An art industry city employment attraction evaluation method and system

Abstract

The application discloses an art industry city employment attraction evaluation method and system, comprising: obtaining network search data of each city in a target region within a preset time period; extracting a plurality of principal components based on the network search data, and constructing an employment attraction comprehensive evaluation index of each city based on the plurality of principal components; obtaining an evolution trend of the employment attraction in a time dimension based on the employment attraction comprehensive evaluation index of each city; performing grade division on a plurality of cities in the target region by using a clustering analysis method based on the plurality of principal components; and performing GIS visualization processing on the employment attraction comprehensive evaluation index of each city, the evolution trend of the employment attraction in the time dimension, or a city employment attraction grade classification result, to generate a thematic map reflecting spatial distribution characteristics of the employment attraction. The application can effectively solve technical shortcomings in regional employment demand evaluation and dynamic prediction, and provide a more scientific decision basis for macro employment market research and judgment.

Description

Technical Field

[0001] This invention relates to the technical field of urban industrial attraction assessment, and in particular to a method and system for evaluating urban employment attraction in the arts industry. Background Technology

[0002] In recent years, with the increasing emphasis placed on art-related majors and the adjustment of the academic discipline catalog, how to cultivate art talents that meet social needs and enhance the attractiveness of the industry for employment has become an important issue. However, current research on employment education mainly focuses on curriculum design and practical teaching, while systematic research on urban employment activity is relatively weak. Existing theoretical methods are insufficient to fully depict the spatial characteristics and dynamic evolution of the art industry's employment market.

[0003] The development of internet technology has made online search data an important source of information reflecting the employment market situation. However, research on constructing employment activity models for urban arts industries based on online search indices is still lacking in China. Therefore, this study utilizes data such as Baidu search indexes to establish an employment demand index model, and combines it with GIS visualization analysis and grey prediction models. This approach can effectively detect the spatiotemporal differentiation characteristics of employment attraction, providing quantifiable intelligence support for industry decision-making.

[0004] The existing employment statistics framework has limitations in reflecting regional development disparities and supply-demand dynamics, especially in the arts sector, lacking a systematic approach to regional imbalances, long-term forecasting capabilities, and a multi-dimensional indicator system. Therefore, there is an urgent need to establish a comprehensive employment attraction evaluation indicator system, integrating methods such as principal component clustering and time-series forecasting, to address current technical shortcomings in evaluating and dynamically forecasting regional employment demand, and to provide a more scientific basis for macro-level employment market analysis. Summary of the Invention

[0005] In order to overcome the shortcomings of existing technologies, the purpose of this invention is to propose a method and system for evaluating the urban employment attraction of the arts industry, which can effectively make up for the current technical shortcomings in the evaluation and dynamic prediction of regional employment demand, and provide a more scientific basis for decision-making in the analysis of the macro employment market.

[0006] To achieve the objectives of this invention, the following technical solution is adopted: One objective of this invention is to provide a method for evaluating the urban employment attraction of the arts industry, the method comprising the following steps: Based on preset art industry keywords, obtain online search data of each city in the target area within a preset time period to form a raw dataset, and preprocess the raw dataset. Based on the preprocessed original dataset, multiple principal components are extracted using principal component analysis, and a comprehensive evaluation index of employment attraction for each city is constructed based on these multiple principal components. Based on the time-series data of the comprehensive evaluation index of employment attraction in each city, a dynamic time-series prediction model of employment attraction index is constructed using the residual correction grey model, and the employment attraction index for future time periods is predicted to obtain the evolution trend of employment attraction in the time dimension. Based on the cross-sectional data of the multiple principal components in the spatial dimension, cluster analysis is used to classify multiple cities in the target area into different levels, so as to obtain the classification results of urban employment attraction levels and to obtain the distribution pattern of employment attraction in the spatial dimension. The comprehensive evaluation index of employment attraction of each city, the evolution trend of employment attraction over time, or the classification results of employment attraction level of each city are processed by GIS visualization to generate a thematic map reflecting the spatial distribution characteristics of employment attraction.

[0007] In the above technical solution, by pre-setting keywords related to the arts industry and acquiring massive amounts of online search data using a web search index platform, a comprehensive evaluation index of employment attraction is constructed based on principal component analysis to extract multiple principal components. This achieves a quantitative assessment of the employment attraction of the urban arts industry, supplementing and enriching the existing employment statistics systems of governments and universities. Preprocessing the acquired raw dataset effectively improves its practicality and reliability. A dynamic time-series prediction model of the employment attraction index is constructed using a residual-corrected grey model, enabling prediction of the employment attraction index for future periods and providing insights into changes in the supply and demand of the arts industry employment market. Dynamic monitoring provides a basis for decision-making regarding macro-level employment market trends; cluster analysis is used to classify the employment attraction of multiple cities within a target region, which can intuitively reveal the spatial distribution pattern of employment attraction among cities, effectively identify cities with different states such as strong, stable, and insufficient employment demand, and provide a scientific basis for formulating employment policies according to local conditions; the evaluation index, predicted trend, or classification results are visualized using GIS to generate thematic maps reflecting the spatiotemporal distribution characteristics of employment attraction, transforming simple employment data into practical intelligence, simplifying abstract data mining results, enhancing data classification understanding, and making it more conducive to government decision-making research.

[0008] Furthermore, the preprocessing of the original dataset includes: The original dataset is subjected to noise reduction and data cleaning, wherein: Delete useless tags, special characters, and irrelevant punctuation marks; For Chinese text segmentation, continuous text strings are divided into individual words; Stem extraction and semantic restoration of English or other text data with word form changes.

[0009] Furthermore, principal component analysis is used to extract multiple principal components, and a comprehensive evaluation index of employment attraction for each city is constructed based on these principal components, specifically including: From the preprocessed original dataset, evaluation index data related to employment attraction in the arts industry are extracted in multiple dimensions. Principal component analysis is used to reduce the dimensionality of the evaluation index data in multiple dimensions and extract multiple principal components. The multiple dimensions include fine arts and art design, music performance and dance, and directing and opera. Determine the score of each principal component and its corresponding variance contribution rate; The comprehensive evaluation index of employment attraction for each city is obtained by weighting and summing the scores of each principal component and their corresponding variance contribution rates.

[0010] Furthermore, a dynamic time-series prediction model for the employment gravity index is constructed using a residual-corrected grey model, specifically including: The original sequence of the comprehensive evaluation index of employment attraction is accumulated to generate the accumulated sequence; Based on the accumulated sequence, a grey differential equation is established and solved to obtain a preliminary prediction model and its fitted sequence; Calculate the residual sequence between the original sequence and the fitted sequence, and establish a residual prediction model based on the residual sequence; The preliminary prediction model is modified using the residual prediction model to obtain the residual modified grey model, which serves as the dynamic time-series prediction model for the employment gravity index.

[0011] Furthermore, the step of refining the preliminary prediction model using the residual prediction model specifically includes: Determine the starting point for residual sequence modeling; For the predicted values ​​after the starting point, the prediction results of the preliminary prediction model and the prediction results of the residual prediction model are superimposed to obtain the corrected predicted values. For the predicted values ​​prior to the starting point, the prediction results of the preliminary prediction model remain unchanged.

[0012] Furthermore, cluster analysis is used to classify multiple cities within the target area into different levels, specifically including: The scores of the multiple principal components are used as the basic data for cluster analysis, wherein each city in the basic data corresponds to a sample point; The distance between sample points in each city is calculated using a distance metric method, and a distance matrix is ​​constructed based on the calculated distance between the sample points in the city. The distance matrix is ​​used to record the similarity between any two sample points. Based on the similarity between any two sample points, the hierarchical clustering method is used to merge the two closest sample points into one cluster, and this process is repeated until all sample points are merged into one class to generate a cluster hierarchy diagram. The number of clusters is determined based on the clustering hierarchy diagram, and the classification results of urban employment attraction levels are obtained.

