Credit Rating System And City Pattern Mining Based On Volunteer Big Data

With the vigorous development of voluntary service activities all over the country,more and more individuals and groups have devoted themselves to the field of voluntary service,and at the same time,they have accumulated a large amount of big data of voluntary service.Since the national voluntary service information system was launched in 2008,by the end of February 2019,more than 120 million volunteers,730,000 volunteer groups and 2.32 million volunteer projects have been accumulated in 31 provincial administrative regions.Under the background of the fast development of big data and Al technologies,how to analyze and mine the big data of volunteer service through machine learning related algorithms and help the subsequent development of volunteer service activities is an urgent challenge.With the rapid development of smart city,the analysis and research results of volunteer service also make difference in guiding the city’s credit reporting,risk control and service.For example,in the recruitment of volunteers for major events such as the Winter Olympics and Entrepreneurs Forums in the city,indicators such as volunteer creditability will help the organizers to screen candidates.And in the case of massive data complexity,the current field of volunteer service is lack of visual management platform,so it is unable to make an intuitive and clear display of the development status of volunteer service.This thesis will build the credit and risk control system of volunteer city through the credit rating model of volunteers,excavate city volunteer pattern from the perspective of volunteer teams and volunteer projects,and then build the activity model of volunteer city.At the same time,we will build a big data visualization platform for volunteers and volunteer cities to help government managers manage and analyze volunteer service information.The main contents of this thesis are as follows.a)According to the credit influencing factors of the volunteer data set,data cleaning and feature engineering are carried out,and the credit rating system of volunteers is established.The score card model is constructed according to the logistic regression algorithm and weight of evidence algorithm.The output of the rating model is standardized to 0-300 by using the transformation of linear function,the correlation between each influence factor and the prediction result was judged by Information Value.The results show that area under curve is 0.9651,KS value is 0.8521,accuracy and partition ability are in line with expectations.Secondly,a multi-layer neural network is designed by using neural network,and a volunteer credit classification model is established by using cross entropy as loss function.Finally,for the accuracy index,the results of the score card model and neural network model are 0.8995 and 0.6667 respectively;for the precision index,the two models are 0.9195 and 0.7253;for the recall index,the two models are 0.9159 and 0.7336 respectively.The results of score card model are better than those of neural network model.b)Based on the city volunteer service information data,the pattern mining of volunteer city is carried out,including the pattern research of volunteer groups and projects,and the analysis of the activity of volunteer city.Firstly,for the volunteer groups in the city,the hierarchical clustering method is used to cluster the influence factors after min-max normalization,the silhouette coefficient reaches 0.9655,and 12 kinds of volunteer groups with different attributes are obtained.Besides,for the volunteer projects in the city,the inverse geocoding of AMAP is used to mine the POIs information.This thesis analyzes the city volunteer projects in time and space filed,and explores the differences of city voluntary projects in the dimension of time and space.Besides,TF-IDF normalization is carried out on POIs of voluntary projects,and clustering analysis is carried out on the transformed data feature vectors.Finally,the thesis uses analytic hierarchy process algorithm to construct the activity model of voluntary city,and gets the relationship between city activity and permanent resident population.c)Based on the research results of volunteer credibility and volunteer city service pattern,the Volunteer Credibility Visual Platform and Volunteer City Service Visualization Platform are built.Using the system processing framework of spring MVC,the database on the server is connected through the framework to complete the changes of the volunteer service related data.At the same time,using Bootstrap’s front-end application service architecture and Asynchronous JavaScript And XML page refresh technology,the platform’s web page shows the visual model and view.Finally,functional tests are carried out for the two visual platforms,and the platforms passes the test cases.