Disclosure of Invention
The present invention has been made in view of the above-described problems occurring in the prior art.
Therefore, the invention provides a method for analyzing the consumption behaviors of users in different scenes in a commercial space, which can solve the problems of data island, insufficient dynamic adaptability, lack of a user feedback mechanism and the like in the prior art, and can enhance the real-time response capability by comprehensively integrating multi-source data, thereby realizing intelligent management in the commercial space.
The technical scheme includes that user behavior data are acquired through a wireless positioning device and a distributed environment sensor, multi-source heterogeneous data are subjected to space-time alignment, commodity granularity data of a transaction system are fused to construct a three-dimensional track model, resident features and moving modes of users are recognized based on the three-dimensional track model, dynamic thermodynamic diagrams are generated by associating commodity distribution, guiding strategies and environment parameters are adjusted according to density gradients and user attribute levels of the thermodynamic diagrams, and user labels are updated by integrating payment feature data and user behavior data to generate cross-region association strategies.
The wireless positioning device comprises a wireless positioning device, a wireless processing device and a control device, wherein the wireless positioning device is used for receiving mobile equipment signals carried by a user to position a moving track of the user in a commercial space;
the distributed environment sensor comprises an RFID reader and a pressure sensing device which are deployed in a functional area and used for recording attribute information and interaction frequency of operated objects.
The invention relates to a method for analyzing consumption behaviors of users in different scenes in a commercial space, which comprises the steps of performing space-time alignment on multi-source heterogeneous data, constructing a central API gateway, receiving and routing data requests from a wireless positioning device and a distributed environment sensor, converting different data structures of each device into a uniform format by utilizing a mapping table, and transmitting the uniform format to a data processing platform;
Introducing a self-adaptive heterogeneous data analyzer, automatically identifying and processing data from different sources according to the data characteristics, and removing noise data and redundant information;
Each piece of data is assigned with a unique time stamp and a geographic positioning identifier, a connection is established between the activity track of the user and dynamic environment data stored in a database, the similarity of the time stamp and the position identifier in space and time is calculated, and the data are dynamically associated.
The method for analyzing the consumption behaviors of the users in different scenes in the commercial space comprises the following steps of constructing a three-dimensional track model, wherein the construction of the three-dimensional track model adopts cooperative processing based on a cloud computing technology, and constructing a multi-level cloud computing architecture;
The edge computing nodes are arranged on a user access layer, user data are collected and primarily processed through edge equipment, a lightweight data processing module is developed in each edge node, real-time analysis is carried out on user behavior data, and the cleaned data are transmitted to a central cloud node through a secure encryption channel for further analysis;
And creating a RESTful API, communicating the edge node with the cloud, and carrying out bidirectional data transmission and command issuing.
The invention is used as a preferred scheme of the user consumption behavior analysis method of different scenes in the commercial space, wherein the identifying of the residence characteristic and the movement mode of the user comprises extracting the behavior characteristics of residence time, frequency and movement speed from the user behavior data and constructing a user activity model;
the behavior cloning method in reinforcement learning is applied to simulate a typical movement mode of a user in a commercial space, the change of user behavior data is continuously monitored through a real-time data feedback mechanism, a stream data analysis technology is adopted to capture the user behavior trend in real time, and the parameters of a user activity model are adjusted;
Through the incremental learning method, under the condition that the user activity model is not retrained, parameter adjustment is carried out only based on newly added data, so that the user activity model always reflects the latest user behavior characteristics.
The method for analyzing the consumption behaviors of the users in different scenes in the commercial space comprises the steps of converting user behavior data into a thermodynamic diagram format and distinguishing the heat of different time periods;
Capturing a nonlinear relation in user behavior data, identifying behavior characteristics of user activities, mapping the behavior characteristics to a thermodynamic diagram space through a quadratic or cubic polynomial function, dividing a user hot spot area into high, medium and low heat areas, wherein each heat area corresponds to different colors, and planning a detour path to a low-density waiting area for a high-consumption-level user preferentially;
And integrating a simple user feedback investigation function, combining user feedback with thermodynamic diagram data, calculating the influence of user satisfaction on the thermodynamic diagram data by using a simple weighting model, and dynamically updating the thermodynamic diagram according to the result.
