Design of enterprise human resources decision support system based on data mining

The management and development of human resources have become one of the core contents of enterprise management. It is from the initial use as an auxiliary tool in human resources business management to the later development of a relatively independent and functional sub-system for the processing of personnel matters between modules, and then to achieve the integration of various sub-systems to form a relatively complete set of human resources information management system. In recent years, with the rapid development of data mining technology, more and more enterprises have begun to attach importance to the assistance of computer information systems to manage the completion of various decision-making activities in operation. This paper designs an enterprise human resource decision support system based on data mining to more effectively support human resource management business information, data standardization, intelligent decision analysis, optimize human resource business processes, improve human resource management business efficiency, provide more effective decision support for enterprise operation and promote enterprise development. Results of the study show the effectiveness of the research.


Introduction
In enterprise management, human resources are known as human capital and it is the core resource of enterprise strategic management. It has become more and more important due to the demands of market and business industry. The competition among enterprises is reflected in the competition for human resources. With the development of science and technology, human resources have become an important resource for enterprises to compete with each other (Wang Communication 2016). Many enterprises can quickly grasp the dynamics of talents because of their advanced human resources management system which saves a lot of time and capital investment for enterprises to select and hire talents. On the whole, the research and development of the human resources decision-making system at home and abroad is less, and the start time is late, so great efforts need to be made to improve the human resources decision-making system (Ouyang and Zhang 2009). The human resources decisionmaking system can be operated from the moment when employees are recruited and conduct a systematic and comprehensive investigation and analysis on employees' pre-job training, skills, salary, and other issues, which can help the company to better manage employees. In addition, the system can avoid incompatibility due to the discrete data stored in the system, make the human resource management move toward standardization, digitization, and networking faster, and improve the overall image of the company (Li 2017). Human resources decision system based on human resources data mining provides decision support for enterprise development from basic analysis, professional analysis, multi-level and multi-dimensional analysis of efficiency and effectiveness, which will be of great significance, which is embodied in the following three aspects (Huang 2019). The first one enterprise management can more scientifically and effectively obtain the internal human resources status of the enterprise, valuable operation reference data, and human resources indicators of external industries in real time, and analyze them in combination with the business decision-making model, to obtain effective decision-making early warning and reference information and make business decisions. The second one is the human resource managers can more accurately and effectively comprehensively manage the enterprise employee information, master the enterprise employee information status in real time, and can quickly and accurately count and generate all kinds of human resource decision-making and analysis reports, to further improve and optimize the enterprise's internal human resource management. The third one is through the collection, data filtering, mining, and analysis of industry human resources data, the human resources index reference library of industry benchmark enterprises is constructed as the reference standard for enterprises to make decisions. The HR data center is composed of internal human resource business data and external related data. The data mining and analysis provide fast and correct reference support for enterprise decision-making, which promotes the continuous improvement of human resources business, forms a good closed-loop, and then promotes the development of enterprises (Zhu 2019). The proposed research work has designed an enterprise human resource decision support system based on data mining to further effectively support human resource management business information, data standardization, intelligent decision analysis, optimize human resource business processes, improve human resource business efficiency, provide more effective decision support for enterprise operation, and promote enterprise development.
The paper is organized with the passion of: Sect. 2 discusses the data mining technologies with the overview of the data mining and the classification algorithms of the data mining. Section 3 presents the detail of the design of enterprise human resources decision support system based on data mining. The simulation test is given in Sect. 4, and the paper is concluded in Sect. 5.

Data mining technology
The following subsections briefly discuss the data mining technologies.

Overview of data mining
Data mining is a new technology rising from 1990s for decision support systems. People usually regard data mining technology as the most important step in the process of knowledge discovery in a database. The so-called data mining is a nontrivial process to obtain people's unknown, novel, effective, potentially useful, and ultimately understandable information and knowledge patterns from a large number of incomplete, noisy, and fuzzy random data. Its main tasks are association analysis, cluster analysis, classification, prediction, time series pattern, and deviation analysis. In short, data mining is to extract or ''mine'' knowledge from a large amount of data, but not all information discovery tasks are data mining (Zhu 2017). Data mining is divided into three processes: data preparation, data mining, conclusion expression, and interpretation, as shown in Fig. 1.