[0013] Furthermore, the distance metric method includes one or more combinations of Euclidean distance, standardized Euclidean distance, Mahalanobis distance, Block distance, and Minkowski distance, used to eliminate the dimensional influence of the principal component scores.

[0014] Furthermore, the system clustering method includes the centroid method, wherein the centroid position of the merged cluster and its distance from any other cluster are jointly determined by the centroid position of each cluster before merging, the number of sample points, and the inter-cluster distance.

[0015] Furthermore, the comprehensive evaluation index of employment attraction for each city, the evolution trend of employment attraction over time, or the classification results of employment attraction levels for each city are subjected to GIS visualization processing, specifically including: The comprehensive evaluation index of employment attraction of each city, the evolution trend of employment attraction over time, or the classification results of employment attraction level of each city are used as attribute data and associated with the urban geographic information spatial data of the target area. By utilizing the symbolization function of GIS software, differentiated visual identifiers are assigned to geographical elements of different cities based on the magnitude or category of attribute values, generating thematic maps that reflect the spatial distribution characteristics of employment attraction.

[0016] The second objective of this invention is to provide an evaluation system for urban employment attraction in the arts sector, the system comprising: The data processing module is used to obtain online search data of each city in the target area within a preset time period based on preset art industry keywords, form a raw dataset, and preprocess the raw dataset. The data extraction module is used to extract multiple principal components based on the preprocessed original dataset using principal component analysis, and to construct a comprehensive evaluation index of employment attraction for each city based on the multiple principal components. The data prediction module is used to construct a dynamic time series prediction model of the employment attraction index based on the time series data of the comprehensive evaluation index of employment attraction in each city, using a residual-corrected grey model, and to predict the employment attraction index for future time periods in order to obtain the evolution trend of employment attraction in the time dimension. The ranking module is used to classify multiple cities in the target area based on the cross-sectional data of the multiple principal components in the spatial dimension, and to obtain the ranking results of urban employment attraction in order to obtain the distribution pattern of employment attraction in the spatial dimension. The data visualization module is used to perform GIS visualization processing on the comprehensive evaluation index of employment attraction of each city, the evolution trend of employment attraction over time, or the classification results of employment attraction levels of the city, to generate thematic maps reflecting the spatial distribution characteristics of employment attraction.

[0017] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention proposes a method and system for evaluating the employment attraction of the arts industry in cities. By pre-setting keywords related to the arts industry and acquiring massive amounts of online search data using a web search index platform, a comprehensive employment attraction evaluation index is constructed based on principal component analysis to extract multiple principal components. This achieves a quantitative assessment of the employment attraction of the arts industry in cities, supplementing and enriching existing government and university employment statistics systems. Preprocessing the acquired raw dataset effectively improves its practicality and reliability. A dynamic time-series prediction model for the employment attraction index is constructed using a residual-corrected grey model, enabling prediction of the employment attraction index for future time periods, thus realizing the evaluation of the arts industry's employment attraction. Dynamic monitoring of supply and demand changes in the employment market provides a basis for decision-making regarding macroeconomic employment market trends. Using cluster analysis to classify the employment attraction of multiple cities within a target region can intuitively reveal the spatial distribution pattern of employment attraction among cities, effectively identifying cities with different states of employment demand, such as strong, stable, and insufficient demand, thus providing a scientific basis for formulating employment policies tailored to local conditions. Visualizing evaluation indices, predicted trends, or classification results using GIS generates thematic maps reflecting the spatiotemporal distribution characteristics of employment attraction, transforming simple employment data into practical intelligence, simplifying abstract data mining results, enhancing data classification understanding, and further facilitating government decision-making research. Attached Figure Description

[0018] Figure 1 A flowchart illustrating the steps of an evaluation method for urban employment attraction in the arts sector, provided in this application embodiment; Figure 2 A schematic diagram of the structure of an urban employment attraction evaluation system for the arts industry provided in this application embodiment; Figure 3 A schematic diagram illustrating the research approach for the synergistic effect of employment competitiveness and high-quality employment development oriented towards new productivity, provided for embodiments of this application. Figure 4 A schematic diagram of a synergistic effect model of employment competitiveness and employment quality oriented towards new productivity provided in an embodiment of this application; Figure 5 A schematic diagram illustrating the knowledge structure detection and description process for the spatiotemporal dimensions of employment in the arts industry, provided for embodiments of this application; Figure 6 A schematic diagram illustrating the employment status and trends of college art students since 2009, provided for embodiments of this application; Figure 7 A schematic diagram illustrating the fitting and prediction effect of the dynamic time-series model of the employment gravity index of the arts industry in Jiangsu Province in the experiment provided for the embodiments of this application; Figure 8 A schematic diagram of the clustering distribution of employment demand in the arts industry in 13 cities of Jiangsu Province in 2021, based on the principal component-based standardized Euclidean distance centroid method, provided for embodiments of this application. Figure 9 A schematic diagram of the clustering distribution of the first and second principal components of employment gravity space dimension in 13 cities of Jiangsu Province in 2021, provided for the embodiments of this application; Figure 10 A schematic diagram illustrating the distribution of overall activity levels in the arts sector across 13 cities in Jiangsu Province in 2021, provided for embodiments of this application. Figure 11 This is a schematic diagram showing the distribution of the overall activity level of the arts industry in different regions of Jiangsu Province in 2021, provided as an example of the present application. Detailed Implementation

[0019] To facilitate understanding of the present invention, a more complete description will be given below with reference to the accompanying drawings. Preferred embodiments of the invention are shown in the drawings. However, the invention can be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided to provide a thorough and complete understanding of the disclosure of the invention.

[0020] Unless otherwise defined, 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 pertains. 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 term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0021] Example 1: This embodiment provides a method for evaluating the urban employment attraction of the arts industry. (See also...) Figure 1 The method includes the following steps: Step S1: Based on preset art industry keywords, obtain online search data of each city in the target area within a preset time period to form an original dataset, and preprocess the original dataset; Step S2: Based on the preprocessed original dataset, extract multiple principal components using principal component analysis, and construct a comprehensive evaluation index of employment attraction for each city based on the multiple principal components; Step S3: Based on the time-series data of the comprehensive evaluation index of employment attraction in each city, a dynamic time-series prediction model of the employment attraction index is constructed using the residual correction grey model, and the employment attraction index for future time periods is predicted to obtain the evolution trend of employment attraction in the time dimension. Step S4: Based on the cross-sectional data of the multiple principal components in the spatial dimension, use cluster analysis to classify the multiple cities in the target area into different levels to obtain the classification results of the city employment attraction level, so as to obtain the distribution pattern of employment attraction in the spatial dimension. Step S5: Perform GIS visualization processing on the comprehensive evaluation index of employment attraction of each city, the evolution trend of employment attraction over time, or the classification results of employment attraction level of the city, to generate a thematic map reflecting the spatial distribution characteristics of employment attraction.

[0022] In a preferred embodiment, the preprocessing of the original dataset in step S1 includes: The original dataset is subjected to noise reduction and data cleaning, wherein: Delete useless tags, special characters, and irrelevant punctuation marks; For Chinese text segmentation, continuous text strings are divided into individual words; Stem extraction and semantic restoration of English or other text data with word form changes.

[0023] For example, noise reduction and data cleaning are performed on the employment science text data in the original dataset, and useless labels, special symbols, and irrelevant punctuation marks are deleted from the original dataset; frequent tone and stop words are filtered out from the original dataset; for Chinese text segmentation, continuous text strings are split into individual words; for English or other text data with word form changes, stemming and semantic restoration are performed.

