, and the fairness and the integrity of the e-commerce platform are maintained; through machine learning technology, a threshold judgment model is trained, and user processing or threshold optimization can be performed according to judgment results. This helps to increase the level of intelligence in the system, making predictions and judgments more automated and efficient.
In summary, the user behavior prediction method based on big data analysis can improve prediction accuracy, personalize user portraits, detect bill swiping behaviors, and improve the intelligent level of the system, so that development of an e-commerce platform and improvement of user experience are effectively promoted.
Drawings
Fig. 1 is a schematic diagram of a user behavior prediction method based on big data analysis according to the present invention.
Detailed Description
In order that the above-recited objects, features and advantages of the present application will be more clearly understood, a more particular description of the application will be rendered by reference to specific embodiments thereof which are illustrated in the appended drawings. It should be noted that, without conflict, the embodiments of the present application and features in the embodiments may be combined with each other.
In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present invention, and the described embodiments are merely some, rather than all, embodiments of the present invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention.
The embodiment provides a user behavior prediction method based on big data analysis, which comprises the following steps:
S1, collecting data of buyers and corresponding sellers on an e-commerce platform; obtaining a first user representation of a buyer;
s2, dividing the first user portrait of the buyer and the corresponding seller sales information into different time periods according to the seasonal attribute and the promotion attribute; obtaining a second user representation of different time periods;
s3, obtaining seller comprehensive evaluation values through seller sales information in different time periods; judging the potential bill probability of the seller according to the comprehensive evaluation value of the seller;
s4, obtaining a first comprehensive evaluation value of the buyer according to the second user portraits in different time periods;
S5, collecting equipment data of the buyer on an e-commerce platform, wherein the equipment data comprises an account IP address; obtaining second behavior data of the buyer according to the equipment data;
s6, obtaining potential bill probability of the buyer according to the first comprehensive evaluation value, the second behavior data and the seller comprehensive evaluation value of the buyer;
And S7, training a threshold judgment model through machine learning, and carrying out user processing and/or threshold optimization according to judgment results.
The working principle of the technical scheme is as follows: data of buyers and corresponding sellers on an e-commerce platform is collected, and a first user representation of the buyer is created based on the data. Such data includes characteristics such as registration time, registration frequency, purchase preference, consumption ability, search records, and purchase frequency; dividing the first user portrait of the buyer into different time periods according to the seasonal attribute and the promotion attribute, and combining corresponding seller sales information to obtain second user portraits of the buyer in different time periods; setting comprehensive evaluation values of sellers according to seller sales information of different time periods; this composite rating is used to determine the reputation and business capabilities of the seller, and thus infer the potential billing probability of the seller. A first comprehensive evaluation value of the buyer is calculated based on the first user representation and the second user representation of the buyer. The comprehensive evaluation value reflects the behavior characteristics and purchasing trends of buyers; collecting equipment data, such as an account IP address, of the buyer on the e-commerce platform to acquire second behavior data of the buyer; these data help to determine the authenticity and trustworthiness of the buyer; and calculating the potential bill probability of the buyer according to the first comprehensive evaluation value of the buyer, the second behavior data and the comprehensive evaluation value of the seller. This probability is used to predict whether the buyer has a swipe action; the threshold decision model is trained by machine learning techniques. The model processes or optimizes the threshold value of the user according to the judging result so as to accurately identify and predict the bill-refreshing behavior; in general, the method can effectively predict and identify potential bill-refreshing behavior by analyzing the behavior data of buyers and sellers and combining equipment data and a machine learning model, and provides important data support and decision basis for an e-commerce platform.
The technical scheme has the effects that: by collecting the data of the buyers and sellers on the e-commerce platform and combining the seasonal attribute and the promotion attribute, the user behavior is divided into different time periods, and the behavior and shopping tendency of the buyers can be predicted more accurately. Meanwhile, the potential bill probability of the seller is judged according to the comprehensive evaluation value of the seller, so that more reliable evaluation and screening of the seller are provided; through the establishment of the first user portrait and the second user portrait of the buyer, the characteristics of purchasing preference, consumption capability, search record and the like of the buyer can be more comprehensively known. The method and the system are beneficial to the E-commerce platform to provide more accurate personalized recommendation and customization service, and improve user experience and satisfaction. The potential billing probability of the buyer is calculated by collecting the equipment data of the buyer on the e-commerce platform, such as the account IP address, and combining the first comprehensive evaluation value, the second behavior data and the seller comprehensive evaluation value of the buyer. The method is favorable for timely finding and preventing the behavior of the brushing, and the fairness and the integrity of the e-commerce platform are maintained; through machine learning technology, a threshold judgment model is trained, and user processing or threshold optimization can be performed according to judgment results. This helps to increase the level of intelligence in the system, making predictions and judgments more automated and efficient.
In summary, the user behavior prediction method based on big data analysis can improve prediction accuracy, personalize user portraits, detect bill swiping behaviors, and improve the intelligent level of the system, so that development of an e-commerce platform and improvement of user experience are effectively promoted.
The embodiment relates to a user behavior prediction method based on big data analysis, wherein the step S2 includes:
dividing the first user portrait of the buyer and the corresponding seller sales information into different time periods according to seasonal attributes and promotional attributes; obtaining second user portraits of different time periods, wherein the different time periods are first statistical time periods; the sales information of the seller comprises seller sales and transaction quantity; the first statistical time period dividing method is as follows:
Spring promotional/non-promotional: wherein the promotion period comprises 3 months, 4 months, 5 months weekend/holiday promotion and spring festival promotion (if any);
summer promotional/non-promotional: wherein the promotion period comprises weekend/holiday promotion of 6 months, 7 months and 8 months; e.g. 618;
Autumn promotional/non-promotional: wherein the promotion period comprises 9 months, 10 months, 11 months of weekend/holiday promotion and shopping festival promotion (such as double 11);
winter promotion period/non-promotion period: wherein the promotion period comprises 12 month, 1 month, 2 month weekend/holiday promotion, christmas promotion and annual end warehouse promotion.
