中国社会幸福感影响因素的数据探索性分析

北京理工大学珠海学院2020届本科生毕业论文An Exploratory Analysis of the Influencing Factors of ChineseSocial HappinessAbstractIn this paper,the questionnaire data from the cross-sectional interview of multi-order strat-ified sampling of China comprehensive social survey(CGSS)project is used as the originaldata set.Selection in the multiple sets of variables and the happiness evaluation indexof high correlation main variables to do analysis,include gender,age,geographic,career,health,marriage,and politics,and so on individual factors,parents,spouse,children,fam-ily,family factors of capital and so on,and justice,credit and social attitudes of the publicservice and many other factors,through the visual analysis of the data packet for the eval-uation of happiness.Firstly,feature engineering is done to understand the data.In the datapreprocessing,the data is cleaned and integrated,and the data format is processed.In thefeature transformation,the useful data is screened to reduce the data volume and reducethe data dimension.Then,the overall situation of the data was observed and analyzed oneby one according to gender,urban and rural area,age group,reasonable degree of income,socio-economic status,and degree of subjective social equity to explore the factors relatedto the evaluation of happiness.Then based on the machine learning method of XGBoostlimit gradient promotion algorithm model and neural network sequential model respective-ly,the happiness evaluation and prediction model was constructed.Finally,after manytimes experimenting,by solving the problems of sample disequilibrium in feature engineer-ing and further proposes the optimization scheme of model fusion,the final reflection madeimprovement of the fitting performance of the prediction model.Keywords:Happiness influence factor analysis Happiness evaluation prediction modelEnsemble leaming Sequential model