[0024] In a preferred embodiment, in step S2, multiple principal components are extracted using principal component analysis, and a comprehensive evaluation index of employment attraction for each city is constructed based on the multiple principal components, specifically including: From the preprocessed original dataset, evaluation index data related to employment attraction in the arts industry are extracted in multiple dimensions. Principal component analysis is used to reduce the dimensionality of the evaluation index data in multiple dimensions and extract multiple principal components. The multiple dimensions include fine arts and art design, music performance and dance, and directing and opera. Determine the score of each principal component and its corresponding variance contribution rate; The comprehensive evaluation index of employment attraction for each city is obtained by weighting and summing the scores of each principal component and their corresponding variance contribution rates.

[0025] In a preferred embodiment, step S3 involves constructing a dynamic time-series prediction model for the employment attraction index using a residual-corrected grey model, specifically including: Step S31: Accumulate the original sequence of the comprehensive evaluation index of employment attraction to obtain the accumulated sequence; Step S32: Based on the accumulated sequence, establish and solve the grey differential equation to obtain the preliminary prediction model and its fitted sequence; Step S33: Calculate the residual sequence between the original sequence and the fitted sequence, and establish a residual prediction model based on the residual sequence; Step S34: Use the residual prediction model to correct the preliminary prediction model to obtain the residual corrected grey model, which serves as the dynamic time-series prediction model for the employment gravity index.

[0026] In a preferred embodiment, step S34, which involves correcting the preliminary prediction model using the residual prediction model, specifically includes: Determine the starting point for residual sequence modeling; For the predicted values ​​after the starting point, the prediction results of the preliminary prediction model and the prediction results of the residual prediction model are superimposed to obtain the corrected predicted values. For the predicted values ​​prior to the starting point, the prediction results of the preliminary prediction model remain unchanged.

[0027] In a preferred embodiment, step S4 involves classifying multiple cities within the target area into different levels using cluster analysis, specifically including: The scores of the multiple principal components are used as the basic data for cluster analysis, wherein each city in the basic data corresponds to a sample point; The distance between sample points in each city is calculated using a distance metric method, and a distance matrix is ​​constructed based on the calculated distance between the sample points in the city. The distance matrix is ​​used to record the similarity between any two sample points. Based on the similarity between any two sample points, the hierarchical clustering method is used to merge the two closest sample points into one cluster, and this process is repeated until all sample points are merged into one class to generate a cluster hierarchy diagram. The number of clusters is determined based on the clustering hierarchy diagram, and the classification results of urban employment attraction levels are obtained.

[0028] As a preferred embodiment, the distance measurement method includes one or more combinations of Euclidean distance, standardized Euclidean distance, Mahalanobis distance, Block distance, and Minkowski distance, used to eliminate the dimensional influence of the principal component scores.

[0029] As a preferred embodiment, the system clustering method includes the centroid method, wherein the centroid position of the merged cluster and its distance from any other cluster are jointly determined by the centroid position of each cluster before merging, the number of sample points, and the inter-cluster distance.

[0030] In a preferred embodiment, step S5 involves performing GIS visualization processing on the comprehensive evaluation index of employment attraction for each city, the evolution trend of employment attraction over time, or the classification results of urban employment attraction levels. Specifically, this includes: The comprehensive evaluation index of employment attraction of each city, the evolution trend of employment attraction over time, or the classification results of employment attraction level of each city are used as attribute data and associated with the urban geographic information spatial data of the target area. By utilizing the symbolization function of GIS software, differentiated visual identifiers are assigned to geographical elements of different cities based on the magnitude or category of attribute values, generating thematic maps that reflect the spatial distribution characteristics of employment attraction.

[0031] In this embodiment, by pre-setting keywords related to the arts industry and acquiring massive amounts of online search data using a web search index platform, a comprehensive evaluation index of employment attraction is constructed based on principal component analysis to extract multiple principal components. This achieves a quantitative assessment of the employment attraction of the arts industry in a city, supplementing and enriching the existing employment statistics systems of governments and universities. Preprocessing the acquired raw dataset effectively improves its practicality and reliability. A dynamic time-series prediction model of the employment attraction index is constructed using a residual-corrected grey model, enabling prediction of the employment attraction index for future periods and realizing dynamic forecasting of supply and demand changes in the arts industry employment market. The system monitors employment trends to provide a basis for decision-making. Cluster analysis is used to classify the employment attraction of multiple cities within a target region, revealing the spatial distribution pattern of employment attraction among cities and effectively identifying cities with different states of employment demand, such as strong, stable, and insufficient demand. This provides a scientific basis for formulating employment policies tailored to local conditions. Furthermore, GIS visualization of evaluation indices, predicted trends, or classification results generates thematic maps reflecting the spatiotemporal distribution characteristics of employment attraction. This transforms simple employment data into practical intelligence, simplifies abstract data mining results, enhances data classification understanding, and is more conducive to government decision-making research.

[0032] Example 2: This embodiment proposes an evaluation system for urban employment attraction in the arts industry. (See also...) Figure 2The system includes: The data processing module is used to obtain online search data of each city in the target area within a preset time period based on preset art industry keywords, form a raw dataset, and preprocess the raw dataset. The data extraction module is used to extract multiple principal components based on the preprocessed original dataset using principal component analysis, and to construct a comprehensive evaluation index of employment attraction for each city based on the multiple principal components. The data prediction module is used to construct a dynamic time series prediction model of the employment attraction index based on the time series data of the comprehensive evaluation index of employment attraction in each city, using a residual-corrected grey model, and to predict the employment attraction index for future time periods in order to obtain the evolution trend of employment attraction in the time dimension. The ranking module is used to classify multiple cities in the target area based on the cross-sectional data of the multiple principal components in the spatial dimension, and to obtain the ranking results of urban employment attraction in order to obtain the distribution pattern of employment attraction in the spatial dimension. The data visualization module is used to perform GIS visualization processing on the comprehensive evaluation index of employment attraction of each city, the evolution trend of employment attraction over time, or the classification results of employment attraction levels of the city, to generate thematic maps reflecting the spatial distribution characteristics of employment attraction.

[0033] Example 3: This embodiment provides corresponding experimental data based on the methods and systems described in Embodiments 1 and 2, as detailed below: Taking the arts industry in Jiangsu Province as an example, this study constructs an employment demand index using online job search data, explores the employment demand and changing trends in the arts industry in Jiangsu Province, establishes a time series prediction model for the employment demand index, and realizes dynamic prediction of changes in the employment market demand of the arts industry in Jiangsu Province.

[0034] 1. A Portrait of the Employment Situation of Art Majors in Universities. This study mines multi-source data on the employment quality of art graduates to provide empirical evidence for changes in social demand trends.

[0035] 2. Extraction and Analysis of Employment Situation Data Features in Jiangsu Province. Based on SSA (Self-Analysis and Analysis), trend and periodic components were extracted from employment situation data. Wavelet analysis was used to identify and analyze the periodic oscillation components of the employment situation index. Using December 27, 2010 as a baseline, the online search volume of employment keywords on the Baidu Index platform was used to map the changing trends in demand for arts-related employment in the internet age. 3. Employment Trend Analysis and Dynamic Forecasting in the Arts Industry of Jiangsu Province. A residual grey model GM(1,1) time series model was established to dynamically predict the supply and demand changes in the arts industry's employment market, enriching existing employment statistics and reconstructing the employment statistics framework.