The working principle of the technical scheme is as follows: first, a first user representation of a buyer and corresponding seller sales information are divided into different time periods according to seasonal and promotional attributes. These time periods are categorized according to seasonal features and promotional campaigns, based on spring, summer, fall and winter. Wherein each season has a specific promotional period and non-promotional period. And acquiring a second user portrait of the buyer in the divided different time periods. The second user representation is a more detailed and comprehensive analysis of the buyer during the first statistical time period, including characteristics of purchase preferences, consumption capabilities, search records, etc. The sales information of the seller includes sales and transaction amounts of the seller. By statistically analyzing this information and combining the customer representation of the buyer, the shopping tendencies and consumption habits of the buyer over different time periods can be inferred. The first statistical time period partitioning method is based on seasonal attributes and promotional attributes. For each season of spring, summer, autumn and winter, the sales promotion period and the non-sales promotion period are divided. For example, spring promotions include weekend/holiday promotions of 3 months, 4 months, 5 months and spring festival promotions (if any); through the steps, the method can combine the user portrait of the buyer and the sales information of the seller, and divide different time periods according to the seasonal attribute and the promotion attribute. Therefore, the shopping behavior and shopping tendency of the buyer can be predicted more accurately, and the basis is used for judging whether to brush a bill or not.
The technical scheme has the effects that: by this method, purchasing behavior and order data of buyers in different time periods can be analyzed. If a buyer purchases a large number of goods frequently in a short time and the purchases are significantly different from those of other real users, the system can identify the abnormal bill-making behavior. By analyzing the first user representation of the buyer and the seller's sales information, the system can capture unusual transaction patterns, such as mass purchasing of promotional items during non-promotional periods or continuous ordering during the same period of time, etc. This helps to find potential brush lines to be effective and take timely action. The statistical time period is divided based on seasonal and promotional attributes, and the system can analyze the purchasing preferences and trends of buyers in different time periods. If a buyer's purchase behavior deviates significantly from the prevailing law, particularly if a large number of items are purchased frequently during a promotion, the system may mark them as potential inventory behavior. The method can timely detect abnormal bill swiping behaviors and provide an early warning mechanism to inform a platform manager or a security team to further investigate and process. By tracking the mode and trend of the bill-refreshing behavior, the e-commerce platform can establish a corresponding prevention and control strategy, and the recognition and prevention capability of the bill-refreshing behavior is improved.
In a word, the user behavior prediction method based on big data analysis can help an e-commerce platform to judge the behavior of the bill, and the accurate identification and timely prevention and control of the behavior of the bill are realized by identifying abnormal modes, analyzing behavior rules and providing an early warning mechanism. The method is beneficial to protecting the fair competition environment of the e-commerce platform, improving the user trust degree and providing more stable operation and development for the platform.
The embodiment relates to a user behavior prediction method based on big data analysis, wherein the step S3 includes:
setting seller thresholds according to sales information of sellers in different time periods, wherein the seller thresholds are sales quantity and good score quantity thresholds in a preset third time period; the preset fourth time period may be one day, two days, three days, one week, etc.;
Wherein the sales number threshold is
S is rounded upwards, C is the score of the current merchant, and C a is the average score of the similar merchants; h a is the average score of the same class of commodity; h is the current commodity score, S avg is the average sales volume in a third preset time in the first statistical time period; standard deviation of sales in all third preset time in sigma first statistical time period; wherein sales volume and standard deviation only count the time period of the success;
T g is the current store registration time, and N is the average sales amount of similar commodities in the same third preset time period; t gavg is the average registration time length of the similar stores;
Commodity good score threshold
Wherein q is rounded upwards, F a is the median score of the similar commodity, and F max is the highest score;
The seller comprehensive evaluation value is:
Wherein W1 and W2 are weight coefficients, and the value range is 0, 1; s a is the sales quantity of merchants corresponding to the commodities purchased by the user at present at the fourth preset time; q a is the number of good scores obtained by merchants corresponding to the commodities purchased currently by the user in the current third preset time period;
judging the potential bill probability of the seller according to the comprehensive evaluation value of the seller; the higher the Y value, the higher the probability of a bill being made.
The working principle of the technical scheme is as follows: according to sales information of sellers in different time periods, a seller threshold is set as a judgment standard. The thresholds include a sales number threshold and a good score number threshold that are preset for a third period of time. The preset fourth period of time may be one day, two days, three days, one week, or the like. The sales volume threshold is calculated according to equation (1).
And (5) calculating a commodity good score threshold according to the formula (2). Where q is the result of rounding up. Fa is the median score for the same class of commodity and Fmax is the highest score. And (3) calculating the comprehensive evaluation value of the seller according to the formula (3). Wherein w1 and w2 are weight coefficients. Judging the potential bill probability of the seller according to the comprehensive evaluation value Y of the seller; the higher the Y value, the higher the probability of a bill being made.
Through the steps, the S3 method can calculate the comprehensive evaluation value of the seller according to the sales information, the grading condition and the purchasing behavior of the user of the seller, and judge the potential bill probability according to the value. Therefore, the bill identification method based on big data analysis can help an e-commerce platform to detect and prevent bill swiping and maintain a fair transaction environment.