Several common classification algorithms of data mining
The construction methods of classifiers mainly include the following: decision tree classification method, mathematical statistics method, artificial intelligence, neural network method, and so on. According to the different research directions of classification algorithms, they can be divided into the following categories: decision tree classification algorithm, association rules, Bayesian classification, neural network, k-nearest neighbor algorithm, genetic algorithm, rough set, and so on. Decision trees and association rules are typically artificial intelligence methods. The decision tree classification algorithm is a widely used data mining classification algorithm, including ID3 and C4.5 algorithms (Cheng 2017). For a given data set, how to generate a classifier quickly and effectively is a big problem, and one of the more feasible is the decision tree algorithm. The decision tree method is widely used (Lu 2019). It mainly analyzes and infers classification rules from a group of unordered and irregular cases, and uses the way of the decision tree as the form of expression. The main idea of the decision tree classification algorithm is to compare attribute values recursively from top to bottom at the internal nodes of the tree. The specific trend mainly depends on different attribute values, and finally, get a conclusion at the leaf nodes (Mach 2019). The decision tree generation algorithm uses information gain to select the attributes that can make the samples best classified. The information gain is calculated as follows: If there are n messages and the probability distribution is p = {p 1 , p 2 ,…, p n }, the expected information of the sample S is: For a given sample s i [ S, the total number of samples is S i , S i is divided into m subsets according to the class attribute value. The number of samples in each subset is s i (1 B j B m), and its probability distribution is According to the formula, the expected information of sample s i is I(s i ) = I(p).

Association rules
Association rules are the most common methods of data mining. The goal of association rules is to find some internal relationship between all objects. The basic methods of association rules mining can be described by the following methods: Let the database D = {t 1 , t 2 , …, t n } be composed of some transactions with only unique identification, I = {i 1 , i 2 , …, i m } be a set of items, and each transaction t i (i = 1, 2, …, n) corresponds to each subset on I.
Definition 1 Let I 1 I I [ I, the support value of I on data set D refers to the proportion of items containing I in D, as shown in the following formula: Definition 2 For the set of items I and the set D of the database, the non-empty subset of I in T that is not smaller than Minsuppport, i.e., all the sets of items that meet the minimum support that one presupposes, are called frequent item sets or large item sets. In the frequent item set, we abstract out all frequent item sets that are not contained by other items, and these particular item sets are called maximal frequent item sets or we call them maximal large item sets.

Bayesian classification
Bayesian classification is based on Bayesian theorem, where H is a certain assumption, P(H) is a priori probability, and P(X|H) is the probability of X under the condition that H is established, and then the posterior probability P(H|X) is: The principle of Bayesian classification is to calculate the posterior probability through the prior probability of an object and Bayesian theorem, that is, the probability that the object belongs to a certain class, and select the class with the maximum posterior probability as the class to which the object belongs.
3 Design of enterprise human resources decision support system based on data mining

Demand analysis of human resource decision system and business requirements
In business decision-making, human resources are the main strategic resources of enterprises. Human resources management plays an important role in enterprise strategy and decision support. In operation, the enterprise supports the work of each business department through the effective management of human resources and provides strategic decision support for realizing the overall business objectives of the enterprise (Ma and Zhang 2018;Zhan and Han 2015). As shown in Fig. 2, human resource strategy and decision-making are an important part of enterprise business strategy and decision-making. Human resource decision analysis, as the main support of human resource strategy, the requirements of enterprise decision-making on human resources, in turn, constantly adjust and optimize the development of human resource management. Human resources decision analysis carries out in-depth mining and analysis from the perspectives of talent planning, talent basic data (Li 2013), professional management of human resources, and efficiency and effectiveness of human resources, to provide support for enterprise strategy and decision-making.

HR business management function requirements
The data generated by basic human resource business management, such as enterprise organization structure information, personnel basic information, organization information, position information, education information, and personnel change information, is an important part of enterprise human resource data (Xie 2021;Sheng 2018).
To obtain this data, it is necessary to informatize, process, and standardize management. Therefore, the realization of the human resource business management function module is the basic requirement of the system. Figure 3 shows the use case diagram of HR organization structure management. HR users have use cases for organization unit management, job management, position management, job-level management, cost center management, salary range management, and location management. Each organization management use case includes functions such as adding, querying, updating, and deleting.

HR decision tree function requirements
By mining and analyzing human resource data, a human resource decision tree model is generated. Under the condition of setting conditions and data reference basis, the decision management user can make decision analysis and calculation by using the decision tree classification algorithm and show the internal correlation and logical law of human resource data in the form of the decision tree (Pan 2019;Liu 2019). Through the decision tree, it can provide more scientific and efficient decision support for enterprises, such as human resources decision-making, market decision-making, product decision-making, and internationalization decision-making. The realization requirements of enterprise human resources decision support are shown in Fig. 4. The enterprise sets decision analysis requirements, sets enterprise decision direction and objectives, calculates the decision tree through systematic comprehensive decision analysis, outputs human resources decision tree, displays early warning information and enterprise decision support information, and provides effective support for enterprise business decision-making (Wu 2020).

