Depression has gradually become one of the most common mental illnesses in modern society. According to the 2023 WHO report, about 280 million people worldwide suffer from the depression. Since social media has become a major platform for expressing emotions, many users often share their feelings there. This large amount of data offers opportunities to detect users' mental states. This study proposes using social media posts as data source to build a depression prediction model to explore the users' mental states. Our goal is to provide doctors with another tool to help diagnose depression.