The technical scheme has the effects that: by analyzing sales information, scoring conditions and purchasing behavior of the user of the seller, S3 can calculate the comprehensive evaluation value of the seller and judge the potential probability of the bill according to the value. The method can improve the accuracy of the recognition of the bill, and help the e-commerce platform to timely find and prevent the bill from appearing; s3, setting a judgment standard based on the sales number of sellers and the good score number threshold value, so that the bill-refreshing action is more difficult to escape detection. The sales conditions of sellers in different time periods are analyzed, so that abnormal sales quantity and good evaluation quantity can be identified, and the occurrence of bill swiping behavior is prevented; the method can improve the transaction fairness on the e-commerce platform and protect the rights and interests of consumers. Through judging the potential bill probability of the seller, the electronic commerce platform can take corresponding measures, such as limiting the activities of the seller and reducing the weight of the seller, so that the consumer can conduct real and effective transactions; s3, the transaction environment of the electronic commerce platform can be effectively purified by identifying the behavior of the swiping platform. The occurrence of the bill-refreshing action is reduced, the reputation and the reputation of the platform can be improved, and more real buyers and sellers are attracted to participate in the transaction; equation (1) precisely sets the seller's sales amount threshold by considering the seller's score, the like merchant score, the commodity score, the sales amount, and the like. This helps to accurately determine whether the seller has a swipe behavior, improving the accuracy and reliability of the threshold. The sales threshold in equation (1) takes into account both sales and standard deviation. Sales and standard deviation in different time periods are counted, so that sales conditions and fluctuation degrees of sellers can be more comprehensively known, and the probability of the bill is more accurately judged. The commodity good score threshold in the formula (2) considers two factors of the median score and the highest score of the similar commodities. Therefore, the potential bill probability of the seller can be further judged according to the evaluation condition of the commodity. Equation (3) comprehensively evaluates the potential bill probability of the seller according to the sales number and the degree of difference between the good number and the average value of the seller. By defining the weight coefficient, the weights of different factors can be adjusted according to actual conditions, and the probability of the bill is judged more accurately.
In summary, the formula in the big data analysis-based user behavior prediction method (S3) considers a plurality of factors, and determines the potential billing probability of the seller by setting a threshold and a comprehensive evaluation value. The application of the formulas can improve the accuracy and reliability of bill prediction, help an e-commerce platform to discover and prevent bill application in time, and protect the rights and interests of consumers.
The embodiment relates to a user behavior prediction method based on big data analysis, wherein the step S4 includes:
Setting a buyer threshold according to second user portraits in different time periods, wherein the buyer threshold comprises a continuous buying frequency threshold in a first preset time period; repeating the purchase times threshold value within a first preset time period of the same commodity; the threshold value of the number of times of evaluation of the same commodity in a second preset time period; the preset time period may be set to one hour, two hours, four hours, half a day, one day, etc.;
the continuous buying times threshold value in the first preset time period is as follows:
Wherein, L is rounded upwards, deltaL avg is the average continuous number of times of the user in the first preset time period in the last statistical period and the maximum continuous number of times of the user in the first preset time period in the last same first time statistical period; l max is the largest continuous number of times of ordering in a first preset time period in a first statistical period on the first statistical time period; d s is the number of times the user was judged to be in the last statistical period in the first statistical period; d z is the total number of times of the user in the last statistical period in the first statistical period; f is the frequency of purchases of the user in the last year, alpha is an adjustment coefficient, and the range is (0, 2); f=g/t, g being the total number of purchases by the user in the last year; t is the total number of active days of the user in one year;
For example, the first preset time period is set to be 2 hours, the first statistical time period is set to be a period from 3 months to 5 months of spring without sales promotion, and the average continuous number of times of the user in2 hours is set to be 10 times and the maximum number of times of the user in the period of the last first statistical time period is set to be 15 times; in the period of the last first statistics time period, the average continuous number of times of the user in2 hours is 5 times, Δl avg is 5 times, if no statistics value is found in the period of the last first statistics time period, Δl avg is L max and 15 times; wherein the average continuous ordering times only count the time period of the ordering of the user; for example, 3 to 5 am, if the user has no order, the user is not in the statistical range; the user makes a total of 30 passes in the period of the last first statistical time period, and 5 passes are determined to be a brush list, then c s is 5 and c z is 30;
threshold value of number of repeated purchases of same commodity in first preset time period
R max is the maximum repetition of the user in one statistical period over the first statistical time period; k is a coefficient, and is related to commodity attributes, quick-release products and durable goods; the fast-eliminating product K is larger than the durable goods; a range (1, 5);
the threshold value of the number of times of evaluation of the same commodity in a second preset time period;
P＝va×L (6)
v a is the average annual rating probability of the user divided by the total number of purchases;
The first comprehensive evaluation value of the buyer is as follows:
T a is the current commodity browsing time, and T p is the average browsing time when the user purchases the commodity; w3, w4 and w5 are weight coefficients; l a is the number of continuous buying times in the current first preset time period; r a is the number of repeated purchases of the same product in a first preset time period currently performed by the user; p a is the number of times of evaluation of the user on the same product in a second preset time period currently; s a is the sales quantity of merchants corresponding to the commodities purchased by the user at present at the fourth preset time; q a is the number of evaluations obtained by the merchant corresponding to the commodity currently purchased by the user in the current third preset time period.