Human resources Web data dynamic mining function requirements
Web data dynamic mining requires dynamically and effectively collect industry human resources-related data and benchmark industry human resources index data through web data mining technology, screen, classify and store the data, and establish an enterprise decision-making reference human resources index database. Through web data mining technology, collect industry human resources-related data of various industries and various analysis dimensions from the Internet, such as the key index data regularly or irregularly published on the Internet by some benchmark listed enterprises, and then filter and filter these data by referring to the defined decision dictionary and semantic rules to obtain effective decision reference data. These data are stored in the database to form the knowledge reference base of human resources decision-making indicators (Xu 2017). Figure 5 shows the web data mining process of industrial human resources.

Overall system design
From the perspective of demand analysis and practical application, the implementation function modules mainly include the design of functional modules such as internal human resources business data management, external data import, industry human resources benchmarking data maintenance, decision analysis report model, human resources decision tree model, and web data mining analysis.

System technical framework design
3.3.1.1 Software architecture Figure 6 shows the software architecture of the enterprise human resources decision-making system. The system constructs a human resources data center by maintaining and managing the internal human resources business management data and status data, and importing other system data, such as enterprise sales, generation, and other data, using online analytical processing, data analysis, and mining, and other technologies (Wei and Wang 2018) re-extract the data used for decision-making, conduct comprehensive analysis through the corresponding decision theoretical model and decision tree algorithm, and obtain valuable data through web data mining and analysis, to provide effective support for decision managers.
The software architecture of the human resources decision system is mainly composed of data source, data dictionary, data warehouse, decision theoretical model, industry data standard, decision analysis report, decision tree algorithm model, and decision-maker user management interface . From the standard input of data, combined with industry standards and the decision demand model, after calculation through relevant algorithms, the decision report analysis results, early warning, and decision information are presented to users through the decision-maker user management interface. The specific functions of each component are described as follows.
Decision data source: Provides internal and external analysis data of the enterprise, mainly from internal human resources business management data, and other ERP systems, such as relevant data of finance, sales and production systems, web mining data, and system interface input. This part needs to build the human resource business management module, realize the integration interface with other external systems, the direct input interface of external data users, and the web data mining function, continuously accumulate all kinds of relevant data into human resource data, and improve the scientificity, effectiveness, and timeliness of data mining analysis.
Decision data dictionary: Data type standard library, which establishes data items, data structure, and logical standards to meet the needs of user report statistics and Design of enterprise human resources decision support system based on data mining 10575 analysis, to make the decision results more scientific and effective. Decision data warehouse: Stores industry-standard reference data after screening and standard specification. The data warehouse is a collection of all types of data that provide support in the decision-making process of enterprises.
Decision report model: The collection of all report models created according to the actual business requirements defines the enterprise decision information requirements from multiple levels and dimensions and constructs the decision analysis report model. Decision theoretical model: To achieve the enterprise objectives, the theoretical models required for prediction, early warning, judgment, and decision-making provided by human resources management are required (Yu 2011). Combined with the report model and the input data, after the set decision inference conditions are met, the system outputs decision early warning information.
Industry decision-making standards: We should collect a large number of standard reference values of various industries in some periods or reference values of industry benchmarks to obtain comparable and more reasonable analysis reference information, to make enterprise decision-making more targeted.
Decision-maker interface: Management desktop and cockpit of enterprise managers. Real-time display of various reports, analysis results, early warning information, decision information, and decision support for managers.

System network topology
The network topology of the enterprise human resources decision system is shown in Fig. 7. The overall system needs four servers, namely HR business server, decision analysis server, web data mining server, and data storage server. The HR business server mainly processes HR business operations. The decision analysis server mainly completes the mining and analysis of human resource data and generates decision analysis reports, and decision trees. The web data mining server mainly completes the mining of Internet data and stores the mined data to the database server; the database server is used to store all human resources business data, other relevant decision-making data imported by ERP system, external input data, and human resources index reference data of relevant industries. In addition, the system adopts a B / S structure, which has the characteristics of distribution and strong interaction, to realize the realtime update and sharing of human resources business data. Decision-makers and users can query information anytime and anywhere, obtain the latest data in real-time, and improve the timeliness of decision-making information.

