In the field of political economy, scholars generally use mathematical and statistical models to analyze data and to test hypotheses about the results. My research intends to analyze the impact of leaders on economic growth by using leader data from the Archigos (Goemans et al. 2009) database from 1835 to the end of 2015 and Maddison Project database. Archigos database is widely used in different areas, scholars use this dataset to analyze the relationship between national leaders and economic growth. The data in my research is modelled by the AutoGluon model developed by Amazon through machine learning. AutoML and AutoGluon automatically extract features from the data and then uses multiple classifiers to train the data. I will use models to evaluate the impact of leaders on economic growth. And I receive the following three conclusions: 1). Machine learning model is very sensitive with numeric values, so we must deal with our data very carefully; 2). The results of classification are better than linear regression, they are large differences between groups, small differences within groups; 3). After optimization, machine learning methods can extract features very efficiently and give statisticians and economists an intuition with AI application in this area.