Suicidal Ideation Prediction Using Machine Learning

In the present world, accidents and health {complications account for the majority of fatalities. The majority of deaths after accidents are caused by suicide due to depression and natural catastrophes. The widespread use of the Internet has given people a new means of communicating their feelings. It is also a platform with a massive amount of content where users may read other users’ opinions, which are divided into several sentiment groups and are becoming more and more important in decision-making process. This paper contributes to the classification that is useful to examine the data in the form of the quantity of tweets where comments are extremely unstructured and either negative or positive or somewhere in between these two. To do this, we first pre-processed the data, then extracted the adjectives with meaning from the tweet, and last utilised machine learning-base classification methods, specifically. When compared to the current system, the accuracy of the TFIDF, N-gram, and LinearSVC algorithms for suicide prediction with tweets including suicidal thoughts was improved to 95 percent. Such testing and observation may help in both individual and population-wide prevention. By establishing a baseline for suicide identification on online social networks, such as Twitter, the experimental work suggests the viability of the approach adopted. In the end, we evaluated the classifier’s performance in terms of accuracy.