This paper aims to understand how people suffering from depression interact with the online world and how it is different from those not afflicted with this mental illness. It focuses on how people interact, react with various communities, and how they in turn convey information. These texts and its authors are used to extract strategic and novel set of features to be fed into machine learning models. Various studies have tackled uncovering depression symptoms from a single post, using various features extracted from the text. This paper aims to understand if it is possible to uncover if a person suffers from depression from their history of posts that need not be the posts dealing explicitly with sadness, or such stereotypical depressive topics. The analysis of results in this paper shows promising results with users being classified with depression with a high accuracy.