The progression of social networking and online communication services lay down a podium to disclose one’s sufferings and feelings in the real world. Suicidal ideation sufferers usually share their opinions and thoughts on social media. Several studies uncovered that contemplation of suicide could be identified by analyzing social networking posts. To tackle the challenge of finding and comprehending patterns in suicidal ideation, creating a machine learning system to automatically identify suicidal thoughts or sudden behavioral shifts in users by examining their social media posts is crucial. Our approach in this work relies on experimental research and uses word-embedding techniques like TF-IDF and TF-IDF N-GRAM utilizing Word and Character Analyzer to construct a Suicide Thought Detection System. The proposed system uses publicly available Reddit datasets and employs machine learning algorithms (XGBoost and CATBoost) for classification. As the feature limit is restricted to 1500 features, TF-IDF NGram didn’t perform well. To overcome this issue, TF-IDF word analyzer can be chosen for actual usage and testing.