Detection of Anxiety Levels in Urdu Text (Multiclass Dataset)

This dataset comprises 24,000 Urdu social media posts that have been manually annotated for anxiety detection. Each post is labeled using a binary scheme: 0 for No Anxiety and 1 for Anxiety (including mild, moderate, severe, and panic). The texts were preprocessed to remove URLs, emojis, and extra white spaces to ensure clean input for research. This dataset provides a valuable resource for studying anxiety patterns in Urdu text and can be used for training and evaluating machine learning and deep learning models for text-based mental health analysis. The authors of this dataset are Sadia Fareed, Dr. Syed Ali Raza, Jawariya Fayyaz, Fareeha Shakoor, and Miara Kahloon, all of whom are affiliated with Government College University, Lahore, Pakistan.