It is of great significance to study general risk preference for different populations to take targeted interventions to reduce potentially risky behaviors. In this paper, the results of Balloon Analogue Risk Task (BART) experiment are analyzed by single factor analysis of variance. Through this analysis, it is found that there are significant differences in people's general risk preference in different situations. Then, This paper is based on BART experiment and risk propensity questions and Frequency of General Risk Propensity Scale (GRiPS), Socio-Economic Panel (SOEP) Results of three questionnaires on Risky Behaviours (FoRB) were used to measure general risk preference using exploratory factor analysis. Then, cluster analysis is used to explore the performance of different risk preferences under different characteristics. The empirical analysis shows that the group with high risk preference has larger data on the two factors of GRiPS and SOEP, while the group with moderate risk preference has more balanced values on the four factors of BART, GRiPS, SOEP and FoRB. The risk-averse group has lower scores on GRiPS, SOEP and FoRB. According to the above conclusions, for specific things, relevant targeted measures can be developed to minimize the impact of potential risks or injuries on people's personal safety and property safety.