Cem Tırınk

Değişkenler arasındaki ilişkiyi matematiksel olarak modelleyen regresyon yöntemleri, gelişen bilim dünyasında istatistiki yöntemlerden biri olarak modernleşmektedir. Bu doğrultuda makine öğrenmesi yöntemleri klasik yöntemlerden ayrılarak daha uygulanabilir analizler sağlayarak karşımıza çıkmaktadırlar. Bu çalışmada 157 adet oğlak verisine ait cidago yüksekliği (CY), sağrı yüksekliği (SY), göğüs derinliği (GD), doğum ağırlığı (DA) ve vücut uzunluğu (VU) değişkenleri kullanılarak canlı ağrlık (CA) tahmini XGBoost metodu kullanılarak analiz edilmiştir. Çoklu doğrusal bağıntının varlığını tespit etmek amacıyla Varyans Şişirme Faktörü analizi sonucunda tüm değişkenler için 5 eşiğinin altında değerler bulunmuştur. Çapraz Doğrulama Kat sayısının, Ortalama Hata Kareleri Kökü üzerine etkisini göstermek amacıyla XGBoost hiperparametre ayarlaması sonucunda 7 katlı çapraz doğrulamanın marjinale ulaştığı belirlenmiştir. Optimum model; eta=0,1, max_depth=5, min_child_weight=1, subsample=0,7, colsample_bytree=0,5, gamma=0 ve best_nrounds=240 parametreleri ile elde edilmiştir. Bu model test verisinde 0,283 RMSE, 0,115 MAE ve 0,988 R² değeri ile yüksek tahmin başarısı göstermiştir. Optimize edilmiş XGBoost modeli için öğrenme eğrisine bakıldığında seçilen modelin iyi dengelenmiş olduğu ve güçlü genelleme yeteneğini koruduğu gözlemlenmiştir. Hangi değişkenlerin önemli olduğunu ve aynı zamanda model tahminlerini nasıl ve hangi yönde etkilediklerini belirlemek amacıyla hem kazanç tabanlı değişken önemi hem de SHAP analizi çıktılarına göre C2 en etkili tahminleyici olarak belirlenmiştir. Elde edilen bulgular, XGBoost algoritmasının oğlaklarda canlı ağırlık tahmininde güvenilir, başarılı ve yorumlanabilir bir yöntem olduğunu göstermektedir.

Light is a significant environmental stimulus affecting embryonic development during incubation, and it is known that different wavelengths can have distinctive effects on incubation performance and post-hatching physiological characteristics. However, the extent to which the traces left by different light colors applied during incubation on the small intestine morphology of chicks can be classified has not yet been sufficiently clarified. Therefore, the aim of this study is to evaluate the classifiability of different light colors (red, white, and green) applied during incubation based on the small intestine morphological characteristics of chicks using the XGBoost algorithm. Fertilized eggs were incubated under red, green, white, or dark conditions. At hatching, the small intestines of chicks were excised and divided into the duodenum, jejunum, and ileum. The villus characteristics in the duodenum, jejunum, and ileum were stained with hematoxylin and eosin and examined using an image analysis programme. Villus height, crypt depth, and morphological ratios based on these, obtained from the duodenum, jejunum, and ileum, were used as classification inputs. The XGBoost model was trained over different hyperparameter combinations, and the best performance was achieved with max_depth = 5, eta = 0.1, and nrounds = 100. Optimized with 5-fold cross-validation, the model achieved 78.95% accuracy in classifying light groups in an independent test set. Furthermore, the macro precision, macro recall, and macro F1-score of the model were 0.7896, 0.7930, and 0.7878, respectively. These findings demonstrate that different light colors applied during incubation create discernible structural changes in the small intestine morphology of chicks, and these differences hold promising potential for classification using machine learning methods. These findings demonstrate that intestinal histomorphology has promising biomarker potential not only for assessing biological effects but also for determining incubation lighting conditions using machine-learning-based approaches. The study reveals that using predictive models to explain the biological effects of incubation lighting can be an important tool for sustainable poultry production.

This study compares the performance of XGBoost and LightGBM models for predicting live weights of Yucatecan Criollo pigs from biometric measurements and examines the structural and algorithmic differences that affect model fit. Detailed analysis of the models’ hyperparameter optimization and variable importance revealed how each model approaches the data and prioritizes features. This study was conducted on 182 Yucatecan Criollo pigs (134 females and 48 males). When model performances were evaluated, the XGBoost model showed superior prediction performance with acceptable accuracy and lower error rates in the test dataset, with R² = 0.905, RMSE = 5.704, and MAE = 3.636. In contrast, the LightGBM model produced acceptable results under certain hyperparameter combinations with R² = 0.824, RMSE = 7.772, and MAE = 5.505. While the robust performance of both models requires strategic decisions in model selection and optimization, it is recommended to consider the dataset’s nature in feature selection and hyperparameter settings. This study provides important insights for simplifying the model and improving its efficiency in machine learning applications, and serves as a reference for more effective model use.

Reliable and cost-effective live weight determination is critical for improving both management and selection efficiency in small-scale sheep farming. In r

The determination of the live body weight of camels (required for their successful breeding) is a rather difficult task due to the problems with handling and restraining these animals. Therefore, the main aim of this study was to predict the ABW of eight indigenous camel (Camelus dromedarius) breeds of Pakistan (Bravhi, Kachi, Kharani, Kohi, Lassi, Makrani, Pishin, and Rodbari). Selected productive (hair production, milk yield per lactation, and lactation length) and reproductive (age of puberty, age at first breeding, gestation period, dry period, and calving interval) traits served as the predictors. Six data mining methods [classification and regression trees (CARTs), chi-square automatic interaction detector (CHAID), exhaustive CHAID (EXCHAID), multivariate adaptive regression splines (MARSs), MLP, and RBF] were applied for ABW prediction. Additionally, hierarchical cluster analysis with Euclidean distance was performed for the phenotypic characterization of the camel breeds. The highest Pearson correlation coefficient between the observed and predicted values (0.84, p < 0.05) was obtained for MLP, which was also characterized by the lowest root-mean-square error (RMSE) (20.86 kg), standard deviation ratio (SD_ratio) (0.54), mean absolute percentage error (MAPE) (2.44%), and mean absolute deviation (MAD) (16.45 kg). The most influential predictor for all the models was the camel breed. The applied methods allowed for the moderately accurate prediction of ABW (average R² equal to 65.0%) and the identification of the most important productive and reproductive traits affecting its value. However, one important limitation of the present study is its relatively small dataset, especially for training the ANN (MLP and RBF). Hence, the obtained preliminary results should be validated on larger datasets in the future.

This study evaluates the effectiveness of XGBoost and LightGBM algorithms for estimating the live weight of Holstein×Zebu crossbred heifers. The study compares the performance of both algorithms using a wide range of biometric measurements and tests various hyperparameter settings. The research results show that the XGBoost algorithm provides almost perfect agreement with an R ² value of 0.999 on the training set and high performance with an R ² value of 0.986 on the test set. The LightGBM algorithm also achieved effective results with R ² values of 0.986 and 0.981 on both training and test sets. The machine learning algorithms used in the current study stand out as having the potential to provide a practical and economical solution for live weight estimation in livestock enterprises and especially for herd management applications in rural areas through input variables such as body measurements, milk yield, etc. However, the obtained results in the current study reveal the potential of machine learning algorithms for live weight estimation in the livestock sector and indicate that advanced research is needed for the optimisation of these algorithms.

The aim of this study is to evaluate the model performance in the classification of FAMACHA© scores using Support Vector Machines (SVMs) with a focus on the estimation of the FAMACHA© scoring system used for early diagnosis and treatment management of parasitic infections. FAMACHA© scores are a color-based visual assessment system used to determine parasite load in animals, and in this study, the accuracy of the model was investigated. The model’s accuracy rate was analyzed in detail with metrics such as sensitivity, specificity, and positive/negative predictive values. The results showed that the model had high sensitivity and specificity rates for class 1 and class 3, while the performance was relatively low for class 2. These findings not only demonstrate that SVM is an effective method for classifying FAMACHA© scores but also highlight the need for improvement for class 2. In particular, the high accuracy rate (97.26%) and high kappa value (0.9588) of the model indicate that SVM is a reliable tool for FAMACHA© score estimation. In conclusion, this study demonstrates the potential of SVM technology in veterinary epidemiology and provides important information for future applications. These results may contribute to efforts to improve scientific approaches for the management of parasitic infections.