[0036] 4. A Study on Employment Demand in the Arts Industry in Jiangsu Province Based on Data Mining. The principal component scores of employment demand in the arts industry in 13 prefecture-level cities of Jiangsu Province in 2021 were visualized to simplify the originally abstract data mining results. Computer graphics and image processing techniques were used to convert the data into graphics or images for display on the screen.

[0037] 5. Evaluation of Urban Employment Demand in Jiangsu Province's Arts Industry Based on Principal Component Clustering. This study uses visualization analysis of Jiangsu Province's arts employment data to statically display the spatial distribution characteristics of basic employment research and dynamically identify the evolution of employment trends in the arts industry across 13 prefecture-level cities in Jiangsu Province. A time-series model of the employment demand index for Jiangsu Province's arts industry is then constructed.

[0038] This study analyzed the Baidu Index of 13 keywords across 7 categories (music, dance, performance, broadcasting and hosting, fine arts and design, calligraphy, and opera) within the arts industry, encompassing 42 professional directions. The search area was categorized according to the research needs: Jiangsu, nationwide, and 13 cities in Jiangsu Province. A Python web crawler was used to precisely extract daily and weekly data for 10 years and 8 months (December 27, 2010 to August 30, 2021). Unsupervised learning methods were employed to clean, integrate, extract features, analyze trends, evaluate models, and represent the massive dataset. A Jiangsu Province arts industry employment demand index model was established, and anomaly detection was performed on the time dimension of the Jiangsu arts industry employment demand index.

[0039] By analyzing the trend of comprehensive index data from 557 weeks (December 27, 2010 to August 30, 2021), a residual-corrected GM(1,1) grey model based on principal component analysis is established. This model reveals the time-domain and frequency-domain job-seeking characteristics of my country's employment population over the past 10 years and 8 months, and a dynamic time-series prediction model for the employment demand index is established. Based on existing Baidu search index data, a Jiangsu Province arts industry employment demand assessment index is synthesized using existing principal component analysis, serving as a prediction model for the future Jiangsu Province arts industry employment demand index.

[0040] For example, see Figure 3First, a study on the employment situation of arts majors in Jiangsu Province based on data mining. This study accurately assesses the "state" (real-time information about the development of events at a given moment) and "momentum" (potential information about the future state of events after changes in the next moment) of the arts employment situation. Utilizing a big data platform combining search keywords and internet indices (mobile and PC), principal component analysis (PCA) combined with wavelet analysis (WA) is used to extract trend and periodic components from employment data. This analysis explores the job-seeking characteristics of the arts workforce in Jiangsu Province over the past 10 years in the time and frequency domains, compares the differences in employment supply and demand in southern, central, and northern Jiangsu, and proposes corresponding countermeasures for government and university employment work in the new era to promote fuller and higher-quality employment for arts graduates.

[0041] Second, an unsupervised learning-based employment index model for arts-related industries in Jiangsu Province was established. Mathematical models such as Grey Model (GM), Stepwise Regression (SR), and Cluster Analysis (CA) were used to analyze the employment index. Principal component analysis was performed to reduce the dimensionality of the 10-dimensional arts-related industry keyword indices. The main components with the greatest impact on arts-related employment were analyzed, and time series plots were created based on these components to establish a time series model, resulting in an employment demand index model for the arts-related industries. The fitting effect was tested using 2021 data, and the development patterns and trends of arts-related employment demand in 2022 were predicted.

[0042] For example, see Figure 4 Art colleges should actively promote domestic and international exchanges in talent cultivation, academic research, and artistic creation. They should explore, discover, co-build, and share art courses, majors, and faculty with renowned universities, associations, and platforms in the industry to continuously improve the professional skills, management level, and artistic vision of teachers and students. They should also establish teams of industry mentors, industry experts, art masters, and career guidance counselors to actively carry out artistic professional creation practices, and hold influential international academic conferences and salon activities to enhance the professional competitiveness and employment competence of college art students. Coupling refers to the phenomenon where two or more systems influence each other and even unite through interaction. Using a multi-source data fusion-based coupling model of employment competitiveness and employment quality, an integrated system for college art student employment education is constructed, encompassing "multi-source data fusion → evaluation indicator construction → employment competitiveness enhancement → employment quality assurance."

[0043] For example, see Figure 5Various research activities related to employment generate a large amount of new data. Scientific data originates from the factual records of scientific research, and literature is the main carrier and best channel for scientific and technological knowledge. This study mines employment science data through processes such as data cleaning, data reduction, model selection, feature extraction, and category description. It uses deep learning text information feature extraction to train a model for dimensionality reduction and extraction of employment science data, and conducts multi-scale evaluation to detect the spatiotemporal distribution and trends of employment science data, achieving multi-dimensional matrix construction. Furthermore, it combines multivariate analysis with strategic coordinates to conduct strategic intelligence analysis on China's employment development over the past 20 years, realizing the application of scientometric methods in discipline construction. Through word-text clustering analysis, it identifies current research hotspots in key employment areas, analyzes the distribution of co-occurring strategic coordinates, and further understands the development trends of various research hotspots, providing decision-making references for university faculty and research management personnel in employment and entrepreneurship.

[0044] Data Collection and Preprocessing: Employment science data is characterized by its accuracy, reliability, large volume, interactivity, and relevance. Various mathematical models are used to mine employment science data, extract and analyze feature information, and study employment trends and spatiotemporal distribution. The data sources are mainly divided into four parts: 1. Quality data quantified through various questionnaires, including employment rate, graduate destination distribution, average monthly income, job satisfaction, professional relevance, turnover rate, enrollment, entrepreneurship, overseas study, military service, and further education; 2. Data encoded through field notes from telephone follow-ups with employers and semi-structured in-depth interviews with previous graduates; 3. Trace data from the "Internet + Big Data" platform, based on massive amounts of internet user behavior data, including Baidu Index, CNKI Knowledge Element Index, Sina Weibo Public Opinion, Google Index, and Weibo Index; 4. Literature data related to universities and employment from the CNKI database and the Web of Science Core Collection.

[0045] The text data on employment science is subjected to noise reduction and data cleaning: useless tags, special symbols, and irrelevant punctuation are deleted; frequent tone and stop words are filtered; for Chinese text, continuous text strings are segmented into individual words; and for English and other text data with word form changes, stemming and semantic restoration are performed.

[0046] Feature Extraction and Multidimensional Evaluation (Model Selection): Deep learning is widely used in employment science data mining technology. It can better understand, process and use data. Through deep learning models such as pre-trained language models (PLMs) → natural language processing (NLP) → text feature extraction (DNN), the dimensionality of employment science data can be reduced to obtain important feature information such as text classification, generation, similarity calculation, information extraction, sentiment analysis, and name recognition.

[0047] This study employs a spatiotemporal perspective to qualitatively and quantitatively identify and evaluate employment science text data, thereby obtaining comprehensive characteristics of the data. Characteristic information is visualized through static exploration across five spatial dimensions: country, institution, core author group, research theme, journal publication volume, and geographical distribution. This analysis examines the research fields of employment sectors, regions, themes, and core authors. Characteristic information is also quantified through dynamic monitoring across five temporal dimensions: time series evolution, disciplinary distribution growth, knowledge theme evolution, journal type changes, and frontier prediction analysis. This analysis predicts the evolution of research characteristics, trajectory hotspots, theme evolution, and future development trends over the past 20 years.

[0048] By scanning, extracting, analyzing, classifying, storing, statistically calculating, and performing matrix analysis on bibliographic information related to employment science, accurate basic data can be obtained. Multidimensional relationship construction can then yield various data matrices applicable to cluster analysis and social network analysis, such as word matrix, co-occurrence matrix, bimodal matrix, and coupling matrix.