The working principle of the technical scheme is as follows: user portraits according to different time periods: constructing portraits of users according to the information of purchasing behavior, evaluating behavior and the like of the users in different time periods; and setting a threshold value of the buyer according to the continuous purchase times, the repeated purchase times and the evaluation times in the preset time period. The preset time period may be one hour, two hours, four hours, half a day, one day, etc.; calculating a continuous buying frequency threshold value in a first preset time period: calculating a threshold value of the number of successive purchases according to formula (4); calculating a threshold value of repeated purchasing times of the same commodity in a first preset time period: calculating a threshold value of the number of repeat purchases according to formula (5); calculating the threshold value of the evaluation times of the same product in a second preset time period: calculating a threshold value of the number of evaluation times according to the following formula (6); calculating the comprehensive evaluation value of the buyer: and (5) calculating the comprehensive evaluation value of the buyer according to the formula (7).
The technical scheme has the effects that: by analyzing the indexes such as purchasing behavior, evaluation times and the like of the user in different time periods, the behavior of the user can be predicted more accurately. The algorithm considers a plurality of factors such as purchase frequency, continuous purchase times, repeated purchase times and the like, so that the prediction accuracy is improved; by analyzing the behavior of the user, the system can conduct personalized recommendation to the user according to the preference and buying habit of the user for the commodity. This may increase user purchase satisfaction, promote user repurchase and loyalty. By setting the buyer threshold, the system can determine whether the user is in a billing activity, thereby reducing the risk to the merchant. Meanwhile, the evaluation times of the user on the same commodity can be judged, the merchant is helped to identify potential bad evaluation risks, and the system can calculate the comprehensive evaluation value of the buyer according to different weight coefficients according to the sales quantity and the evaluation quantity of the commodity corresponding to the commodity purchased by the user currently, so that the preference degree of the user on the commodity is judged. Merchants can accurately market according to the information, so that the sales volume of commodities is increased; the method is based on big data analysis, and can automatically predict the user behavior, thereby saving manpower and time cost. Merchants can focus on formulating targeted marketing strategies, so that the efficiency and the effect are improved; by setting the continuous purchasing frequency threshold L and combining the frequency of the bill which is judged to be the bill in the last statistical period of the user and the total bill ordering frequency, whether the user has the bill ordering behavior can be effectively judged. The fairness and the integrity of online transactions can be maintained by reducing the bill swiping behavior; by setting the repeat purchase number threshold R, the number of repeat purchases of the same commodity by the user within the first preset time period can be limited. This helps to prevent the user from making artificial transactions, maintaining the authenticity and validity of the sales of the goods; setting an evaluation frequency threshold according to the proportion of the annual evaluation frequency of the user to the total purchase frequency, and identifying the behavior of the potential bill according to the past evaluation habit of the buyer; by comprehensively considering the purchase behaviors (continuous purchase times, repeated purchase times) and evaluation behaviors (evaluation times) of the user, and combining the sales number of merchants and the obtained evaluation number, a comprehensive evaluation value B can be calculated; the comprehensive evaluation value is used as a basis for judging whether to bill. In general, the model has the advantages that by setting different threshold parameters, whether the user has bad behaviors or not can be judged according to the purchase behaviors and the evaluation behaviors of the user, and corresponding limitation and incentive are carried out, so that the fairness, the authenticity and the credibility of the transaction are maintained.
The method for predicting user behavior based on big data analysis in this embodiment, the S5 includes:
collecting equipment data of a buyer on an e-commerce platform, wherein the equipment data comprises an account IP address; obtaining second behavior data of the buyer according to the equipment data; the second behavior data includes the number of registered accounts of the buyer, the registered account time, the same IP address ordering time and the number of times, for example: and registering a plurality of accounts frequently in a short time in a preset fourth time period.
The working principle of the technical scheme is as follows: first, the system will collect the buyer's device data on the e-commerce platform, including the account IP address. These device data may be used to identify different users and devices.
Obtaining second behavior data: based on big data analysis and machine learning algorithms. By collecting and analyzing the buyer's device data and the second behavioral data, the system can build a user behavioral model. The model may identify abnormal behavior that may exist for the buyer, such as registering multiple accounts frequently for a short period of time. The system automatically predicts these abnormal behaviors and discovers potential problems or risks in advance. Specifically, by analyzing the device data, the system may obtain second behavior data of the buyer. The data includes information such as the number of registered accounts of the buyer, the time of registering the accounts, the time and the number of times of placing orders for the same IP address. For example, if the system detects that the buyer registers a plurality of accounts frequently for a short period of time during a preset fourth period of time, this behavior is recorded as second behavior data. Based on the obtained second behavior data, the system may perform further analysis and prediction. By analyzing the behavior patterns of the buyer's registered account, registration time, and the same IP address, the system can infer the buyer's possible behavior trends and characteristics.
The technical scheme has the effects that: potential behavioral trends of buyers, such as frequent account registration, are predicted in advance. Abnormal behavior of the buyer is detected and identified to prevent fraud, billing or other malicious activity. The safety and the transaction environment of the e-commerce platform are improved, and risks and losses are reduced. In a word, the S5 method provides an effective user behavior prediction mode by collecting equipment data and second behavior data and combining big data analysis and a machine learning algorithm, and can be used for improving the operation and management of an e-commerce platform.
The embodiment relates to a user behavior prediction method based on big data analysis, wherein the step S6 includes:
Obtaining a final evaluation value of the buyer according to the first comprehensive evaluation value of the buyer, the second behavior data and the comprehensive criticizing value of the seller:
m is the number of registered and/or ordered accounts in a fourth preset time period of the same IP address; b i is the first comprehensive evaluation value of the same IP ith account number of the buyer;
And obtaining the potential bill probability of the buyer according to the final evaluation value of the buyer, wherein the larger the G m is, the higher the probability is.