The invention relates to a method for analyzing consumption behaviors of users in different scenes in a commercial space, which comprises the steps of generating a behavior sequence comprising a time stamp, coordinates and stay time length by collecting a user movement track;
establishing a multidimensional fusion index, associating user behaviors and consumption characteristics into unified data objects, and marking the unified data objects as potential associated behaviors when a user stays in any functional area for a long time but does not consume the data objects and generates guest prices higher than the average value in the remaining area;
Constructing a multistage label system, wherein the multistage label system comprises a basic attribute layer, a behavior derivative layer and a deep interaction type, and dynamically correcting label weights by adopting an incremental learning model, namely reducing the association confidence coefficient of a current label and a class when the behavior of a label user deviates from a historical mode;
when the user leaves the current area, the optimal association area is matched according to the label, wherein the adjacent area with the promotion activity is preferentially pushed to the 'price sensitive' user for navigation;
When any functional area gathers label users, automatically triggering a cooperative strategy, adjusting commodity display density of adjacent areas, and highlighting related products;
And a module for deploying behavior feedback analysis, wherein the user behavior change after the policy execution is used as an optimization signal, and when any associated policy does not reach an expected target, the current policy priority is automatically degraded.
The invention relates to a preferable scheme of a user consumption behavior analysis system of different scenes in a commercial space, which comprises a data acquisition module, a data processing module, a user behavior analysis module, a dynamic thermodynamic diagram generation module and a strategy generation and execution module;
the data acquisition module acquires behavior data of a user in real time through the wireless positioning device and the distributed environment sensor, is responsible for receiving equipment signals carried by the user, environment sensor data and data of the RFID reader, and records the moving track, interaction frequency and related attribute information of the user in a commercial space;
the data processing module performs space-time alignment on the collected multi-source heterogeneous data, performs data cleaning and conversion, and constructs a structured data set for analysis;
The user behavior analysis module is used for extracting resident features and movement modes of the user based on the cleaned user behavior data and optimizing a user activity model by using a self-adaptive learning mechanism;
The dynamic thermodynamic diagram generation module converts the processed user behavior data into a thermodynamic diagram format, generates user activity thermodynamic diagrams of different time periods, and displays the region of interest of the user for the merchant;
The strategy generation and execution module generates targeted business strategies based on analysis results, including cross-regional association strategies, and adjusts commodity display and user guidance in real time.
A computer device comprising a memory storing a computer program and a processor implementing the steps of a method of user consumption behavior analysis of different scenarios in a business space when the computer program is executed.
A computer readable storage medium having stored thereon a computer program which when executed by a processor implements the steps of a method of consumer behavior analysis of different scenarios within a business space.
The method has the beneficial effects that the real-time understanding and response capability of the consumption behavior of the user are improved through the comprehensive analysis of the multi-source data in the commercial space. By fusing the wireless positioning technology and the user behavior data of the distributed environment sensor, the invention constructs a new user activity model and a three-dimensional track model, thereby being capable of more accurately identifying the resident characteristics and the movement modes of the user.
The comprehensive capture and deep analysis of the user behaviors are realized, and the merchant can clearly identify the distribution and consumption trend of the good of the user through dynamic thermodynamic diagram generation, so that the commodity layout and guiding strategy are optimized, and the user experience is improved. Meanwhile, a user feedback investigation function and a dynamic updating mechanism are introduced, so that the system can adjust a behavior analysis strategy according to real-time data and user satisfaction, the subjective experience and objective data of the user are fused, and a more targeted decision basis is provided for merchants.
Detailed Description
So that the manner in which the above recited objects, features and advantages of the present invention can be understood in detail, a more particular description of the invention, briefly summarized above, may be had by reference to the embodiments, some of which are illustrated in the appended drawings. All other embodiments, which can be made by one of ordinary skill in the art based on the embodiments of the present invention without making any inventive effort, shall fall within the scope of the present invention.
In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present invention, but the present invention may be practiced in other ways other than those described herein, and persons skilled in the art will readily appreciate that the present invention is not limited to the specific embodiments disclosed below.
Further, reference herein to "one embodiment" or "an embodiment" means that a particular feature, structure, or characteristic can be included in at least one implementation of the invention. The appearances of the phrase "in one embodiment" in various places in the specification are not necessarily all referring to the same embodiment, nor are separate or alternative embodiments mutually exclusive of other embodiments.
While the embodiments of the present invention have been illustrated and described in detail in the drawings, the cross-sectional view of the device structure is not to scale in the general sense for ease of illustration, and the drawings are merely exemplary and should not be construed as limiting the scope of the invention. In addition, the three-dimensional dimensions of length, width and depth should be included in actual fabrication.
