Prediction of the Success of Startup Companies Based on Support Vector Machine and Random Forset

Startup companies are a huge driving force for economic development, and the success of these high-risk companies can bring huge returns to venture capital companies. The ability to predict the success of startups is a great advantage for investors to overtake their competitors. With the development of information technology, highly reliable results can be obtained by using complex machine learning algorithms or data mining. Data were acquired from the world's largest structured database for startup companies. These data included nearly 40 characteristics of almost 22,000 companies. The main purpose of this research is to establish a model to classify start-ups and analyze the important features for the success of startup companies. Random Forest and Support Vector Machine were used to explore the important features that determine the success of startup companies and explain some of the features. It also compared the effects of different machine learning methods, illustrating the effectiveness of machine learning methods for investors to predict the success of startups. These analyses will provide investors and venture capital companies with effective methods, reduce their large manpower input for prediction, and improve the efficiency of their analysis of startup companies.