[0049] Multivariate Analysis and Category Description: Employment science data classification refers to the process of dividing text mining into multiple categories. It can be divided into multiple subcategories based on different methods and techniques. Cluster analysis is an important statistical method in multivariate analysis, dividing a dataset into several clusters. Due to the similarity of data within clusters, it's not possible to simply summarize the results of each cluster; it's necessary to analyze the commonalities within the text of each cluster and group them into multiple clusters based on the degree of similarity between clusters. The category of a cluster is described by calculating the high-frequency words in the text within each cluster (#1, #2, ..., #C). The first column of the word matrix contains keywords (or cited references), and the first row contains document record numbers (or source documents). The numbers 1 and 0 in the matrix represent whether the keywords appear in the employment science data or whether they are cited by the source documents, respectively. The similarity of the word matrix is ​​calculated using the Ochiai's Coefficient, and a suitable inter-cluster distance algorithm is used to generate a cluster map of the employment science data. The core semantic interpretation of each category in the cluster map is analyzed, and similarities and differences are identified to summarize the meaning of each cluster in the employment field.

[0050] For example, see Figure 6This study analyzes the employment status and trends of college art students since 2009. The analysis reveals that the number of college graduates in 2025 is projected to reach 12.22 million, an increase of 430,000 compared to the previous year. In 2024, a total of 56,600 people (including those applying for other majors) registered for the provincial unified art examination. In June 2023, the unemployment rate for young people aged 16 to 24 peaked at 21.3%, showing an upward trend. Since 2020, the diffusion index of residents' perception of the current employment situation has been in a contractionary state. With the revolution of new productive technologies such as Internet+, big data+, cloud computing, and AI+, the traditional art industry is constantly transforming and upgrading. The temporary instability in industry supply and demand has created structural contradictions in employment, putting certain pressure on the employment of art students.

[0051] In this embodiment, the analysis and dynamic forecast of employment trends in the arts industry in Jiangsu Province specifically includes: 1. Graph representation optimization based on nonlinear transformation: Graphical representations of high-dimensional data are generally quite complex, which is detrimental to subsequent processing. Based on the requirements of subsequent clustering methods, optimization methods based on nonlinear functions can make the graphical representations more consistent with practical processing requirements and the principles of visualization and interactivity. The basic criteria for selecting the nonlinear function f(x) when describing data nonlinearly are as follows.

[0052] (1) ,when ; (2) and Scope, when Sometimes, .

[0053] The first condition is the range requirement; the second requirement is that the function must be monotonically non-decreasing. Functions that satisfy both requirements can be used as transformation functions for the optimization of graphical representations.

[0054] For certain multivariate graphs, piecewise functions can be used to process them based on the characteristics of their distribution.

[0055]

[0056] Each sub-function in the formula should satisfy the constraint condition of x. The "magnifying glass function" has a good local magnification effect, and the formula is:

[0057] In the formula: .

[0058] 2. Grey System Model Due to differences in their environments and levels of knowledge, people have varying degrees of understanding of many natural phenomena in the objective world. Based on their level of understanding of specific systems, systems are generally categorized as "white-box systems," "black-box systems," and "grey-box systems." A "white-box system" refers to a system whose internal structure is fully understood, and in many cases, a mathematical model of the system has been established. A "black-box system" refers to systems whose internal structure is completely unknown; only stimulus and response information can be obtained, and sometimes even this information is difficult to obtain. A "grey-box system" lies between "white-box systems" and "black-box systems," meaning that some basic information about the system is known, but the system is not fully understood; research can only be conducted based on statistical inference or some kind of logical thinking. This research method is known as the grey system method. Regarding the uncertainty prediction of the employment demand index for the arts industry in Jiangsu Province, an extended model of the GM(1,1) model, the residual GM(1,1) model, will be introduced.

[0059] (1) GM(1,1) Grey Model Given the original sequence ,in , ; for A positive accumulation generates a 1-AGO sequence. ,in, We obtain the original GM(1,1) model:

[0060] A grey GM(1,1) difference model can be established

[18] :

[0061] The development coefficient -a and the grey action b are solved using the least squares algorithm. The differential equation is then solved, and the accumulated data is restored to obtain the estimated sequence.

[0062] The accuracy of the obtained gray model is checked to determine its accuracy level.

[0063] The time response of the mean GM(1,1) model is:

[0064] Further solving yields Time restoration formula is

[0065] The GM(1,1) model is tested using residual tests, correlation tests, and posterior variance tests.

[0066] (2) Grey model of residual GM(1,1) When the simulation accuracy fails to meet the requirements of various forms of the GM(1,1) model, it is advisable to establish a GM(1,1) model for the residual sequence and modify the original model to improve the simulation accuracy.

[0067] Due to the difference between the derived value and the cumulative derived value, to reduce errors caused by round-trip calculations, a method is usually used. The residual, correction Simulated values .

[0068] set up for The residual sequence, where If it exists Satisfy: ① , Consistent symbols; ② .

[0069] but The constructible residual tail sequence is denoted as .

[0070] Its 1-AGO sequence is .

[0071] By cumulative subtraction and restoration, the residual estimate sequence can be obtained.

[0072] like The corresponding residual corrected time series reconstruction is:

[0073] The above equation is the GM(1,1) residual correction model with cumulative reduction.

[0074] 3. Establishment and prediction of a dynamic time series model for the employment demand index Baidu Index is a web information platform developed based on massive search volume, analyzing the attention and changing trends of internet users towards keywords. It scientifically calculates the weighted average search frequency of various keywords on Baidu web pages, featuring diverse data types and high-efficiency processing. The comprehensive search index is divided into PC-based and mobile-based indices based on data source. In the fields of economics, sociology, psychology, and education, scholars primarily use Baidu search data for data mining, trend prediction, and public opinion monitoring.

[0075] (1) Model establishment Principal component analysis of the Jiangsu Province Arts Industry Employment Demand Comprehensive Assessment Index for 557 weeks revealed a significant shift before and after week 7 of 2016. Therefore, the data was divided into two parts: week 1 of 2011 to week 7 of 2016, and week 8 of 2016 to week 35 of 2021. Time series models were fitted to these two parts respectively.

[0076] A predictive analysis was conducted on the comprehensive assessment index of employment demand in the arts sector of Jiangsu Province for week 557, such as... Figure 7 As shown, the fitting and prediction effect of the dynamic time series model of employment demand index in the arts industry in Jiangsu Province is shown. The blue solid line "—" represents the original employment demand index sequence, and the pink dotted line "-----" represents the predicted index sequence. The trend fitting effect of the past 10 years and 8 months is good.

[0077] (2) Prediction and verification Table 1. Simulation error of residual GM(1,1)

[0078] From Table 1, the sum of squared residuals can be calculated as follows:

[0079] The average relative error is

[0080] The simulation accuracy of residual correction GM(1,1) is significantly improved. If the correction accuracy is still unsatisfactory and the residual sequence no longer meets the modeling requirements, then it is necessary to consider using other models or making appropriate selections from the original data sequence.

[0081] Table 1 shows the prediction of the employment demand index for four weeks from week 32 to week 35 of 2021 using the established residual-corrected GM(1,1) grey model. Based on existing Baidu search indexes, the employment demand assessment index for the Jiangsu Province arts industry in weeks 554, 555, 556, and 557 was obtained using principal component analysis, which yielded values ​​of -1.2030, -1.1903, -1.1798, and -1.1531, respectively. The sum of squared residuals from the predicted values ​​was 0.0214, with an average relative error of 5.2875%, consistent with the actual trend, indicating that the model's prediction effect is good.

