Disclosure of Invention
The invention aims to provide an enterprise digital operation management system based on artificial intelligence so as to solve the problems in the background technology.
In order to achieve the aim, the invention provides the technical scheme that the enterprise digital operation management system based on artificial intelligence comprises a data integration module, a behavior prediction module, a feature recommendation module, a cluster subdivision module and a decision support module;
The data integration module is used for collecting and arranging the original data and transmitting the arranged original data to the behavior prediction module and the characteristic recommendation module;
the behavior prediction module is used for constructing a prediction model according to an artificial intelligent algorithm, analyzing the client behavior and the purchase mode and predicting the client demands and the business intentions;
the feature recommendation module is used for providing service recommendation according to the client features according to the collaborative filtering algorithm;
the cluster subdivision module is used for subdividing the customer groups according to a cluster analysis algorithm;
the decision support module is used for grading the service grade according to the type of the customer group.
Preferably, the method for constructing the prediction model according to the artificial intelligence algorithm comprises the following steps:
Collecting client information, interaction data, sales data and use data, removing repeated data, processing missing values and abnormal values, normalizing or standardizing the collected data, wherein the client information comprises employee number Nu and tax payment amount Ta, the interaction data comprises E-mail interaction number En, telephone record number Tn and conference record number Mn, the sales data comprises purchase times Pn, contract amount Ac and purchase period Bc, and the use data comprises product use times Uc and fault feedback times Fn;
according to the collected data extraction characteristics, dividing the data into a training set and a testing set, setting initial super parameters by using a random forest algorithm, and training a random forest model by using the training set;
evaluating the performance of the model by using a test set, calculating the F1 score of the random forest model, and checking the stability of the random forest model through cross verification;
Performing super parameter tuning, and adjusting the number of trees and the maximum depth parameter;
And deploying the optimized random forest model, and predicting the purchase intention of the customer.
Preferably, in extracting the features according to the collected data, the feature selection method includes:
The method is characterized by comprising the steps of scaling a weight coefficient, an interaction coefficient, a consumption trend coefficient, calibrating the scale weight coefficient to Sc, the interaction coefficient to In, the consumption trend coefficient to Ct, calibrating the feedback coefficient to Tf, calculating the scale weight coefficient to Sc=Nu+Ta, calculating the interaction coefficient to In=Mn (En+Tn), and calculating the consumption trend coefficient to CtThe computational expression using feedback coefficients is
Preferably, the method for recommending the service by using the collaborative filtering algorithm comprises the following steps:
Collecting behavior data of scoring, browsing times and purchasing times of the service by a user;
performing service characteristic evaluation according to the behavior data of the service by the user;
Based on the service feature evaluation, calculating the similarity between users by using a similarity algorithm;
For each user, finding a plurality of similar users with highest similarity with the users;
recommending the service selected by the similar user for the target user according to the behavior of the similar user;
The definition logic of the service characteristic evaluation is as follows:
service feature evaluation=score× (number of purchases/number of browses).
Preferably, based on the service feature evaluation, the logic for calculating the similarity between the users by using the similarity algorithm is as follows:
Calibrating s to be the number of the service, and s= {1,2, 3..n }, wherein n is a positive integer, calculating the pearson correlation coefficient between users, and calculating the expression to beWherein An、Bn is the service characteristic evaluation of the user A and the user B to the service n,The service characteristics for all purchased services are evaluated for user a as a mean,And evaluating the mean value for the service characteristics of all purchased services for the user B.
Preferably, the cluster subdivision module is used for subdividing the customer group according to a cluster analysis algorithm, and the method comprises the following steps of;
The method comprises the steps of collecting employee number, annual purchase amount, consumption age, income and historical consumption amount of clients, determining a K value according to a K-means algorithm through an elbow method or a contour coefficient method, analyzing characteristics of each cluster according to a cluster analysis result, and classifying client types.
Preferably, the logic for analyzing the result of the cluster analysis is:
When the K-means cluster K value is 2, for B2B business clients, if the staff number and annual purchasing quantity of the clients in the cluster 1 are larger than those in the cluster 2, the clients in the cluster 1 are marked as large clients, the clients in the cluster 2 are marked as medium and small clients, and for B2C business clients, if the historical consumption limit and income of the clients in the cluster 1 are larger than those in the cluster 2 and the consumption age of the clients in the cluster 1 is smaller than those in the cluster 2, the clients in the cluster 1 are marked as high-income clients, and the clients in the cluster 2 are marked as medium-income clients.
