The Research On The Factors And Prediction Models Of College Students’ Entrepreneurial Success From The Perspective Of Information Ecology

Under the background of “mass entrepreneurship and innovation”,the government has introduced corresponding entrepreneurial policies including taxation,financing,administrative assistance as the guarantee for college students to start their own businesses.However,as the main force of entrepreneurship,the entrepreneurial success rate of college students continues to decline due to the asymmetry of entrepreneurial information.In this context,it is crucial to improve the success rate of entrepreneurship and maintain the balance between the overall factors of the information ecology.Therefore,analyzing the success factors of entrepreneurship from the perspective of information ecology and constructing the ecological prediction model of entrepreneurial information is conducive to analysing the incentive effect of information ecological factors on the transition of college students’ entrepreneurial upgrading,so that the information factors in the whole entrepreneurial ecosystem can timely adjust the information niche and effectively promote endogenous economic growth.Firstly,this thesis sorts out relevant theories in the field of information ecology,college students’ entrepreneurship and machine learning,and defines the concepts of college students,college students’ entrepreneurship and the measurement standards of college students’ entrepreneurial success.By combining literature review and expert interview,the factors that affect the entrepreneurial success of college students are divided into three levels in terms of information person,information environment and information resources.The index system and conceptual model of college students’ entrepreneurial success factors are preliminarily constructed and proposed research hypothesis.Then,pre-survey and formal survey were conducted by means of questionnaire survey.Reliability and validity analysis,factor analysis and path analysis were carried out on the collected questionnaire data with SPSS17.0 and AMOS21.0 respectively to verify the validity of the collected data and the rationality of the hypothesis proposed.On the basis of using Python3.7 successively set up random forests and BP neural network prediction model,after parameter tuning and training,respectively from the perspective of discriminant ability and generalization ability,two prediction model performance optimization analysis,found that the random forest model can be more sensitive to select the entrepreneurial success and failure of the project and to ensure that the accuracy is above 80%,with a final optimized random forest model as the prediction model in this thesis,we study college students’ entrepreneurial success.Finally,according to the prediction results of the model,countermeasures and suggestions are put forward from the three levels of individual micro,medium view of school and national macro view for the stakeholders,as the information environment and information providers,are available for reference.From the perspective of information ecology,the research in this thesis not only extends the research scope of entrepreneurship and information ecology,but also promotes the collaborative development of entrepreneurship information ecological population(government,start-up enterprises,universities and research institutes,entrepreneurship and innovation organizations and investment institutions,etc)to a certain extent,and promotes the information of a virtuous cycle in the whole entrepreneurship process.At the same time,the entrepreneurial success prediction model constructed can help college students entrepreneurs and investors to conduct more efficient project selection and preparation,risk control and improve the operational efficiency of entrepreneurial projects.