FORMATION OF A UNIFIED DATA SET FOR ANALYZING THE RELATIONSHIP BETWEEN DOG BREEDS AND DISEASES

Abstract

This article presents an approach to creating a dataset containing information about dog breeds, their classification in accordance with international standards, and hereditary diseases associated with these breeds. The main objective of the study was to integrate disparate sources of information and standardize breed and disease names, which enabled further comprehensive analysis. Official databases and specialized resources were used to form the final dataset: iDog, Canine Inherited Disorders Database (CIDD), The Kennel Club (UK), and American Kennel Club (AKC). The developed standardization algorithm, based on natural language processing methods, made it possible to eliminate inconsistencies in names, ensure data comparability, and form a structured dataset including 4,093 records. As part of the study, exploratory data analysis (EDA) was conducted to identify patterns in the distribution of breeds by size, classification groups, and disease types. The analysis showed that large and small breeds are characterized by a wider range of hereditary pathologies compared to medium-sized breeds. In addition, differences in the structure of diseases between The Kennel Club and American Kennel Club classifications were identified, which emphasizes the importance of taking national characteristics into account in international comparative studies. The results of the study demonstrate the significance of the developed dataset as a basis for building predictive models and decision support tools in veterinary medicine. The resulting dataset can be used to more accurately predict dogs' predisposition to diseases and develop effective prevention and treatment strategies.

Keywords: Сreating a dataset, data standardization, Exploratory Data Analysis, NLP