Research On Government Affairs Evaluation Data Of A Government Based On Cluster Analysis

The quality of government services directly affects public's satisfaction,and high-quality services play a vital role in people's livelihood and economic development.So the “Government Service Evaluation System” has emerged.A city responds positively to the “Government Service Evaluation System” policy and formulates relevant implementation rules for its work.Firstly,based on the data of more than 70,000 collected from this city in “Government Service Evaluation System” during the first half of 2021,this paper provides descriptive statistics on each indicator,combines charts and images to understand the characteristics of the data of “Government Service Evaluation System” in the city,and analyzes the differences in services among regions and departments.Then,based on the characteristics of the city's government service evaluation data,combined with the understanding of the implementation rules,practical and effective indicators with clear classification effects are constructed: The five evaluation indicators of favorable evaluation rate,active evaluation rate,fake favorable rates,“Ji Ban” rates,and the number of cases are identified,which indicate the satisfaction and participation of the public in the policies and services,the integrity and efficiency of the staff,and the administrative scale of the service institutions,which are the main factors affecting the quality of government services.Subsequently,a K-means clustering model was constructed based on the standardized data,and the appropriate number of cluster,K,is determined by the Elbow Method and the Silhouette Coefficient Method.Finally,the models are established with region and department respectively,and the problems are analyzed according to the clustering results,and suggestions for improvement of government services are put forward.The clustering results are more desirable and more in line with the actual situation,and the differences in the quality of government services among the subjects can be analyzed at the clusters.However,in the process of analyzing the results,it was found that for different subjects,the clustering performance of the same set of indicators differed.In the regional clustering,the “Ji Ban” rates is not very useful for the clustering results and can be considered to be removed.While in the departmental clustering,it is found that the positive rating rate,active evaluation rate,and fake favorable rates in the four clusters are relatively consistent,so it can be considered that in the departmental clustering,these three indicators are weighted and combined into a new comprehensive indicator before clustering.