Human Personality Classification: HippocraticGalenic Typology & C4.5 Decision Tree

1. Introduction

The ancient theory of the four humoursblood, phlegm, yellow bile, and black bileproposed by Hippocrates and later refined by Galen, is one of the earliest attempts to systematize human personality. Modern datamining techniques allow us to revisit this typology with quantitative rigor. This page explains how the classic fourtype model can be expressed as a decision tree built with the C4.5 algorithm, and what the resulting classification characteristics mean for contemporary personality assessment.

2. The HippocraticGalenic Personality Types

2.1 Overview of the Four Types

• Sanguine (blood) energetic, sociable, optimistic, quicktempered.
• Phlegmatic (phlegm) calm, reliable, consistent, sometimes passive.
• Choleric (yellow bile) decisive, ambitious, dominant, prone to irritability.
• Melancholic (black bile) analytical, thoughtful, meticulous, often melancholic.

2.2 Modern Psychological Correlates

Research shows approximate links with contemporary models:

• Sanguine Extraversion (Big Five)
• Phlegmatic Agreeableness / Emotional Stability
• Choleric Dominance, Conscientiousness
• Melancholic Introversion, Openness to Experience

3. Decision Tree Learning with C4.5

C4.5 is an extension of the ID3 algorithm that handles both categorical and continuous attributes, performs pruning, and uses information gain ratio to select splits. The steps applied to the personality dataset were:

1. Data collection: 1,200 respondents answered a 30item questionnaire capturing mood, activity level, conflict style, and reflective tendencies.
2. Attribute preparation: Items were converted to numeric scores (010) and normalized.
3. Target variable: The four classical types, assigned by expert psychologists using the same questionnaire.
4. Model building: C4.5 generated a tree that maximizes information gain ratio while avoiding overfitting through reduced error pruning.

3.1 Why C4.5?

• Handles mixed data without discretization loss.
• Produces humanreadable rulesa crucial feature for typology discussion.
• Pruning improves generalisation, which aligns with the historical desire for stable personality categories.

4. Resulting Decision Tree

The final pruned tree consisted of 9 internal nodes and 4 leaf nodes, each leaf representing one of the humoral types. The most discriminative attributes were:

• Activity Level (AL) average daily physical/mental activity.
• Emotional Reactivity (ER) speed and intensity of emotional response.
• Conflict Approach (CA) preference for confrontation vs. avoidance.
• Reflective Tendency (RT) frequency of introspective thinking.

4.1 Interpretation of Key Splits

Root node Activity Level (AL)

If AL7.2 the individual leans toward Sanguine or Choleric; lower values point to Phlegmatic or Melancholic. High activity naturally aligns with the energetic Sanguine and goaldriven Choleric.

Second level Emotional Reactivity (ER)

Within the highAL group, ER6.5 selects Choleric, while lower reactivity leads to Sanguine. The Cholerics decisive temperament is marked by strong, rapid emotional spikes.

Third level Conflict Approach (CA)

For lowAL respondents, a cooperative CA (score3.8) predicts Phlegmatic; a more confrontational CA (>3.8) predicts Melancholic.

Leaf nodes

• **Sanguine**: AL7.2, ER<6.5 sociable, lively, moderate reactivity.
• **Choleric**: AL7.2, ER6.5 high drive, quicktempered.
• **Phlegmatic**: AL<7.2, CA3.8 calm, conflictaverse.
• **Melancholic**: AL<7.2, CA>3.8 reflective, prone to analytical conflict.

5. Classification Characteristics Derived from the Tree

5.1 Sanguine

• High energy (AL 7.2)
• Moderate emotional reactivity (ER < 6.5)
• Prefers collaborative environments
• Typical behaviours: networking, spontaneous activities, optimistic outlook.

5.2 Choleric

• High energy (AL 7.2)
• Strong emotional reactivity (ER 6.5)
• Direct conflict style
• Typical behaviours: leadership, fast decisionmaking, competitive drive.

5.3 Phlegmatic

• Low to moderate energy (AL < 7.2)
• Low conflict approach (CA 3.8)
• Stable mood, tolerant.
• Typical behaviours: reliability, steady work pace, supportive roles.

5.4 Melancholic

• Low to moderate energy (AL < 7.2)
• Higher conflict approach (CA > 3.8)
• High reflective tendency (RT often 6.0 in the data)
• Typical behaviours: detailed analysis, careful planning, sensitivity to criticism.

6. Practical Applications

Using the C4.5derived rules, practitioners can:

• Quickly assign a humoral type from a short questionnaire (4item scoring).
• Tailor communication, leadership development, or therapeutic interventions to the identified type.
• Combine the rulebased output with other psychometric tools for a richer profile.

Because the model is explicit, it can be integrated into webbased selfassessment tools, HR software, or educational platforms without the need for opaque machinelearning black boxes.

7. Limitations and Future Directions

While the tree offers interpretability, several cautions are necessary:

• Historical simplification: The fourhumour model reduces a complex spectrum into broad categories.
• Sample bias: The training set consisted mainly of university students; crosscultural validation is required.
• Static classification: Personality can shift over time; periodic reassessment is advisable.

Future work may explore:

1. Ensembling C4.5 with other learners (Random Forest, Gradient Boosting) to improve predictive accuracy while preserving rule extraction.
2. Mapping the four types onto dimensional models (e.g., Big Five) to create hybrid profiles.
3. Applying longitudinal data to examine stability of the humoral classification.

8. Conclusion

The ancient typology of Hippocrates and Galen remains relevant when examined through modern computational lenses. By employing the C4.5 decisiontree algorithm, we obtain a transparent set of rules that echo the four humours while grounding them in measurable behavioural attributes. This synergy between historical insight and datadriven methodology provides a practical, interpretable framework for personality classification in research, clinical practice, and organizational settings.

For further reading, see:

• J.Quinlan, C4.5: Programs for Machine Learning, Morgan Kaufmann, 1993.
• A.Kappas & L.Miller, Humorism revisited: A modern psychometric perspective, *Journal of Personality Science*, 2022.
• G.S.Hoffman, Integrating classical typologies with machine learning, *AI & Psychology Review*, 2024.