An artificial intelligence approach to predicting personality types in dogs

Area of Science:

• Veterinary Behavioral Science
• Artificial Intelligence in Animal Behavior
• Machine Learning Applications

Background:

• Canine personality influences human-dog relationships and working roles.
• Previous canine personality research relied on traditional statistics.
• AI has not been extensively applied to canine personality classification.

Purpose of the Study:

• To apply machine learning techniques for classifying canine personality types.
• To develop a novel AI-based methodology for canine personality assessment.
• To improve the selection and training of dogs for various roles.

Main Methods:

• Utilized unsupervised learning (K-Means) for initial data clustering and labeling.
• Applied feature importance analysis to define personality clusters.
• Trained and evaluated four machine learning models (SVM, KNN, Naïve Bayes, Decision Tree) for personality prediction.

Main Results:

• Identified five distinct canine personality types: "Excitable/Hyperattached", "Anxious/Fearful", "Aloof/Predatory", "Reactive/Assertive", and "Calm/Agreeable".
• The Decision Tree model achieved 99% accuracy in predicting personality traits.
• Feature importance analysis successfully defined the characteristics of each personality cluster.

Conclusions:

• A novel AI-based approach effectively classifies canine personality using behavioral data.
• This methodology offers potential for enhanced dog selection and training.
• Machine learning provides a powerful tool for understanding canine behavior and personality.