Menu

[Synthetic] [Self-Promotion] Screen time, sleep & mental health: 15,300 people, one row each, Low/Moderate/High stress target (CC BY 4.0)

#datasets#redditdatasets#time#sleep#health
Reading 0:00
15s threshold

Hi everyone, I just published a new dataset exploring the intersection of daily screen time, sleep habits, and mental health.

Full disclosure upfront: This is 100% synthetic data generated using a tool I built (GenerateRealData). There is no real-world PII, and no actual patient surveys are included.

Why it might be useful:

Ready for Multi-Class ML: It features a clean Low, Moderate, or High stress target variable, making it easy to drop into a classification project. A quick gradient-boosting model gets about 84% accuracy on held-out people.

Logically Consistent: Unlike standard random mock data, the statistical relationships between heavy screen time, sleep deprivation, and stress are built into the generation rules, so correlations and EDA show sensible patterns. For example, 95% of people averaging under 6.5 h of sleep land in High stress, vs 20% of those over 7.5 h.

Clean & Flat: 15,300 rows with exactly one row per person. No missing values and no complex SQL joins required to start analyzing.…

Continue reading — create a free account

Join HashtagPLUS to read full articles, follow hashtags, vote, and join the conversation.