Table 5. Association rules generated by association rule mining.

Obese rules
(Rule 1) Prevotella = {4}, Faecalibacterium = {1}, Oscillibacter = {1} → Obese (0.86)
(Rule 2) Bacteroides = {3}, Oscillibacter = {1} → Obese (0.86)
(Rule 3) Prevotella = {2}, Faecalibacterium = {1}, Oscillibacter = {1} → Obese (0.83)
(Rule 4) Bacteroides = {1}, Alistipes = {1}, Faecalibacterium = {2} → Obese (0.80)
(Rule 5) Bacteroides = {2}, Faecalibacterium = {1}, Oscillibacter = {1} → Obese (0.80)
Normal rules
(Rule 6) Bacteroides = {4}, Prevotella = {1} → Normal (0.95)
(Rule 7) Bacteroides = {4}, Oscillibacter = {4} → Normal (0.89)
(Rule 8) Bacteroides = {3}, Faecalibacterium = {4}, Oscillibacter = {4} → Normal (0.86)
(Rule 9) Bacteroides = {4}, Oscillibacter = {2} → Normal (0.86)
(Rule 10) Bacteroides = {3}, Oscillibacter = {4} → Normal (0.85)
(Rule 11) Bacteroides = {2}, Faecalibacterium = {1}, Oscillibacter = {4} → Normal (0.80)
Rule explanation (Rule 1)
Original rule	Prevotella = {4}, Faecalibacterium = {1}, Oscillibacter = {1} → Obese (0.86)
Rule explanation	If the compositions of Prevotella > 66, Faecalibacterium < 0.7 and Oscillibacter < 0.1; the sample is obese with 86% accuracy
Applied category for rule generation
Category	Bacteroides	Alistipes	Prevotella	Faecalibacterium	Oscillibacter
1	<4	<0.04	0	<0.7	<0.1
2	4–36	0.04–0.5	0–0.1	0.7–2	0.1–0.3
3	36–60	0.5–2	0.1–66	2–7	0.3–1
4	>60	>2	>66	>7	>1