The working principle of the technical scheme is as follows: the method calculates a final evaluation value of the buyer using the first comprehensive evaluation value of the buyer, the second behavior data, and the comprehensive criticizing value of the seller. And then judging the potential bill probability of the buyer according to the final evaluation value. This process is implemented by big data analysis, incorporating the buyer and seller's evaluation information to determine possible billing behavior; first, the system collects the first comprehensive evaluation value of the buyer, the second behavior data, and the comprehensive criticizing value of the seller. These rating values can be used to evaluate the reputation and behavior of the buyer. The final evaluation value of the buyer can be obtained by multiplying the first comprehensive evaluation value of the buyer by the corresponding weight for weighted summation. Wherein m represents the number of accounts registered and/or ordered under the same IP address, and Bi represents the first comprehensive evaluation value of the ith account under the same IP address of the buyer. Based on the final rating of the buyer, the system can infer the potential billing probability of the buyer. In general, the higher the final evaluation value of the buyer, the higher the probability of the bill being made.
The technical scheme has the effects that: the final evaluation value of the buyer can be obtained by comprehensively considering the first comprehensive evaluation value of the buyer, the second behavior data, and the comprehensive criticizing value of the seller. The comprehensive evaluation can more comprehensively and accurately reflect the credit and behavior condition of the buyer. By calculating the final evaluation value of the buyer, the potential billing probability of the buyer can be predicted. The higher the final evaluation value (Gm) of the buyer, the higher the probability of the order. This can be used to discover in advance the possible presence of a billing action, taking corresponding countermeasures. The method is based on big data analysis, and by means of collection and processing of large-scale data, modes and rules hidden behind the data can be well mined. Through analysis of a large amount of data, user behaviors can be more comprehensively known, and accurate prediction and judgment can be performed. By predicting the potential bill-swiping probability of the buyer, the platform or the seller can be helped to take measures in time, such as reinforcing auditing, increasing risk prompt and the like, so as to reduce the influence of the bill-swiping on transaction safety. This helps maintain good order of the transaction platform and trust of the user. By predicting the potential bill swiping probability of the buyer, the transaction with risk can be warned or screened, so that the transaction experience of the user is improved. The occurrence of the bill-swiping action is reduced, and the efficiency and the quality of the real transaction can be improved.
In a word, the user behavior prediction method based on big data analysis can accurately predict potential bill probability of buyers through comprehensive analysis of evaluation values, help a platform or sellers to identify and cope with potential risks, and improve transaction safety and user experience. At the same time, it also relies on adequate and high quality data support, as well as reasonable weight and algorithm selection.
The embodiment relates to a user behavior prediction method based on big data analysis, wherein the step S7 includes:
training a threshold judgment model of the comprehensive evaluation value of the buyers and the buyers through machine learning, and checking the terminal larger than the threshold according to threshold judgment;
User processing and/or threshold optimization are carried out according to the obtained terminal rechecking result; the user processing comprises a first warning prompt, a second reputation reduction prompt and a third temporary freezing transaction, and the user is required to send an application to a background terminal for thawing.
The working principle of the technical scheme is as follows: first, necessary information including behavior data of a buyer, a comprehensive evaluation value, and the like is collected from a large amount of user behavior data. These data will be used to train the model and make the threshold decisions.
And processing and extracting the characteristics of the data by using a machine learning algorithm, and constructing a threshold judgment model of the comprehensive evaluation values of the buyers. The model predicts whether the buyer exceeds a preset threshold based on his behavioral data and the comprehensive evaluation value. And judging the behavior and the comprehensive evaluation value of the buyer through a threshold judgment model. If the behavior and the evaluation value of the buyer are larger than the preset threshold, the system triggers the terminal rechecking operation. And according to the result of terminal rechecking, the system can perform corresponding processing and/or threshold optimization on the buyer. The processing mode comprises a first warning prompt, a second credit reducing prompt and a third temporary freezing transaction. If the buyer needs to defrost the transaction, an application needs to be sent to the background terminal. If the manual review and threshold prediction are different, then threshold optimization is performed, and if the buyer accepts the alert and improves the behavior, the system may also gradually increase the threshold. If the buyer repeatedly violates or does not improve, the system may take more stringent treatment. Through the working principle, S7 can predict and judge the behavior of the buyer through the machine learning model and threshold judgment, so that accurate terminal rechecking operation and corresponding processing measures are provided. The method can effectively identify risk users, reduce malicious behaviors and protect the safety of a transaction platform and the interests of the users. Meanwhile, through user processing and threshold optimization, buyers can be encouraged to follow rules, behaviors are improved, and the quality of the whole transaction environment is improved.
The technical scheme has the effects that: risk control: by means of the threshold judgment model trained by machine learning, whether the behavior of the buyer exceeds a set threshold can be accurately judged. For buyers with behaviors exceeding a threshold, the system can timely check the terminal, so that malicious behaviors and risks on a transaction platform are effectively reduced. According to the result of terminal rechecking, the system can adopt different processing modes for buyers, such as a first warning prompt, a second reputation reduction prompt and a third temporary freezing transaction. These treatments may alert buyers to improve behavior and penalize violations. Meanwhile, through threshold optimization, the system can predict more accurately, and meanwhile, the threshold can be gradually increased according to feedback and behavior improvement conditions of buyers, so that more freedom and trust are given to buyers conforming to rules. The method can effectively identify the risk users, reduce fraud and malicious behaviors and improve the safety of a transaction platform. At the same time, for buyers, they can also better evaluate and select trading partners, improving the reputation of the trade. By predicting and processing the behavior of the buyer, the system can more accurately provide alerts and reminders, help the buyer to recognize improper behavior, and correct errors. This will help improve the quality of the overall transaction environment, enhancing the satisfaction and experience of the user.