Preferably, the decision support module is used for grading the service grade according to the customer group type as logic;
When the B2B business client is marked as a large client, setting the service level as V1 level, and providing customized solutions and bulk purchasing offers;
When B2B business clients are marked as small and medium-sized clients, setting the service level as V2 level, and providing flexible purchasing plans and supporting services;
When the B2C business clients are marked as high-income clients, setting the service level as S1 level, and providing high-end fashion and technological product recommendation and preferential activities;
When the B2C service client is marked as a medium income client, the service level is set to S2 level, and discount promotion of daily necessities and home service are provided.
In the technical scheme, the invention has the technical effects and advantages that:
According to the application, the artificial intelligence technology is used, and the digitized operation of an enterprise is optimized and expanded by combining with a customer relationship management system, so that the transfer development of the enterprise in the B2B service field and the B2C service field is satisfied, the situation that the enterprise uses two customer relationship management systems of different service tracks simultaneously to cause data island and information isolation barriers is avoided, the clients of different service types are analyzed and predicted based on the artificial intelligence technology, the stability of the relation of the clients is ensured, the cooperation state of the clients of different types is accurately controlled, and the energy cost of management staff is reduced.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. 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.
Referring to FIG. 1, the invention relates to an enterprise digital operation management system based on artificial intelligence, which comprises a data integration module, a behavior prediction module, a feature recommendation module, a cluster subdivision module and a decision support module;
The data integration module is used for collecting and arranging the original data and transmitting the arranged original data to the behavior prediction module and the characteristic recommendation module;
the behavior prediction module is used for constructing a prediction model according to an artificial intelligent algorithm, analyzing the client behavior and the purchase mode and predicting the client demands and the business intentions;
the feature recommendation module is used for providing service recommendation according to the client features according to the collaborative filtering algorithm;
the cluster subdivision module is used for subdividing the customer groups according to a cluster analysis algorithm;
the decision support module is used for grading the service grade according to the type of the customer group.
In conducting enterprise digitized operation management, enterprises have different requirements and conditions for customer relationship management of B2B services and B2C services, for B2B services, customers are typically companies or organizations, data sources including sales records, contracts, customer access records, industry reports, enterprise public data, etc., data types including enterprise scale, industry classifications, financial status, purchasing history, business requirements, etc., while for B2C services, customers are individual consumers, data sources including purchase records, browsing records, social media interactions, customer feedback, mobile application usage data, etc., data types including demographic information, purchase history, interest preferences, consumption behavior, etc., demographic information typically including age, gender, income level, etc., analysis targets for conducting B2B sales and conducting B2C sales are also different for B2B services, the layering logic of the analysis targets is to layer clients according to factors such as enterprise scale, purchasing quantity, cooperation time and the like, identify key clients and potential clients, evaluate the total value of the clients in the whole cooperation period, formulate a targeted client maintenance strategy, focus on value analysis on maintaining client relationship, track and optimize each stage from the potential clients to final ordering by adopting a sales funnel analysis mode, improve sales conversion rate, predict purchasing behavior, demand change, offer possibility and the like of the clients, and in B2C business, the subdivision logic of the analysis targets is to conduct client subdivision according to behavior, preference and demographic characteristics of consumers so as to conduct individuation, the aim of value analysis is to evaluate the value of the clients in the whole life period, formulate a retention and retrieval strategy and focus on analysis of the repeated consumption capability of the clients, in the field of analysis methods, B2B business and B2C business are also different, B2B business is analyzed through association rules, purchasing association among different products or services is identified so as to carry out cross sales or up-sales, business relationship and networks among enterprises are analyzed through relational network analysis so as to find potential cooperation opportunities, trend analysis is utilized to track and analyze industry trend and market dynamics, change of customer demands is predicted, personalized products or services are recommended according to historical behaviors of customers and behaviors of similar users for B2C business, a clustering analysis algorithm is utilized to divide customers into different groups so as to carry out targeted marketing and services, natural language processing technology is utilized to analyze comments and feedback of customers so as to know trends and demands of customers, use time sequence analysis is utilized to track and analyze industry trend and market dynamics, change of customer demands is predicted, and marketing activity time sequence is optimized;
Although big data analysis of B2B and B2C customer relationship management is different in data source, analysis target and analysis method, the final objective is the same, namely, the customer demand is known, the customer satisfaction is improved, sales and business growth are increased, so that the B2B business data and the B2C business data can be integrated and analyzed according to the common target to adapt to the rapidly-changing market environment.
