Predicting Suicidal Ideation Via Social Media

• Main Author: Boon, Kar Lih
• Format: Undergraduates Project Papers
• Language: English
• Published: 2023
• Subjects: QA75 Electronic computers. Computer science
• Online Access: http://umpir.ump.edu.my/id/eprint/40173/1/CA19046.pdf; http://umpir.ump.edu.my/id/eprint/40173/

Nowadays, suicide is one of the leading causes of mortality worldwide, with over 800,000 people dying by suicide each year. Suicidal ideation is a contemplations and preoccupations about suicide. Most of the people who got suicidal ideation are active in social media and send out signs about their intentions. However, an accurate classifier can identify the data which may potentially hint towards suicidal ideation. The aim of this research is to study the suicidal ideation via Subreddits on Reddit dataset. The dataset is collected from Kaggle websites which dataset is collected from 2008 until 2021. The model will predict if the individual has suicidal or non-suicidal ideation based on the dataset. Three Machine Learning algorithms are implemented to predict the result and outcome which are Support Vector Machine (SVM), Decision Tree and Naive Bayes (NB). SVM give the most precise result for accuracy with 93.40% among the others. It can accurately predict the suicidal ideation via social media data.