Analysis of Suicidal Tweets from Twitter Using Ensemble Machine Learning Methods

Suicidal ideation is basically to think about taking one’s own life and is related to mental health problems, but people with mental disorders, insecurity, stress and a sense of losing control receive no treatment for their condition due to the limited accessibility to mental health care facilities, awareness or social support. Social media like Facebook, Twitter, Reddit, Instagram is our most preferable area to express our feelings, stress and gratitude. The main challenge is to prevent suicidal cases and detect a suicidal note from one’s status, or tweet which will help to provide proper mental support to that person. The main motive of this paper is to anticipate whether a person’s tweet contains suicidal ideation or not with the help of machine learning. To attain the objectives, we have used an accurate ensemble classifier that can identify content on Twitter that may potentially hint towards suicidal activity. In this research, we have also used several sets of word embedding and tweet features, and we have compared our model among twelve classifiers models. The results from our study reflect that our proposed model can accurately predict the target outcome and come up with an excellent standard for suicidal ideation prediction on active social media such as Twitter.