In summary, the user behavior prediction method based on big data analysis can provide more accurate risk control and user processing, so that the safety and credibility of a transaction platform are improved, and the transaction experience of a user is optimized. This approach has significant effects and benefits in protecting user benefits, reducing risk, and improving transaction quality.
The embodiment provides a user behavior prediction system based on big data analysis, the system includes:
A buyer first user portrayal acquisition module: collecting data of buyers and corresponding sellers on an e-commerce platform; obtaining a first user representation of a buyer;
the time period dividing module: dividing the first user portrait of the buyer and the corresponding seller sales information into different time periods according to the seasonal attribute and the promotion attribute; obtaining a second user representation of different time periods;
The seller judging module: obtaining seller comprehensive evaluation values through seller sales information in different time periods; judging the potential bill probability of the seller according to the comprehensive evaluation value of the seller;
The first comprehensive evaluation acquisition module: obtaining a first comprehensive evaluation value of the buyer according to the second user portraits in different time periods;
The second behavior data acquisition module: collecting equipment data of a buyer on an e-commerce platform, wherein the equipment data comprises an account IP address; obtaining second behavior data of the buyer according to the equipment data;
the buyer comprehensive judgment module: obtaining potential bill probability of the buyer according to the first comprehensive evaluation value, the second behavior data and the seller comprehensive evaluation value of the buyer;
and the optimization processing module is used for: through machine learning, a threshold judgment model is trained, and user processing and/or threshold optimization are performed according to judgment results.
The working principle of the technical scheme is as follows: data of buyers and corresponding sellers on an e-commerce platform is collected, and a first user representation of the buyer is created based on the data. Such data includes characteristics such as registration time, registration frequency, purchase preference, consumption ability, search records, and purchase frequency; dividing the first user portrait of the buyer into different time periods according to the seasonal attribute and the promotion attribute, and combining corresponding seller sales information to obtain second user portraits of the buyer in different time periods; setting comprehensive evaluation values of sellers according to seller sales information of different time periods; this composite rating is used to determine the reputation and business capabilities of the seller, and thus infer the potential billing probability of the seller. A first comprehensive evaluation value of the buyer is calculated based on the first user representation and the second user representation of the buyer. The comprehensive evaluation value reflects the behavior characteristics and purchasing trends of buyers; collecting equipment data, such as an account IP address, of the buyer on the e-commerce platform to acquire second behavior data of the buyer; these data help to determine the authenticity and trustworthiness of the buyer; and calculating the potential bill probability of the buyer according to the first comprehensive evaluation value of the buyer, the second behavior data and the comprehensive evaluation value of the seller. This probability is used to predict whether the buyer has a swipe action; the threshold decision model is trained by machine learning techniques. The model processes or optimizes the threshold value of the user according to the judging result so as to accurately identify and predict the bill-refreshing behavior; in general, the method can effectively predict and identify potential bill-refreshing behavior by analyzing the behavior data of buyers and sellers and combining equipment data and a machine learning model, and provides important data support and decision basis for an e-commerce platform.
The technical scheme has the effects that: by collecting the data of the buyers and sellers on the e-commerce platform and combining the seasonal attribute and the promotion attribute, the user behavior is divided into different time periods, and the behavior and shopping tendency of the buyers can be predicted more accurately. Meanwhile, the potential bill probability of the seller is judged according to the comprehensive evaluation value of the seller, so that more reliable evaluation and screening of the seller are provided; through the establishment of the first user portrait and the second user portrait of the buyer, the characteristics of purchasing preference, consumption capability, search record and the like of the buyer can be more comprehensively known. The method and the system are beneficial to the E-commerce platform to provide more accurate personalized recommendation and customization service, and improve user experience and satisfaction. The potential billing probability of the buyer is calculated by collecting the equipment data of the buyer on the e-commerce platform, such as the account IP address, and combining the first comprehensive evaluation value, the second behavior data and the seller comprehensive evaluation value of the buyer. The method is favorable for timely finding and preventing the behavior of the brushing, and the fairness and the integrity of the e-commerce platform are maintained; through machine learning technology, a threshold judgment model is trained, and user processing or threshold optimization can be performed according to judgment results. This helps to increase the level of intelligence in the system, making predictions and judgments more automated and efficient.
In summary, the user behavior prediction method based on big data analysis can improve prediction accuracy, personalize user portraits, detect bill swiping behaviors, and improve the intelligent level of the system, so that development of an e-commerce platform and improvement of user experience are effectively promoted.
It will be apparent to those skilled in the art that various modifications and variations can be made to the present invention without departing from the spirit or scope of the invention. Thus, it is intended that the present invention also include such modifications and alterations insofar as they come within the scope of the appended claims or the equivalents thereof.
