Yes. The dataset splits naturally into three groups by `Overall_Impact` (Beneficial, Neutral, Negative). These groups differ sharply on almost every measure, and the differences run in one consistent direction. | Segment | Share of students | Daily hours | Sleep (hrs) | Stress | Mental health | GPA | Late-night use | |---|---|---|---|---|---|---|---| | Beneficial | 81.8% (3,681) | 4.42 | 7.03 | 11.16 | 84.26 | 3.55 | 53.5% | | Neutral | 14.5% (654) | 8.50 | 5.48 | 22.30 | 63.42 | 3.00 | 82.3% | | Negative | 3.7% (165) | 12.25 | 4.23 | 31.12 | 47.13 | 2.54 | 81.8% | The rendered result blocks above show these figures per segment. - **Usage is the main dividing line.** Average daily use rises from about 4.4 hours (Beneficial) to 8.5 (Neutral) to 12.3 (Negative). Sleep, sleep quality, mental health and GPA fall along that same ordering, and stress rises. - **Weekend extra hours do not separate the groups.** They are nearly identical across segments (about 1.7–1.8 hours), so weekday and overall volume matter more than weekend extra. - **Late-night use marks a threshold.** About 82% of Neutral and Negative students use social media late at night, compared with about 54% of Beneficial students. The two harmed groups look alike on this measure, so it separates "Beneficial" from "not Beneficial" but not Neutral from Negative. - **The segments are very unequal in size.** The Negative group is only 165 students, so any further breakdown of it (by platform, age or similar) will rest on small counts. **Caveats** - `Overall_Impact` looks like an outcome label rather than an independent attribute, since it tracks the measures so closely. It is a good way to describe the groups, but it is not a cause. - This is observational data, so these are associations only. - I did not break the segments down by age, gender, academic level, platform or device. Those are other possible groupings, and there are pre-aggregated tables for them if you want to see who falls into the Neutral and Negative groups. - If you want a segmentation that does not depend on the label, daily usage hours combined with late-night use is the most natural basis, given how cleanly they track the outcome groups.