Over the past century, people have suffered from depression more frequently due to changes in lifestyles. Many cases of mental illness remain undiagnosed, even though the rates of diagnosis have improved in recent years It can be helpful to identify individuals at risk of depression or depressed using automated detection methods. In order to understand depression detection, it is necessary to represent and analyze features in language. Detecting depression using text classifiers is the topic of this article. Hybrid and ensemble methods are examined and compared with the aim of improving depression detection performance. Compared to hybrid models, ensemble models perform better in classification.Multiplying features and selecting the most appropriate features can result in enhanced performance.