[0082] In this embodiment, the evaluation of urban employment demand in the arts sector of Jiangsu Province based on principal component clustering is specifically as follows: 1. Distance coefficient From a statistical perspective, cluster analysis is a method of dimensionality reduction through data modeling. There are two types of metrics for measuring similarity: distance and similarity coefficient. Distance measures the similarity between samples; similarity coefficient measures the similarity between variables.

[0083] 1) Euclidean distance Assume there are two Dimensional Samples and Then their Euclidean distance is

[0084] 2) Standardized Euclidean distance Assume there are two Dimensional Samples and Then their standardized Euclidean distance is

[0085] in: express The variance matrix of each sample. , Indicates the first The variance of the column. 3) Mahalanobis distance Assuming there is a total The first indicator, the A total of 10 indicators were measured. Data (requirements) ): ,

[0086] Therefore, we obtain Data matrix of order Each row represents a sample data point. 1-order data matrix of The covariance matrix of order is denoted as . two Dimensional Samples and The Mahalanobis distance is as follows:

[0087] Mahalanobis distance takes into account the standardization of the dimensions of each indicator, and is an improvement over other distance methods. Mahalanobis distance not only eliminates the influence of dimensions, but also reasonably considers the correlation of indicators. 4) Block distance two Dimensional Samples and The Brock distance is as follows:

[0088] 5) Minkowski distance two Dimensional Samples and The Minkowski distance is as follows:

[0089] Note: The time is the Brock distance; The time is the Euclidean distance.

[0090] 2. Similarity coefficient Similarity coefficient (cosine of the angle, correlation coefficient) For correlation coefficient (similarity coefficient) and cosine of the angle, the similarity coefficient is a quantity that describes the degree of similarity between samples. The result is expressed as "1-correlation coefficient = 1-correlation" or "1-cosine of the angle = 1-cosine". Because the similarity coefficient and the cosine of the angle cannot be directly used to represent distance, only their difference from 1 can satisfy the requirements of the distance axiom.

[0091] 1) Similarity coefficients (cosine of the included angle, cosine distance, dissimilarity matrix) two Dimensional Samples and The cosine distance is

[0092] This is a standard inspired by the geometric principle of similarity. When recognizing images and text, the cosine of the included angle is often used as the standard. =1, indicating two samples and Completely similar; Approaching 1 indicates and Closely similar; =0, indicating and Completely different; Approaching 0 indicates and The differences are significant. Based on the dissimilarity matrix constructed from cosine distances, n samples can be classified, grouping similar samples into one class and others into different classes.

[0093] 2) Similarity distance (correlation coefficient) two Dimensional Samples and The similarity distance is

[0094] in, The correlation coefficient between variables is used to characterize the similarity between samples.

[0095] 3. Clustering methods Matlab uses hierarchical clustering (also known as hierarchical distance method). Hierarchical clustering is one of the most widely used methods in cluster analysis. Its basic principle is: first, a certain number of samples or indicators are each considered as a separate class; then, based on the degree of similarity between the samples or indicators, the two classes with the highest similarity are merged; this process is repeated until all samples are grouped into one class.

[0096] (1) Basic idea of ​​system clustering Clustering begins by assigning each of the n samples (or p variables) to a separate cluster, defining the distances between samples (or variables) and between clusters. The two closest clusters are then merged into a new cluster (called a merge). The distances between this new cluster and other clusters are calculated, and this process is repeated, reducing the number of clusters by one each time, until all samples (or variables) are merged into one cluster. This results in a clustering tree diagram (or phylogenetic diagram), which clearly shows the number of clusters and the samples (or variables) contained in each cluster. Alternatively, statistical measures can be used to determine the classification results.

[0097] In cluster analysis, a class is usually represented by G, which is assumed to have m elements (i.e., samples or variables). Without loss of generalization, a column vector is used. To indicate, express and Inter-space distance, Representation Class With class The distance between classes. Different methods of defining distance between classes result in different hierarchical clustering methods. Hierarchical clustering, also known as hierarchical distance methods, includes methods such as shortest distance, longest distance, average distance, weighted distance, and centroid method. The shortest distance method is too restrictive, while the longest distance method is too expansive. These two methods are simple and easy to understand, but they represent extreme classification methods. A suitable centroid method is used to analyze the employment demand in the arts industry in Jiangsu Province.

[0098] (2) Shortest distance method (single linkage method) The distance between classes is defined as the distance between the nearest samples in the two classes (the most basic and commonly used definition).

[0099] If a certain step class With class They are grouped into a new class, denoted as ,kind With any existing class The distance between them is

[0100] The steps of shortest distance clustering are as follows: ① Initially, each sample (or variable) is treated as a separate class, and the distance between samples (or variables) is defined, usually using Euclidean distance. Calculate the distance matrix for n samples (or p variables). It is a symmetric matrix.

[0101] ② Search Let the smallest element in the middle be denoted as . ,Will and They are grouped into a new class, denoted as ,Right now

[0102] ③ Calculate the new class With any class The recursive formula for the distance between them is:

[0103] For distance matrix Make modifications, and The row and column are merged into a new row and column, corresponding to The new distances in the new rows and columns are calculated using the above formula, while the values ​​in the remaining rows and columns remain unchanged. The resulting new distance matrix is ​​denoted as... .

[0104] ④ Repeat the above for The two-step operation yields the distance matrix. This continues until all elements are merged into one category.

[0105] (3) Centroid hierarchical method The distance between classes is defined as the Euclidean distance between their centroids (i.e., class means). Let... There is One element, There is One element, defining a class and The centers of gravity are respectively ,

[0106] but and The square distance between them is

[0107] The recursive formula for the squared distance between classes is:

[0108] 4. Distance and Method Selection The key to cluster analysis lies in using appropriate distance and clustering methods. The application of distance and clustering methods varies depending on the research object, and can be summarized in three points: ① When choosing a distance, it's best to eliminate the influence of dimensions. The dimensions of variables need to be consistent. Using the same distance and clustering method, differences in dimensions can aggravate or mitigate the impact of some variables. Clustering results based on standardized and non-standardized data can be diametrically opposed. Only by using standard Euclidean distance or precision-weighted distance can the negative impact of dimensions be limited.

[0109] ② Regarding the choice of clustering methods, the shortest distance method is too restrictive, while the longest distance method is too expansive. These two methods are simple and easy to understand, but they represent extreme classification approaches. More suitable methods include the centroid method, the average method, and the weighted method. ③ Issues related to variable correlation should be avoided. When different variables are correlated, some information is amplified while others is relatively less so, leading to discrepancies between the clustering results and the actual situation. In such cases, using Mahalanobis distance or cluster analysis based on principal component scores can yield a satisfactory classification scheme.

[0110] Clustering methods are not difficult to understand, but their implementation is quite complicated. Sometimes, to achieve a satisfactory classification result, researchers need to experiment, compare, and analyze repeatedly to arrive at a better clustering scheme.

[0111] 5. Clustering effect evaluation Cophenetic (cophenetic correlation coefficient) refers to the correlation coefficient of similar appearance, the correlation coefficient of similar type, and the common classification correlation coefficient CPCC. It calculates the correlation between the distance between the clustering tree information and the original data. The larger this value, the better.

[0112] The cophenetic correlation coefficient, or joint phenotypic correlation coefficient, is used in clustering progress tables (Z-matrix) to determine a critical scale for class merging. Regardless of the classification method used, the distance between samples must be considered. The aforementioned pdist provides the distances above and below the diagonal of the distance matrix, denoted by D. If the clustering method is appropriate, the critical scale for clustering should have a high correlation with the distance matrix; this correlation is measured by the cophenetic correlation coefficient.