Claims (2)
1. A user behavior prediction method based on big data analysis, the method comprising:
S1, collecting data of buyers and corresponding sellers on an e-commerce platform; obtaining a first user representation of a buyer;
s2, dividing the first user portrait of the buyer and the corresponding seller sales information into different time periods according to the seasonal attribute and the promotion attribute; obtaining a second user representation of different time periods;
s3, obtaining seller comprehensive evaluation values through seller sales information in different time periods; judging the potential bill probability of the seller according to the comprehensive evaluation value of the seller;
s4, obtaining a first comprehensive evaluation value of the buyer according to the second user portraits in different time periods;
s5, collecting equipment data of the buyer on an e-commerce platform, and obtaining second behavior data of the buyer according to the equipment data;
s6, obtaining potential bill probability of the buyer according to the first comprehensive evaluation value, the second behavior data and the seller comprehensive evaluation value of the buyer;
S7, training a threshold judgment model through machine learning, and performing user processing and/or threshold optimization according to judgment results;
The S1 comprises the following steps:
Collecting behavior data of a buyer user on an e-commerce platform and extracting characteristics; the behavior data includes first behavior data; the first row of data comprises account registration time, search records, purchase frequency and transaction data;
Collecting information of sellers in transaction data to obtain seller characteristics; the seller characteristics comprise seller name, seller credit and seller sales information;
a first user representation of the buyer is obtained based on the first row of data features and the seller features in the transaction record.
The step S2 comprises the following steps:
dividing the first user portrait of the buyer and the corresponding seller sales information into different time periods according to seasonal attributes and promotional attributes; obtaining second user portraits of different time periods, wherein the different time periods are first statistical time periods; the sales information of the seller comprises seller sales and transaction quantity;
The step S3 comprises the following steps:
setting seller thresholds according to sales information of sellers in different time periods, wherein the seller thresholds are sales quantity and good score quantity thresholds in a preset third time period;
Wherein the sales number threshold is
Wherein S is rounded upwards, C is the current merchant score,Average scoring for the same class of merchants; /(I)Average scoring of the same type of commodity; h is the current commodity score,/>Average sales volume in a third preset time in the first statistical time period; /(I)Standard deviation of sales in all third preset time in the first statistical time period; wherein sales volume and standard deviation only count the time period of the success; /(I)The current store registration time length, N is the average sales amount of similar commodities in the same third preset time period; /(I)Average registration duration for the similar stores;
Commodity good score number threshold q=s×
Wherein q is rounded up and down upwards,Score for similar commodity median,/>Top scoring;
The seller comprehensive evaluation value is: y=w1× w2×/>
Wherein W1 and W2 are weight coefficients; Selling the quantity of the merchants corresponding to the commodities purchased currently by the user at the fourth preset time; /(I) The method comprises the steps that the good scoring quantity obtained by a merchant corresponding to the commodity purchased currently by a user in a current third preset time period is obtained;
Judging the potential bill probability of the seller according to the comprehensive evaluation value of the seller;
The step S4 comprises the following steps:
Setting a buyer threshold according to second user portraits in different time periods, wherein the buyer threshold comprises a continuous buying frequency threshold in a first preset time period; repeating the purchase times threshold value within a first preset time period of the same commodity; the threshold value of the number of times of evaluation of the same commodity in a second preset time period;
the continuous buying times threshold value in the first preset time period is as follows:
L=α×
Wherein, L is rounded upwards, For a user in a first statistical time period, the average continuous number of times of making a bill in a first preset time period in the last statistical period is the same as the maximum continuous number of times of making a bill in the first preset time period in the last first time statistical time period; /(I)The method comprises the steps of providing a maximum continuous order number for a user in a first preset time period in a first statistical period; /(I)The number of times of the bill is judged in the last statistical period in the first statistical period for the user; /(I)The method comprises the steps that the total number of times of ordering is counted for the user in a first counting period in the last counting period; f is the frequency of purchases of the user in the last year, alpha is an adjustment coefficient, and the range is (0, 2); f=g/t, g being the total number of purchases by the user in the last year; t is the total number of active days of the user in one year;
repeated purchasing times threshold value R=within first preset time period of same commodity ; />Maximum repetition for a user in a statistical period over a first statistical time period; k is a coefficient related to the property of the commodity;
Threshold p=number of evaluations within a second preset time period for the same product ×L;/>Dividing the average annual evaluation probability of the user and the annual evaluation times by the total purchase times;
The first comprehensive evaluation value of the buyer is as follows:
browsing time for current commodity of ordering/> Average browsing time when the user purchases the commodity; w3, w4 and w5 are weight coefficients; /(I)Continuing to purchase times for the current first preset time period; /(I)Repeating the purchasing times of the user in a first preset time period for the same commodity; /(I)The number of times of evaluation of the same commodity in a second preset time period is currently set for the user; /(I)Selling the quantity of the merchants corresponding to the commodities purchased currently by the user at the fourth preset time; /(I)The method comprises the steps that the evaluation quantity obtained by a merchant corresponding to the commodity purchased currently by a user in a current third preset time period is obtained;
The step S5 comprises the following steps:
Collecting equipment data of a buyer on an e-commerce platform, wherein the equipment data comprises an account IP address; obtaining second behavior data of the buyer according to the equipment data; the second behavior data comprise the number of registered accounts of the buyers, the time of registering the accounts, and the same IP address presets the time and the times of ordering in a fourth time period;
the step S6 comprises the following steps:
Obtaining a final evaluation value of the buyer according to the first comprehensive evaluation value of the buyer, the second behavior data and the comprehensive criticizing value of the seller:
=/>
m is the number of registered and/or ordered accounts in a fourth preset time period of the same IP address; a first comprehensive evaluation value of the same IP ith account number for the buyer;
Obtaining potential bill probability of the buyer according to the final evaluation value of the buyer;
The step S7 comprises the following steps:
training a threshold judgment model of the comprehensive evaluation value of the buyers and the buyers through machine learning, and checking the terminal larger than the threshold according to threshold judgment;
and carrying out user processing and/or threshold optimization according to the obtained terminal rechecking result.