The data integration module is used for collecting and arranging the original data and transmitting the arranged original data to the behavior prediction module and the characteristic recommendation module;
The original data comprises sales data, visitor information, financial information, purchasing history information, browsing data and individual information;
the method comprises the steps of storing original data by using a data lake, cleaning, converting and loading the original data by using a data warehouse, cleaning the data, identifying, correcting or deleting error, repeated and inconsistent data items in the data, converting the data from a source system format into a format applied to the data warehouse by using data conversion, converting data types, remapping fields, aggregating or decomposing the data, and the like, wherein the cleaned and converted data is loaded into the data warehouse, and the loading process involves sorting, indexing and optimizing the data, so that the query performance can be improved.
When the B2B and B2C business is involved, the integration and management of the data are critical to enterprises, the integration and management of the data can ensure that the data of all data sources are accurate, errors and redundant data are reduced, the reliability and effectiveness of business decisions can be improved by high-quality data, the consistency of data formats and structures of different business departments is ensured through data standardization, the problem of inconsistent data islands and information is avoided, deep data analysis can be performed through integration and management of the data, potential business opportunities and market trends are mined, the scientificity and the accuracy of business strategies can be improved by data integration and management, the decision risk is reduced, the data integration provides a basis for business process automation and intellectualization, the operation efficiency can be improved by robot process automation and artificial intelligence technology, human intervention and errors are reduced, the integrated data can help enterprises to manage and optimize resource allocation better, the resource waste and repeated construction are avoided, and the enterprises can optimize inventory management, supply chain operation and marketing strategies through data analysis, and the overall efficiency is improved.
The behavior prediction module is used for constructing a prediction model according to an artificial intelligent algorithm, analyzing the client behavior and the purchase mode and predicting the client demands and the business intentions;
The step of constructing a prediction model by using an artificial intelligence algorithm comprises the following steps:
Collecting client information, interaction data, sales data and use data, removing repeated data, processing missing values and abnormal values, normalizing or standardizing the collected data, wherein the client information comprises employee number Nu and tax payment amount Ta, the interaction data comprises E-mail interaction number En, telephone record number Tn and conference record number Mn, the sales data comprises purchase times Pn, contract amount Ac and purchase period Bc, and the use data comprises product use times Uc and fault feedback times Fn;
according to the collected data extraction characteristics, dividing the data into a training set and a testing set, setting initial super parameters by using a random forest algorithm, and training a random forest model by using the training set;
evaluating the performance of the model by using a test set, calculating the F1 score of the random forest model, and checking the stability of the random forest model through cross verification;
Performing super parameter tuning, and adjusting the number of trees and the maximum depth parameter;
And deploying the optimized random forest model, and predicting the purchase intention of the customer.
The random forest predicts by integrating a plurality of decision trees, each decision tree is trained on different random samples and feature subsets, the final prediction result is a consistency result of all trees, the accuracy and robustness of the model can be effectively improved, and because the random forest predicts by using a plurality of decision trees, the overfitting risk of a single tree is dispersed, even if some decision trees can be overfitted, the overall prediction result can still keep better generalization capability, the random forest can provide importance assessment of each feature, which features can be represented to be most contributed to the prediction result, the random forest can still keep better prediction performance when facing missing values and can be more robust when facing data noise and abnormal values, and is not easily influenced by extreme values by training by using different feature subsets.
In extracting features according to collected data, the feature selection method comprises the following steps:
The method is characterized by comprising the steps of scaling a weight coefficient, an interaction coefficient, a consumption trend coefficient, calibrating the scale weight coefficient to Sc, the interaction coefficient to In, the consumption trend coefficient to Ct, calibrating the feedback coefficient to Tf, calculating the scale weight coefficient to Sc=Nu+Ta, calculating the interaction coefficient to In=Mn (En+Tn), and calculating the consumption trend coefficient to CtThe computational expression using feedback coefficients is
The feature recommendation module is used for providing service recommendation according to the client features according to the collaborative filtering algorithm;
The service recommendation method by using the collaborative filtering algorithm comprises the following steps:
Collecting behavior data of scoring, browsing times and purchasing times of the service by a user;
performing service characteristic evaluation according to the behavior data of the service by the user;
Based on the service feature evaluation, calculating the similarity between users by using a similarity algorithm;
For each user, finding a plurality of similar users with highest similarity with the users;
And recommending the service selected by the similar user for the target user according to the behavior of the similar user.