[0113] in, Representing clustering Original data matrix; Pairwise distances (column vectors) calculated using different distance methods (Euclidean distance, standard Euclidean distance, Mahalanobis distance, street distance, Minkowski distance, Chebyshev distance, etc.); Through the The distance matrix is ​​obtained by arranging the pairwise distances into a square matrix (based on the pairwise distances calculated from the pairwise distances). Based on distance matrix The results were classified using different methods (shortest distance method, longest distance method, weighted average method, centroid method, median distance method, etc.) (hierarchical clustering function). The three-column matrix represents the clustering schedule.

[0114] This correlation coefficient measurement formula is isomorphic to the simple correlation coefficient formula, but the specific calculation process is more complicated because the third column elements of the D matrix and the Z matrix are inconsistent.

[0115] The Cophenetic correlation coefficient ranges from [0,1]. The closer the value is to 1, the stronger the correlation between Z and D. A higher Cophenetic correlation coefficient generally indicates better clustering, but the reverse is not necessarily true. A high Cophenetic correlation coefficient does not necessarily mean the clustering results are acceptable. An increase in the correlation coefficient from 0.5 to 0.7 suggests an improvement in clustering performance from a purely statistical perspective, but the clustering results should be analyzed realistically based on the actual situation.

[0116] 6. Clustering of Employment Demand Index for Arts-Related Industries in 13 Cities of Jiangsu Province Because of the differences among principal component indices in principal component analysis (PCA), and the variations in analytical indices across spatial scales, this study aims to further understand the trend classification of employment demand indices in the arts sector across 13 prefecture-level cities in Jiangsu Province. The principal components of PCA were used as the source data for cluster analysis, which was then employed to verify the accuracy of its hierarchical classification and ranking. Cluster analysis includes the following steps: Construct n classes, each containing only one sample; Clustering between each class is calculated using the standardized Euclidean distance and centroid distance method to obtain the distance matrix. The common distance between classes is...

[0117] Merge the two closest classes into a new class; Calculate the distance between the merged new class and other classes; Draw a cluster hierarchy distribution map and identify the number of clusters based on the cluster map.

[0118] To avoid the correlation problem among the indicators in principal component analysis, principal components F1, F2, and F3 were used as cluster analysis data. MATLAB programming was used to perform cluster analysis on the employment demand for arts-related industries in 13 prefecture-level cities in Jiangsu Province, yielding the following results: Figure 8 The chart shows the clustering distribution of employment demand in the arts sector across 13 cities in Jiangsu Province in 2021. The joint phenotypic correlation coefficient is 0.9677, indicating a good clustering effect.

[0119] Principal component cluster analysis was used to evaluate the employment demand in the arts sector in 13 prefecture-level cities of Jiangsu Province using 13 multidimensional indicators. The results were categorized into three levels: high demand, stable demand, and insufficient demand. The specific clustering levels are shown in Table 2. The results are consistent with the actual situation of the arts employment market in Jiangsu Province, reflecting the scientific validity of this study. This research can provide decision-making and reference for improving the assessment of employment demand in the arts sector in various prefecture-level cities of Jiangsu Province and for the scientific formulation of relevant policies.

[0120] Table 2. Cluster Rank of 13 Cities in Jiangsu Province in 2021 Based on Principal Component Analysis and Standardized Euclidean Distance Centroid Method

[0121] 8. Countermeasures and Suggestions By constructing a comprehensive demand evaluation index system for the arts industry in Jiangsu Province, it was found that Jiangsu Province has obvious regional development imbalances, with some areas experiencing lagging economic development and insufficient employment demand. To address these issues, the following countermeasures and suggestions are proposed: (1) For the top four cities in Jiangsu Province, namely Suzhou, Nanjing, Wuxi and Nantong, which are in the first tier, firstly, they should firmly grasp the major opportunities at the national level, such as the Yangtze River Economic Belt and the integrated development of the Yangtze River Delta region, improve their international competitiveness, and focus on the "qualitative improvement" of the art market; secondly, they should give full play to their radiating and driving role in central and northern Jiangsu, and improve the overall strength of the province and the Yangtze River Delta region.

[0122] (2) For Xuzhou, Changzhou, Huai'an, Yancheng and Lianyungang, which are in the second tier, they should take advantage of their good transportation and regional advantages, actively strengthen their ties with southern Jiangsu, promote the free flow of production factors such as resources, technology and information, accelerate the pace of integration with southern Jiangsu, and achieve the rapid rise of the art industry.

[0123] (3) For Zhenjiang, Taizhou, Yangzhou and Suqian, which are in the third tier, there are gaps in all aspects with southern Jiangsu due to their relatively weak economic foundation. To achieve comprehensive catch-up, the primary task is to make up for the shortcomings in the regional development of the arts industry as soon as possible, develop local traditional advantageous arts industries according to local conditions, vigorously improve economic strength, and strengthen economic and technological cooperation, innovation and talent exchange with southern and central Jiangsu.

[0124] For example, see Figure 9 Cluster analysis of the first and second principal components of data from 13 cities in Jiangsu Province in 2021. Figure 9 The spatial distribution of employment attraction in the arts sector was clustered, revealing variations across different cities in other principal factors. Principal component clustering analysis was used to comprehensively evaluate the employment attraction of arts-related cities in Jiangsu Province using 13 multidimensional indicators, categorizing them into four levels: strong employment attraction (#1), rising employment attraction (#2), stable employment competitiveness (#3), and insufficient employment competitiveness (#4). Suzhou, Nanjing, Wuxi, and Nantong significantly outperformed Suzhou in the comprehensive economic attraction factor score for the arts sector. Suzhou consistently ranked first in the province for many years in indicators related to fine arts and art design, music performance and dance, and directing and opera. Nanjing, as the capital city of Jiangsu Province, is the political, economic, and cultural center with a highly advantageous geographical location, performing well in GDP, fine arts and art design, music performance and dance, and directing and opera. Wuxi, known as a land of plenty, is an important regional center city in the Yangtze River Delta region, with closer ties to Shanghai, effectively promoting its rapid development; its per capita GDP has remained the highest in the province for many years. Nantong, with a GDP exceeding one trillion yuan, is known as the "Northern Shanghai" and is an economic center and modern port city on the northern wing of the Yangtze River Delta, boasting a large population. In terms of professional scores in music performance and dance, Nantong, Lianyungang, and Wuxi stand out among the 13 prefecture-level cities, with professional singers and music editors ranking among the top in the province. In terms of professional scores in directing and opera, Huai'an, Nanjing, and Yangzhou perform particularly well, maintaining high levels in indicators such as drama actors and film and television directors.

[0125] In this embodiment, the GIS visualization analysis of the spatial dimension of urban employment activity specifically includes: "Visibility" refers to visual accessibility, that is, the range and extent of visibility from one or more locations. ArcGIS is a comprehensive system that users can use to collect, organize, manage, analyze, communicate, and publish geographic information. It is a world-leading platform for building and applying geographic information systems, enabling people around the world to apply geographic knowledge to government, business, science and technology, education, and media.

[0126] Using GIS for visibility analysis can transform the sensory perception of employment attraction into quantifiable indicators, becoming a scientific and effective decision-making tool for leadership. The results of visibility analysis can serve as an important basis for decision-making and layout. With the development of GIS technology, it is widely used in visibility analysis.

[0127] The principal component scores of employment attraction in the arts sector in 13 prefecture-level cities of Jiangsu Province in 2021 were visualized to simplify the originally abstract mining results. Computer graphics and image processing technologies were used to visualize the data and enhance the understanding of data classification.

[0128] The results of spatial analysis can be presented in maps and reports. Spatial analysis can be used to discover hidden patterns and relationships. By conducting GIS visualization spatial analysis of the employment attraction index of the arts industry in 13 prefecture-level cities in Jiangsu Province, simple data can be transformed into practical intelligence, which is more conducive to decision-making research.