2. A system for implementing a big data analysis based user behavior prediction method according to claim 1, the system comprising:
A buyer first user portrayal acquisition module: collecting data of buyers and corresponding sellers on an e-commerce platform; obtaining a first user representation of a buyer;
the time period dividing module: dividing the first user portrait of the buyer and the corresponding seller sales information into different time periods according to the seasonal attribute and the promotion attribute; obtaining a second user representation of different time periods;
The seller judging module: obtaining seller comprehensive evaluation values through seller sales information in different time periods; judging the potential bill probability of the seller according to the comprehensive evaluation value of the seller;
The first comprehensive evaluation acquisition module: obtaining a first comprehensive evaluation value of the buyer according to the first user portraits in different time periods;
The second behavior data acquisition module: collecting equipment data of a buyer on an e-commerce platform, wherein the equipment data comprises an account IP address; obtaining second behavior data of the buyer according to the equipment data;
the buyer comprehensive judgment module: obtaining potential bill probability of the buyer according to the first comprehensive evaluation value, the second behavior data and the seller comprehensive evaluation value of the buyer;
and the optimization processing module is used for: through machine learning, a threshold judgment model is trained, and user processing and/or threshold optimization are performed according to judgment results.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311161402.5A CN117236996B (en) | 2023-09-07 | 2023-09-07 | User behavior prediction method and system based on big data analysis |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311161402.5A CN117236996B (en) | 2023-09-07 | 2023-09-07 | User behavior prediction method and system based on big data analysis |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117236996A CN117236996A (en) | 2023-12-15 |
| CN117236996B true CN117236996B (en) | 2024-05-28 |
Family
ID=89092336
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311161402.5A Active CN117236996B (en) | 2023-09-07 | 2023-09-07 | User behavior prediction method and system based on big data analysis |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117236996B (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110288362A (en) * | 2019-07-03 | 2019-09-27 | 北京工业大学 | Brush single prediction technique, device and electronic equipment |
| KR20200009542A (en) * | 2018-07-19 | 2020-01-30 | 한국기술교육대학교 산학협력단 | Intelligent Shopping Platform Management System and Its Methods Based on Consumer Big Data |
| CN113362126A (en) * | 2021-05-21 | 2021-09-07 | 萨科(深圳)科技有限公司 | Order data detection method and system applied to online business |
| CN115511582A (en) * | 2022-10-31 | 2022-12-23 | 深圳市快云科技有限公司 | Artificial intelligence based Commodity recommendation system and method |
-
2023
- 2023-09-07 CN CN202311161402.5A patent/CN117236996B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20200009542A (en) * | 2018-07-19 | 2020-01-30 | 한국기술교육대학교 산학협력단 | Intelligent Shopping Platform Management System and Its Methods Based on Consumer Big Data |
| CN110288362A (en) * | 2019-07-03 | 2019-09-27 | 北京工业大学 | Brush single prediction technique, device and electronic equipment |
| CN113362126A (en) * | 2021-05-21 | 2021-09-07 | 萨科(深圳)科技有限公司 | Order data detection method and system applied to online business |
| CN115511582A (en) * | 2022-10-31 | 2022-12-23 | 深圳市快云科技有限公司 | Artificial intelligence based Commodity recommendation system and method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117236996A (en) | 2023-12-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108573432B (en) | Transaction supervision system and method for electronic commerce | |
| CN108038696B (en) | Method and system for detecting bill swiping based on equipment identification code and social group information | |
| US8175908B1 (en) | Systems and methods for constructing and utilizing a merchant database derived from customer purchase transactions data | |
| US20150332414A1 (en) | System and method for predicting items purchased based on transaction data | |
| CN104866484B (en) | A kind of data processing method and device | |
| KR100613891B1 (en) | Customer Asset Valuation System | |
| US8620746B2 (en) | Scoring quality of traffic to network sites | |
| Fougère et al. | Heterogeneity in consumer price stickiness: a microeconometric investigation | |
| US20160321661A1 (en) | Systems and methods for organizing, visualizing and processing consumer transactions data | |
| US20090307028A1 (en) | A method and a system for identifying potentially fraudulent customers in relation to electronic customer action based systems, and a computer program for performing said method | |
| US20180130071A1 (en) | Identifying and labeling fraudulent store return activities | |
| CN107103548A (en) | The monitoring method and system and risk monitoring and control method and system of network behavior data | |
| CN106506454A (en) | Fraud business recognition method and device | |
| CN110246030A (en) | In many ways risk management method, terminal, device and storage medium after the loan to link | |
| CN115953021B (en) | Vendor risk analysis method and device based on machine learning | |
| US20090192878A1 (en) | Method and apparatus for utilizing shopping survey data | |
| CN108346088A (en) | Brand official store malice panic buying preventing control method and system | |
| CN116611796B (en) | Exception detection method and device for store transaction data | |
| CN117236996B (en) | User behavior prediction method and system based on big data analysis | |
| CN116629998A (en) | Automatic information counting method and device, electronic equipment and readable storage medium | |
| CN109544299A (en) | Buyer's identity ranking method, equipment and the storage medium of platform are ensured based on transaction | |
| CN114463022A (en) | Intelligent sales management system based on big data | |
| US7606727B1 (en) | System and method for identifying optimal marketing offers | |
| JP6064963B2 (en) | Sales management device and program | |
| CN115131070B (en) | Online commodity predetermined quantity virtual mark identification and processing method and device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |