A New Perspective on Türkiye\'s Sheep Population: Classification with Decision Trees

Authors : Çağrı Kandemir, Çiğdem Takma, Turgay Taşkın

Pages : 966-979

Doi:10.33462/jotaf.1390307

View : 94 | Download : 103

Publication Date : 2024-09-20

Article Type : Research

Abstract :Decision trees are data mining algorithms that make interpreting the cause-effect relationship or classification between features with visual diagrams easy and do not require parametric assumptions. The aim of this study is to classify different sheep breeds raised in Turkey according to their origin and tail structure characteristics using a decision tree algorithm. It has been seen that the CART (Classification and Regression Trees) algorithm is sufficient for the classification of sheep breeds by obtaining low-risk values. According to the results of the study, it is seen that domestic sheep breeds are distributed to Eastern Anatolia, Mediterranean, Black Sea, and Central Anatolia Regions and imported breeds are distributed to all regions of Türkiye. The total population of these breeds was determined as 43.889.918 heads. As it can be understood, the Akkaraman breed ranks first with a rate of 40.88% of the total sheep stock. Next comes the Morkaraman breed with a rate of 11.68%. Thirdly, the most cultivated breed is Kıvırcık, whose rate is 8.82%. Merino sheep comes in fourth place with 8.43%. Awassi sheep are among the most common breeds in the fifth place. Hereby the distribution of sheep breeds in Türkiye has also been revealed and the distribution of breeds according to environmental conditions has been determined. The decision tree model generated using the CART algorithm for the distribution of sheep breeds in Turkey based on tail structures has been found to have a value (0.47). The proportion of sheep with a thin tail structure is 52.2%, while those with a semi-fatty tail structure are 17.5%, and those with a fatty tail structure are 30.2%. It has been determined that sheep breeds with thin tails are more commonly raised in Turkey. In the study, a decision tree model was also created using the CART algorithm to analyze the distribution of sheep populations in regions based on their origin, and it was found to have a value (0.26) with an accuracy rate of 74%. According to the research, 73.9% of sheep are domestic, while 26.1% are imported. When examining the distribution of origin by region, the region with the least number of imported sheep is Eastern Anatolia, while the highest proportions are found in the Mediterranean and Black Sea regions. The region with the lowest proportion of domestic sheep is the Mediterranean and Black Sea (70.4%), whereas the highest proportion is in the Eastern Anatolia region (82.6%).
Keywords : Koyun sayısı, Sınıflandırma, Karar ağacı, Koyun ırkları