Also in the description of the present invention, it should be noted that the orientation or positional relationship indicated by the terms "upper, lower, inner and outer", etc. are based on the orientation or positional relationship shown in the drawings, are merely for convenience of describing the present invention and simplifying the description, and do not indicate or imply that the apparatus or elements referred to must have a specific orientation, be constructed and operated in a specific orientation, and thus should not be construed as limiting the present invention. Furthermore, the terms "first, second, or third" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance.
The terms "mounted, connected, and coupled" as used herein, unless otherwise specifically indicated and defined, shall be construed broadly and include, for example, fixed, removable, or integral, as well as mechanical, electrical, or direct, as well as indirect via intermediaries, or communication between two elements. The specific meaning of the above terms in the present invention will be understood in specific cases by those of ordinary skill in the art.
Embodiment 1, referring to fig. 1, is a first embodiment of the present invention, which provides a method for analyzing consumer behavior of different scenes in a business space, including:
s1, acquiring user behavior data through a wireless positioning device and a distributed environment sensor.
Further, the wireless positioning device comprises a mobile device signal receiving device, a Bluetooth network base station, a Wi-Fi network base station, a wireless positioning device and a wireless positioning device, wherein the mobile device signal receiving device is used for receiving a mobile device signal carried by a user to position a moving track of the user in a commercial space;
After a user enters a commercial space, the positioning device is paired with a mobile phone of the user through Bluetooth, and Wi-Fi signal intensity is obtained. And fusing the accurate position of the Bluetooth signal with Wi-Fi information through multipath propagation and signal back propagation technologies, and calculating the real-time position of the user.
The distributed environment sensor comprises an RFID reader and a pressure sensing device which are arranged in a functional area and are used for recording attribute information and interaction frequency of the operated objects.
And S2, carrying out space-time alignment on the multi-source heterogeneous data, and fusing commodity granularity data of a transaction system to construct a three-dimensional track model.
Further, a central API gateway is constructed, data requests from the wireless positioning device and the distributed environment sensor are received and routed, and different data structures of all the devices are converted into a uniform format by utilizing a mapping table and are transmitted to a data processing platform;
Introducing a self-adaptive heterogeneous data analyzer, automatically identifying and processing data from different sources according to the data characteristics, and removing noise data and redundant information;
Each piece of data is assigned with a unique time stamp and a geographic positioning identifier, and a connection is established between the dynamic environment data stored in the database and the user's activity track. The space-time alignment adopts an improved sliding window protocol, namely an RFID trigger event is taken as a reference time stamp, visual data is matched with a skeleton gesture in a window of +/-300 ms, wi-Fi signal strength is subjected to interpolation positioning after multipath effect compensation, and the attribute distribution of the non-transacted article is calculated by comparing RFID activation times with POS transaction record difference values.
The space index is built according to the position (business district, floor), the time index is built according to the time period (hour, day), and the time and the space information are associated through the bi-directional association pointer, so that when one index is accessed, the associated other dimension data can be quickly obtained, and the data query efficiency is improved.
It should be noted that, the construction of the three-dimensional track model adopts cooperative processing based on the cloud computing technology to construct a multi-level cloud computing architecture;
The edge computing node is arranged at a user access layer, and is used for collecting and primarily processing user data through the edge equipment, processing the original data flow provided by the mobile equipment, and the processing throughput is designed to be 10 ten thousand pieces/second. The data stream includes the user's location information, a time stamp, and additional environmental data (e.g., temperature, humidity, etc.).
In each edge node, a lightweight data processing module is developed, lightweight data preprocessing comprises data cleaning, noise filtering and the like, user behavior data are analyzed in real time, and the cleaned data are transmitted to a central cloud node through a secure encryption channel for further analysis;
Furthermore, when the central node performs user behavior analysis, a cubic spline interpolation method is used for smoothing the moving speed, and the user speed data is smoothed with high precision, so that noise is eliminated, and more real speed characteristics are obtained;
And performing stream data processing by using Spark Structured Streaming, setting the size of a sliding window to be 60 seconds, and performing real-time aggregation analysis. This process includes aggregate statistics on user behavior, such as calculating the average residence time and movement speed of each user over the last 60 seconds.
Further, a multidimensional feature vector is established, corresponding features are established for each commodity by analyzing commodity categories associated with each resident event, and accuracy is improved. And setting a trigger mechanism of newly-increased user behavior data every day, and performing three-dimensional track model fine adjustment on the basis.
