The two samples of 15 rows each are rendered above. One is a spread of Student_IDs and the other is the first 15 IDs in the table, so together they show the range of the data. They are small samples, so the patterns below are only illustrative. **What each row is:** one student, with demographics (age, gender, academic level), usage habits (platform, daily and weekend hours, device, late-night use, social comparison frequency), wellbeing measures (sleep hours and quality, stress score, mental health index, GPA) and an `Overall_Impact` label. **What the samples show:** - **Range of students:** ages run from 15 to 23, covering High School, Undergraduate and Postgraduate, with Male, Female and Non-Binary students. Platforms include TikTok, YouTube, Instagram, Snapchat, Reddit, X (Twitter) and LinkedIn. Smartphones dominate, with a few Tablet and Laptop/PC rows. - **Label mix:** in the first sample, 14 of 15 rows are "Beneficial" and 1 is "Negative". In the second, 11 are "Beneficial" and 4 are "Neutral". "Beneficial" is the most common label in both, though two samples this small don't give reliable proportions. - **Heavier use goes with worse outcomes:** the "Negative" row (STU_2025378) uses TikTok 10.4 hours a day, sleeps 4.8 hours, has a sleep quality score of 1, a mental health index of 56, a stress score of 29 and a GPA of 2.37. The "Neutral" rows tend to show heavy use (7.5 to 10.9 hours a day), shorter sleep (about 5 to 6 hours) and lower mental health scores (64 to 73). - **Light use looks healthier:** low-usage students such as STU_2026053 (3.1 hours a day) sleep about 8.9 hours, score 95 on mental health and have a 4.0 GPA. I haven't tested whether these patterns hold across all 4,500 rows. I can run that comparison, for example impact label by usage band, if you want.