Machine Learning Model for Sentiment Analysis on Mental Health Issues

Abstract

Social media study concerning mental health has increasingly piqued researchers' interest. One of the most well-known sites where users can express their ideas, feelings, and opinions is Reddit. Research statistics on important topics are available on social media. People's mental health is one of the main issues because it is a new area of interest. Different issues, including anxiety, tension, anger, and despair, can be brought on by mental disorders. In recent years, people appear to be busy and have less time to contact one another. Instead, they seem to prefer to engage in online discussion forums. Data has been collected from social networking sites like Reddit, and 10,000 posts were aggregated for investigating the posts among suicidal and non-suicidal. Machine learning algorithms such as Support Vector Machine, Logistic Regression, and Multinomial Navïe Bayes segregated the posts into two classes. Pre-processing of data is done which is the vital step in text analysis, which includes tokenization, removal of stop words and special characters, etc. The performance in terms of accuracy and precision Logistic Regression outperforms the other algorithms. Multinomial Naive Bayes yields great recall. The study also signifies the current research trends and proffers an overview of the researcher’s accomplishments in the related fields.