And dynamically slicing the real-time data according to the time range, the user identity or the geographic position. And creating RESTful API to make the edge node and cloud end intercommunicate to make bidirectional data transmission and command issuing.
And S3, identifying resident features and movement modes of the user based on the three-dimensional track model, associating commodity distribution to generate a dynamic thermodynamic diagram, and adjusting a guiding strategy and environmental parameters according to density gradients of the thermodynamic diagram and a user attribute level.
Further, behavior features of residence time, frequency and moving speed are extracted from the user behavior data, a Deep Neural Network (DNN) is used for fitting a moving track of a user and constructing a user activity model;
it should be noted that the input layer of the user activity model receives a variety of status information including, but not limited to:
The user location information, i.e., the current coordinates (x, y), may be the real-time coordinate location of the user within the business space.
Time stamps for capturing temporal characteristics of user behavior (e.g., hours, weekends/weekdays).
User historical behavior, including the user's recent several behaviors (time and location since last stay), to help the user activity model capture the user's habits.
The hidden layer of the user activity model uses 3-5 layers of fully connected hidden layers, each layer containing multiple neurons:
The method uses a ReLU (RECTIFIED LINEAR Unit) activation function to introduce nonlinearity and accelerate convergence, adopts Dropout (discard method) to reduce the risk of overfitting and improve generalization capability.
The output layer is used for predicting the next behavior of the user:
When coarse behavior prediction is carried out, two nodes are output as two classification problems, and the probabilities of 'movement' and 'stay' are respectively represented, wherein F1-score of movement/stay classification is more than or equal to 0.92, and coarse granularity behavior judgment reliability is ensured.
When accurate next position prediction is performed, a regression layer is created to output the next position coordinates (x ', y ') of the user, providing a specific position of the user's intended behavior.
Further, the mean value of Euclidean distance error (MAE) between the predicted position and the true coordinates, whenAnd judging convergence when the rice is in time. The calculation mode is as follows:
;
Wherein,As a total number of samples,、Respectively the horizontal and vertical coordinate values of the predicted position,、Respectively the abscissa and ordinate values of the true position.
The behavior cloning method in reinforcement learning is applied, a typical movement mode of a user in a commercial space is simulated, the track of the user is divided into different activity stages (such as browsing, purchasing and resting), the change of behavior data of the user is continuously monitored through a real-time data feedback mechanism, the behavior trend of the user is captured in real time by adopting a stream data analysis technology, and the parameters of the activity model of the user are adjusted;
Through the incremental learning method, under the condition that the user activity model is not retrained, parameter adjustment is carried out only based on newly added data, so that the user activity model always reflects the latest user behavior characteristics.
It should be noted that, the regularized loss function (such as L2 regularization) is used to calculate the error gradient of the newly added data to the current model, where the formula is:
;
Wherein,For error gradients, α is the dynamic learning rate (adaptive adjustment as the data distribution changes), λ is the regularization coefficient (preventing overfitting),As a gradient of the new data loss function,Is input data;
And parameters of a model bottom layer characteristic extraction layer (a front 3-layer fully-connected network) are fixed, and only the top layer behavior prediction layer is subjected to parameter fine adjustment, so that the basic characteristic expression capability is ensured not to be damaged. And if the confidence coefficient of the prediction of the newly added data is lower than 0.7, triggering local parameter updating, otherwise, retaining the original parameters, and updating the central node according to micro-batch (mini-batch) because the edge node uploads a data fragment every 60 seconds, wherein the single parameter adjustment range is limited by the dynamic learning rate alpha, so that the whole model oscillation is avoided.
Further, converting the user behavior data into a thermodynamic diagram format to distinguish the heat of different time periods;
Capturing a nonlinear relation in user behavior data, identifying behavior characteristics of user activities, mapping the behavior characteristics to a thermodynamic diagram space through a quadratic or cubic polynomial function, dividing a user hot spot area into high, medium and low heat areas, wherein each heat area corresponds to different colors, and planning a detour path to a low-density waiting area for a high-consumption-level user preferentially.
And integrating a simple user feedback investigation function, combining user feedback with thermodynamic diagram data, calculating the influence of user satisfaction on the thermodynamic diagram data by using a simple weighting model, and dynamically updating the thermodynamic diagram according to the result.
And S4, integrating the payment characteristic data and the user behavior data to update the user tag, and generating a cross-region association strategy.