The similarity algorithm comprises cosine similarity, pearson correlation coefficient and the like;
The definition logic of the service characteristic evaluation is as follows:
service feature evaluation=score× (number of purchases/number of browses);
logic for calculating the similarity between users with pearson correlation coefficients is:
Calibrating s to be the number of the service, and s= {1,2, 3..n }, wherein n is a positive integer, calculating the pearson correlation coefficient between users, and calculating the expression to beWherein An、Bn is the service characteristic evaluation of the user A and the user B to the service n,The service characteristics for all purchased services are evaluated for user a as a mean,And evaluating the mean value for the service characteristics of all purchased services for the user B.
Calculating the similarity between users using pearson correlation coefficients can effectively measure the similarity of scoring behavior of users, thereby providing personalized service recommendations for users in collaborative filtering algorithms.
The cluster subdivision module is used for subdividing the customer groups according to a cluster analysis algorithm;
The method comprises the steps of collecting employee number, annual purchase amount, consumption age, income and historical consumption amount of clients, determining a K value according to a K-means algorithm through an elbow method or a contour coefficient method, analyzing characteristics of each cluster according to a cluster analysis result, and classifying client types.
The decision support module is used for grading the service grade according to the type of the customer group;
When the K-means cluster K value is 2, for B2B business clients, if the staff number and annual purchasing quantity of the clients in the cluster 1 are larger than those in the cluster 2, the clients in the cluster 1 are marked as large clients, and the clients in the cluster 2 are marked as medium and small clients;
When the B2B business client is marked as a large client, setting the service level as V1 level, and providing customized solutions and bulk purchasing offers;
When B2B business clients are marked as small and medium-sized clients, setting the service level as V2 level, and providing flexible purchasing plans and supporting services;
When the B2C business clients are marked as high-income clients, setting the service level as S1 level, and providing high-end fashion and technological product recommendation and preferential activities;
When the B2C service client is marked as a medium income client, the service level is set to S2 level, and discount promotion of daily necessities and home service are provided.
According to the application, the artificial intelligence technology is used, and the digitized operation of an enterprise is optimized and expanded by combining with a customer relationship management system, so that the transfer development of the enterprise in the B2B service field and the B2C service field is satisfied, the situation that the enterprise uses two customer relationship management systems of different service tracks simultaneously to cause data island and information isolation barriers is avoided, the clients of different service types are analyzed and predicted based on the artificial intelligence technology, the stability of the relation of the clients is ensured, the cooperation state of the clients of different types is accurately controlled, and the energy cost of management staff is reduced.
The above formulas are all formulas with dimensions removed and numerical values calculated, the formulas are formulas with a large amount of data collected for software simulation to obtain the latest real situation, and preset parameters in the formulas are set by those skilled in the art according to the actual situation.
The above embodiments may be implemented in whole or in part by software, hardware, firmware, or any other combination. When implemented in software, the above-described embodiments may be implemented in whole or in part in the form of computer program product. The computer program product comprises one or more computer instructions or computer programs. When the computer instructions or computer program are loaded or executed on a computer, the processes or functions described in accordance with embodiments of the present application are produced in whole or in part. The computer may be a general purpose computer, a special purpose computer, a computer network, or other programmable apparatus. The computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another computer-readable storage medium, for example, the computer instructions may be transmitted from one website site, computer, server, or data center to another website site, computer, server, or data center by wired or wireless means (e.g., infrared, wireless, microwave, etc.). The computer readable storage medium may be any available medium that can be accessed by a computer or a data storage device such as a server, data center, etc. that contains one or more sets of available media. The usable medium may be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. The semiconductor medium may be a solid state disk.
It should be understood that, in various embodiments of the present application, the sequence numbers of the foregoing processes do not mean the order of execution, and the order of execution of the processes should be determined by the functions and internal logic thereof, and should not constitute any limitation on the implementation process of the embodiments of the present application.
Those of ordinary skill in the art will appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware, or combinations of computer software and electronic hardware. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the solution. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present application.
It will be clear to those skilled in the art that, for convenience and brevity of description, reference may be made to the corresponding process in the foregoing method embodiment for the specific working process of the above-described system, which is not described herein again.
The functions, if implemented in the form of software functional units and sold or used as stand-alone goods, may be stored in a computer readable storage medium. Based on this understanding, the technical solution of the present application may be embodied essentially or in a part contributing to the prior art or in the form of software goods stored in a storage medium, including several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method according to the embodiments of the present application. The storage medium includes various media capable of storing program codes, such as a usb disk, a removable hard disk, a read-only memory (ROM), a random access memory (RANLMM ACCESS memory), a magnetic disk, or an optical disk.
The foregoing is merely illustrative of the present application, and the present application is not limited thereto, and any person skilled in the art will readily recognize that variations or substitutions are within the scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.