Yes. The most natural grouping is by sleep outcome, and the best explanatory segments are chronotype, occupation (shift work), and primary bedtime app. Gender and blue-light filter use barely separate the records. The numbers come from the two segment summaries I ran, which show record count, average sleep hours, sleep latency, fatigue and bedtime phone minutes for each group. **1. Sleep debt category is the clearest grouping, but it is an outcome** - `sleep_debt_category` is calculated from `total_sleep_hours`, so it describes the result rather than explaining it. - It splits the 8,500 records into four tiers that differ sharply: | Category | Records | Avg sleep (h) | Avg latency (min) | Avg fatigue | Avg bedtime phone (min) | |---|---|---|---|---|---| | Optimal Recovery | 1,387 | 8.35 | 28.9 | 1.08 | 33.5 | | Mild Deficit | 2,004 | 7.23 | 34.7 | 1.83 | 45.5 | | Moderate Debt | 4,462 | 5.55 | 42.4 | 4.68 | 64.1 | | Severe Sleep Debt | 647 | 3.76 | 72.7 | 9.58 | 123.6 | - Fatigue, latency and bedtime phone minutes all rise steadily as sleep falls. The Severe tier is a distinct small cluster of 647 records. **2. Chronotype and occupation separate people on sleep** - Chronotype gives three well-separated groups. Morning Larks average 7.38 h of sleep and fatigue of 2.53. Intermediates average 6.44 h and 3.48. Night Owls average 5.01 h and 5.40. - Occupation is mostly one split. Healthcare / Shift Workers (997 records) average 5.02 h of sleep and fatigue of 5.54. The other four occupations sit close together at roughly 6.4 h and fatigue of 3.5–3.7. - Shift workers and Night Owls look like the high-risk segments, since both average about 5 h of sleep. **3. Primary bedtime app separates people on sleep latency** - Latency runs from 33.2 min for News / Reading, to 36.4 for Streaming, to 46.1 for TikTok / Reels. Instagram / Reddit is also high at 42.5. - The app groups differ little on total sleep (6.17–6.38 h), so app mainly tells you about falling-asleep time. **4. Weak segmenters** - Gender gives almost identical groups: Female 6.26 h, Male 6.27 h, Non-Binary 6.31 h, with similar latency and fatigue. The Non-Binary group is small at 248 records. - Blue-light filter use shows a modest gap: 6.33 h, 39.6 min latency and fatigue of 3.58 with the filter, against 6.21 h, 41.6 min and 3.99 without. It is small compared with the chronotype and occupation gaps. **Practical suggestion:** use chronotype × occupation (or a shift-worker flag) as the main grouping, with bedtime app as a second dimension for latency. Keep `sleep_debt_category` as the outcome label to compare the groups against. Average bedtime phone minutes are nearly flat across chronotype, occupation and app (about 57–61), so that behaviour doesn't define these segments. It only differs sharply across the outcome tiers. This is observational data, so these groupings describe association and don't show cause.