System function structure design
The design of enterprise human resources decision system is mainly divided into four functional modules; human resources business management function module, human resources data management function module, decision analysis report management function module, human resources decision tree model, and web data mining analysis module. Each functional module contains several subfunctional modules. The overall functional architecture design of the system is shown in Fig. 8.

HR business management module
This part mainly includes three resource modules: enterprise organization structure management, personnel information management, and personnel event management, to realize the daily business management of enterprise human resources and the update and maintenance of basic human resources data. Human resources business data is the main data source basis for decision analysis and the main component of human resources data for decision analysis, Master the internal status data of the enterprise in real-time.

HR data management module
This part mainly realizes the preprocessing of internal HR status data, and further extracts and filters various personnel status data needed in decision analysis based on HR daily business data. This part realizes the functions of direct data generation, interface maintenance, and import.

Decision analysis report module
This part mainly realizes the decision-making analysis function of internal human resource data of the enterprise, and statistically analyzes the human resource status of the enterprise. It is divided into three types of analysis reports: basic decision-making analysis, professional decisionmaking analysis, efficiency, and efficiency decision-making analysis.

Simulation tests
The information of some employees in the enterprise unit was input into the HR system for the usage analysis. Ten groups of these employees were selected as research objects, and the clustering method and the scoring method were used to classify the assessment, respectively, and then the results were compared and analyzed to draw several Objects R1-R10 were selected as needed for the study, and the real names were omitted here. The seven aspects of these ten objects were scored separately and then clustered using the K-means method. The final classification results were classified into four categories: excellent, competent, basic competent, and incompetent. The scoring principle is based on the expert scoring method, which means that a group of experts is selected in the unit to score the employees, with each item scoring 10 out of 10 and the lowest score being 0. The scoring results are shown in Table 1.
The K-means clustering algorithm was used to cluster ten objects, treating the data as a total of ten groups of seven-dimensional spatial data, according to the distance nearest neighbor principle. What is needed is only the clustering result, thus the clustering center has little influence on the result. After the clustering algorithm operation, the clustering results are obtained as shown in Table 2.
If the traditional classification method is followed, four grades are assigned according to the level of the score, which is referred to here as the scoring method, where the full score is 63, 45 including 45 is excellent, 40-45 is competent, 35-40 is basic competent and below 35 is incompetent, and the resulting classification results are shown in Table 3.
From the results of the two methods of classification, there is a large difference, only the third basic competence one classification result is consistent, the other three classification results are not the same. The outstanding individuals are four, and the outstanding ones in the clustering method are R1, R2, R3, R8; the outstanding ones in the scoring method are R1, R2, R8, R9; the common objects of both are R1, R2, R8; the objects of difference are R3 and R9; from the raw data, the score of R9 is 46, while the score of R3 is 42; on the surface, it seems that R9 is more outstanding, but analyzing the data, we can see the following phenomenon which is shown in Fig. 9.
The total score of R9 is higher than that of R3, but the difference between the scores of R9 and R3 is greater, including a score of 9 for innovation and 3 for communication and coordination. The clustering method can classify the nearest individuals into one category according to the spatial location of the individual score data, which can take into account the balance and comprehensiveness of individual development and is more convincing.
The use of data mining methods for HR decision support systems is a major departure from traditional methods. The system completely breaks the traditional appraisal model of Fig. 8 Overall function structure of the system relying on manpower to conduct assessments and enter and output results. Staff only needs to input indicator data scores to produce individual data results for the employee being assessed automatically by the system, which increases objectivity and saves time and costs.

Conclusion
This paper mainly expounds on the design of enterprise human resources decision-making systems based on data mining. Firstly, the complete requirements of the system are studied and analyzed in detail, and then the overall functional architecture of the system is designed on this basis. It realizes the daily business of human resources, such as organizational structure management. Through the standardized management of human resources business processes, it establishes human resources standard data center and sharing center to provide basic data guarantee   Design of enterprise human resources decision support system based on data mining 10579 for data analysis. Further, extract the data needed in the decision analysis operation, and the system realizes the maintenance function of the analysis data. By designing three types of reports based on the human resources decision cockpit, you can analyze the current situation of internal human resources from different dimensions, understand the current situation and make more effective decisions.
Funding The paper did not receive any financial support.
Data Availability Enquiries about data availability should be directed to the authors.

Declarations
Conflict of interest The authors declared that they have no conflicts of interest to this work.
Ethical approval The paper does not deal with any ethical problems.
Informed consent We declare that all authors have informed consent.