This study aimed to compare the performance of five growth function models (Brody, Von Bertalanffy, Logistic, Gompertz, and Morgan-Mercer-Flodin) in predicting the growth patterns of Merino crossbred sheep in West Java, Indonesia. The analysis was conducted separately for male and female sheep, and the models were evaluated based on various statistical indicators. The Brody model yielded the highest coefficient of determination (R2) of 0.64 for male sheep, with an associated root mean square error (RMSE) of 6.825. The AIC and BIC values were 16413.30 and 16429.56, respectively, indicating the goodness of fit for the model. The Durbin-Watson (DW) statistic was 1.52, suggesting the absence of autocorrelation. Among the other models, Von Bertalanffy, Logistic, Gompertz, and Morgan-Mercer-Flodin also exhibited reasonably good fits, with comparable R2 values and model evaluation criteria. In the case of female sheep, the growth function models showed different performance characteristics compared to males. The Logistic model produced the highest R2 value of 0.63, indicating a good fit to the data. The RMSE was 5.036, and the AIC and BIC values were 10127.14 and 10143.25, respectively. The DW statistic was 0.93, indicating the absence of autocorrelation. Similar to the male sheep, the other models (Brody, Von Bertalanffy, Gompertz, and Morgan-Mercer-Flodin) also demonstrated reasonable performance. These results provide valuable insights into the growth patterns of Merino crossbred sheep in West Java, Indonesia, and can assist in optimizing breeding and management strategies. The findings highlight the importance of selecting appropriate growth function models based on sex-specific characteristics to accurately predict sheep populations' growth trajectories. Further research is recommended to validate and refine these models for broader application in the sheep farming industry.

Prediction of body weight (BW) using biometric measurements is an important tool especially for animal welfare and automatic phenotyping tools that needs mathematical models. In this study, it was aimed to predict the BW using body length (BL), chest girth (CG) and width of the waist (WW) for rabbits of the maternal form of Hyla NG. The standard rabbit‐raising practices were applied for the animals. A highly efficient gradient‐boosting decision tree (LightGBM), eXtreme gradient‐boosting (XGBoost) and support vector machine (SVM) algorithms were evaluated and compared to the prediction of BW. The coefficient of determination, root mean square error and mean absolute error values were used as comparison criteria. The results showed that LightGBM, XGBoost and SVM algorithms were well fit for the BW using the biometric measures with over 95% accuracy for both train and test sets. The BL was determined as the most explanatory variable on body weight.

The aim of this study was to define the growth by using nonlinear functions in intensive feedlot system with XGBoost algorithm. To achieve this aim, five nonlinear functions were implemented. To implementation of the study, Brown Swiss (n=41) and Simental (n=95) breed were used. Each nonlinear functions were examined for each breed. According to the results of the nonlinear functions, logistic model was the best prediction model for defining the growth of each breed. In this study, the parameters in the best prediction model were calculated individually and the relationship of these parameters with body weight was evaluated with the XGBoost algorithm. Model comparison criteria such as standard deviation ratio (SDratio), Pearson’s correlation coefficient (PC), determination of coefficient (R2) and Akaike’s information criteria (AIC) were used to evaluate the XGBoost algorithm. In conclusion, the XGBoost algorithm can be an effective and optional approach that allows breeders to estimate live weight from growth parameters. This algorithm can operate on large data sets with high accuracy and speed, leading to significant improvements in agricultural productivity and animal health management. XGBoost enables more accurate predictions by analyzing the effects of various characteristics (e.g., nutritional level, breed, age). Therefore, this method can be used to determine critical parameters such as body weight in animal breeding practices, serving as a powerful support tool for operational decisions.