Further, generating a behavior sequence containing a time stamp, coordinates and stay time through the acquired user movement track, aligning the payment time with stay events in the user track, and binding the same user entity;
establishing a multidimensional fusion index, associating user behaviors and consumption characteristics into unified data objects, and marking the unified data objects as potential associated behaviors when a user stays in any functional area for a long time but does not consume the data objects and generates guest prices higher than the average value in the remaining area;
Constructing a multistage label system, wherein the multistage label system comprises a basic attribute layer, a behavior derivative layer and a deep interaction type, and dynamically correcting label weights by adopting an incremental learning model, namely reducing the association confidence coefficient of a current label and a class when the behavior of a label user deviates from a historical mode;
when the user leaves the current area, the optimal association area is matched according to the label, wherein the adjacent area with the promotion activity is preferentially pushed to the 'price sensitive' user for navigation;
When any functional area gathers label users, automatically triggering a cooperative strategy, adjusting commodity display density of adjacent areas, and highlighting related products;
And a module for deploying behavior feedback analysis, wherein the user behavior change after the policy execution is used as an optimization signal, and when any associated policy does not reach an expected target, the current policy priority is automatically degraded.
Embodiment 2a second embodiment of the invention, which differs from the previous embodiment, is:
The functions, if implemented in the form of software functional units and sold or used as a stand-alone product, may be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention may be embodied essentially or in a part contributing to the prior art or in a part of the technical solution, in the form of a software product stored in a storage medium, comprising several instructions for causing a computer device (which may be a personal computer, a server, a network device, etc.) to perform all or part of the steps of the method according to the embodiments of the present invention. The storage medium includes a U disk, a removable hard disk, a Read-Only Memory (ROM), a random access Memory (RAM, random Access Memory), a magnetic disk, an optical disk, or other various media capable of storing program codes.
Logic and/or steps represented in the flowcharts or otherwise described herein, e.g., a ordered listing of executable instructions for implementing logical functions, can be embodied in any computer-readable medium for use by or in connection with an instruction execution system, apparatus, or device, such as a computer-based system, processor-containing system, or other system that can fetch the instructions from the instruction execution system, apparatus, or device and execute the instructions. For the purposes of this description, a "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device.
More specific examples (a non-exhaustive list) of the computer-readable medium include an electrical connection (an electronic device) having one or more wires, a portable computer diskette (a magnetic device), a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). In addition, the computer readable medium may even be paper or other suitable medium on which the program is printed, as the program may be electronically captured, via, for instance, optical scanning of the paper or other medium, then compiled, interpreted or otherwise processed in a suitable manner, if necessary, and then stored in a computer memory.
It is to be understood that portions of the present invention may be implemented in hardware, software, firmware, or a combination thereof. In the above-described embodiments, the various steps or methods may be implemented in software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, may be implemented using any one or a combination of techniques known in the art, discrete logic circuits with logic gates for implementing logic functions on data signals, application specific integrated circuits with appropriate combinational logic gates, programmable Gate Arrays (PGAs), field Programmable Gate Arrays (FPGAs), and the like.
Embodiment 3, referring to fig. 2, provides a system for analyzing consumer behavior of users in different scenes in a business space, which comprises a data acquisition module, a data processing module, a user behavior analysis module, a dynamic thermodynamic diagram generation module and a strategy generation and execution module;
The data acquisition module is used for acquiring behavior data of a user in real time through the wireless positioning device and the distributed environment sensor, and is responsible for receiving equipment signals carried by the user, environment sensor data and data of the RFID reader, and recording the moving track, interaction frequency and related attribute information of the user in a commercial space;
The data processing module performs space-time alignment on the collected multi-source heterogeneous data, performs data cleaning and conversion, and constructs a structured data set for analysis;
the user behavior analysis module is used for extracting resident characteristics and a movement mode of a user based on the cleaned user behavior data and optimizing a user activity model by using a self-adaptive learning mechanism;
the dynamic thermodynamic diagram generation module is used for converting the processed user behavior data into a thermodynamic diagram format, generating user activity thermodynamic diagrams in different time periods and displaying the user focused areas for merchants;
and the strategy generation and execution module is used for generating targeted business strategies based on analysis results, including cross-regional association strategies, and adjusting commodity display and user guidance in real time.
It should be noted that the above embodiments are only for illustrating the technical solution of the present invention and not for limiting the same, and although the present invention has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that modifications and equivalents may be made thereto without departing from the spirit and scope of the technical solution of the present invention, which is intended to be covered by the scope of the claims of the present invention.