---
language:
- en
license: mit
task_categories:
- text-classification
task_ids:
- regression
tags:
- burnout
- android
- telemetry
- stress-detection
- keyboard-dynamics
pretty_name: Synthetic Burnout Telemetry
size_categories:
- n<1K
---

# FRIDAY Synthetic Burnout Telemetry

Synthetic Android telemetry dataset for training a lightweight burnout / urgency
regression model (e.g. RoBERTa fine-tune).  Each row represents a single
keyboard + notification event captured on a simulated Android device.

## Dataset structure

| Split      | Rows |
|------------|------|
| train      | 500  |
| validation |  62  |
| test       |  63  |

## Fields

| Field        | Type    | Description                                              |
|--------------|---------|----------------------------------------------------------|
| `input_text` | string  | Serialised signal string fed directly to the tokeniser   |
| `label`      | float32 | Burnout / urgency score in **[0, 1]**                    |
| `app`        | string  | Source application (Slack, Gmail, WhatsApp, Teams, System) |
| `wpm`        | int32   | Typing speed in words per minute                         |
| `backspaces` | int32   | Correction / backspace count in the session              |
| `hour`       | int32   | Hour-of-day the event was captured (0–23)                |
| `session_min`| int32   | Continuous screen-on duration in minutes                 |
| `notif_count`| int32   | Pending notifications at event time                      |
| `text`       | string  | Raw message or notification body                         |



## Citation

```bibtex
@misc{friday_burnout_2026,
  title  = {FRIDAY Synthetic Burnout Telemetry},
  author = {Your Name},
  year   = {2026},
  note   = {Synthetic dataset for mobile stress detection research}
}
```