README.md

---
dataset_info:
  features:
  - name: text
    dtype: string
  - name: label
    dtype: string
  splits:
  - name: train
    num_bytes: 105750984.56504539
    num_examples: 152946
  - name: test
    num_bytes: 22660826.48866134
    num_examples: 32774
  - name: val
    num_bytes: 22661517.91560003
    num_examples: 32775
  download_size: 65442094
  dataset_size: 151073328.96930677
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
  - split: test
    path: data/test-*
  - split: val
    path: data/val-*
---

## Dataset Card for "vibhorag101/suicide_prediction_dataset_phr"
- The dataset is sourced from Reddit and is available on [Kaggle](https://www.kaggle.com/datasets/nikhileswarkomati/suicide-watch).
- The dataset contains text with binary labels for suicide or non-suicide. 
- The dataset was cleaned minimally, as BERT depends on contextually sensitive information, which can worsely effect its performance.
  - Removed numbers
  - Removed URLs, Emojis, and accented characters.
  - Remove any extra white spaces and any extra spaces after a single space.
  - Removed any consecutive characters repeated more than 3 times.
  - The rows with more than 512 BERT Tokens were removed, as they exceeded BERT's max token limit.
- The cleaned dataset can be found [here](https://huggingface.co/datasets/vibhorag101/phr_suicide_prediction_dataset_clean_light)   
- The evaluation set had ~33k samples, while the training set had ~153k samples, i.e., a 70:15:15 (train:test:val) split.