[0129] like Figure 10 The image shows the distribution of overall activity levels in the arts sector across 13 cities in Jiangsu Province in 2021. The color differentiation clearly reveals the differences in attractiveness among the cities; darker colors indicate higher activity levels, while lighter colors indicate lower activity levels. Suzhou and Nanjing are in Tier I, Wuxi and Nantong are in Tier II, Xuzhou, Changzhou, and Yancheng are in Tier III, and Huai'an, Lianyungang, Yangzhou, Suqian, Zhenjiang, and Taizhou are in Tier IV (the lightest color).

[0130] like Figure 11 As shown, the distribution of the overall activity level of the arts industry in various prefecture-level cities of Jiangsu Province in 2021 shows that the attraction of the arts industry in southern Jiangsu is higher than that in central Jiangsu, while northern Jiangsu ranks last. This indicates that the overall activity level of the arts industry in Jiangsu Province is positively correlated with the employment market and economic development level.

[0131] The above description is merely an embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A method for evaluating the urban employment attraction of the arts industry, characterized in that, The method includes the following steps: Based on preset art industry keywords, obtain online search data of each city in the target area within a preset time period to form a raw dataset, and preprocess the raw dataset. Based on the preprocessed original dataset, multiple principal components are extracted using principal component analysis, and a comprehensive evaluation index of employment attraction for each city is constructed based on these multiple principal components. Based on the time-series data of the comprehensive evaluation index of employment attraction in each city, a dynamic time-series prediction model of employment attraction index is constructed using the residual correction grey model, and the employment attraction index for future time periods is predicted to obtain the evolution trend of employment attraction in the time dimension. Based on the cross-sectional data of the multiple principal components in the spatial dimension, cluster analysis is used to classify multiple cities in the target area into different levels, so as to obtain the classification results of urban employment attraction levels and to obtain the distribution pattern of employment attraction in the spatial dimension. The comprehensive evaluation index of employment attraction of each city, the evolution trend of employment attraction over time, or the classification results of employment attraction level of each city are processed by GIS visualization to generate a thematic map reflecting the spatial distribution characteristics of employment attraction.

2. The method for evaluating the urban employment attraction of the arts industry according to claim 1, characterized in that, The preprocessing process for the original dataset includes: The original dataset is subjected to noise reduction and data cleaning, wherein: Delete useless tags, special characters, and irrelevant punctuation marks; For Chinese text segmentation, continuous text strings are divided into individual words; Stem extraction and semantic restoration of English or other text data with word form changes.

3. The method for evaluating the urban employment attraction of the arts industry according to claim 1, characterized in that, Principal component analysis was used to extract multiple principal components, and a comprehensive evaluation index of employment attraction for each city was constructed based on these principal components, specifically including: From the preprocessed original dataset, evaluation index data related to employment attraction in the arts industry are extracted in multiple dimensions. Principal component analysis is used to reduce the dimensionality of the evaluation index data in multiple dimensions and extract multiple principal components. The multiple dimensions include fine arts and art design, music performance and dance, and directing and opera. Determine the score of each principal component and its corresponding variance contribution rate; The comprehensive evaluation index of employment attraction for each city is obtained by weighting and summing the scores of each principal component and their corresponding variance contribution rates.

4. The method for evaluating the urban employment attraction of the arts industry according to claim 1, characterized in that, A dynamic time-series prediction model for the employment gravity index is constructed using a residual-corrected grey model, specifically including: The original sequence of the comprehensive evaluation index of employment attraction is accumulated to generate the accumulated sequence; Based on the accumulated sequence, a grey differential equation is established and solved to obtain a preliminary prediction model and its fitted sequence; Calculate the residual sequence between the original sequence and the fitted sequence, and establish a residual prediction model based on the residual sequence; The preliminary prediction model is modified using the residual prediction model to obtain the residual modified grey model, which serves as the dynamic time-series prediction model for the employment gravity index.

5. The method for evaluating the urban employment attraction of the arts industry according to claim 4, characterized in that, The step of revising the preliminary prediction model using the residual prediction model specifically includes: Determine the starting point for residual sequence modeling; For the predicted values ​​after the starting point, the prediction results of the preliminary prediction model and the prediction results of the residual prediction model are superimposed to obtain the corrected predicted values. For the predicted values ​​prior to the starting point, the prediction results of the preliminary prediction model remain unchanged.

6. The method for evaluating the urban employment attraction of the arts industry according to claim 1, characterized in that, Cluster analysis was used to classify multiple cities within the target region into different levels, specifically including: The scores of the multiple principal components are used as the basic data for cluster analysis, wherein each city in the basic data corresponds to a sample point; The distance between sample points in each city is calculated using a distance metric method, and a distance matrix is ​​constructed based on the calculated distance between the sample points in the city. The distance matrix is ​​used to record the similarity between any two sample points. Based on the similarity between any two sample points, the hierarchical clustering method is used to merge the two closest sample points into one cluster, and this process is repeated until all sample points are merged into one class to generate a cluster hierarchy diagram. The number of clusters is determined based on the clustering hierarchy diagram, and the classification results of urban employment attraction levels are obtained.

7. The method for evaluating the urban employment attraction of the arts industry according to claim 6, characterized in that, The distance measurement method includes one or more combinations of Euclidean distance, standardized Euclidean distance, Mahalanobis distance, Block distance, and Minkowski distance, used to eliminate the dimensional influence of the principal component scores.

8. The method for evaluating the urban employment attraction of the arts industry according to claim 6 or 7, characterized in that, The hierarchical clustering method includes the centroid method, wherein the centroid position of the merged cluster and its distance from any other cluster are jointly determined by the centroid position of each cluster before merging, the number of sample points, and the inter-cluster distance.

9. The method for evaluating the urban employment attraction of the arts industry according to claim 1, characterized in that, The comprehensive evaluation index of employment attraction for each city, the evolution trend of employment attraction over time, or the classification results of employment attraction levels for each city will be visualized using GIS, specifically including: The comprehensive evaluation index of employment attraction of each city, the evolution trend of employment attraction over time, or the classification results of employment attraction level of each city are used as attribute data and associated with the urban geographic information spatial data of the target area. By utilizing the symbolization function of GIS software, differentiated visual identifiers are assigned to geographical elements of different cities based on the magnitude or category of attribute values, generating thematic maps that reflect the spatial distribution characteristics of employment attraction.

10. A city employment attraction evaluation system for the arts industry, characterized in that, The system includes: The data processing module is used to obtain online search data of each city in the target area within a preset time period based on preset art industry keywords, form a raw dataset, and preprocess the raw dataset. The data extraction module is used to extract multiple principal components based on the preprocessed original dataset using principal component analysis, and to construct a comprehensive evaluation index of employment attraction for each city based on the multiple principal components. The data prediction module is used to construct a dynamic time series prediction model of the employment attraction index based on the time series data of the comprehensive evaluation index of employment attraction in each city, using a residual-corrected grey model, and to predict the employment attraction index for future time periods in order to obtain the evolution trend of employment attraction in the time dimension. The ranking module is used to classify multiple cities in the target area based on the cross-sectional data of the multiple principal components in the spatial dimension, and to obtain the ranking results of urban employment attraction in order to obtain the distribution pattern of employment attraction in the spatial dimension. The data visualization module is used to perform GIS visualization processing on the comprehensive evaluation index of employment attraction of each city, the evolution trend of employment attraction over time, or the classification results of employment attraction levels of the city, to generate thematic maps reflecting the spatial distribution characteristics of employment attraction.

Images