Browsing by Author "Cakmakci, Cihan"

Now showing 1 - 14 of 14

Associations Between Cyp17 Gene Polymorphisms, Temperament and Maternal Behavior in Ewes, and Growth in Their Lambs
(Elsevier Science inc, 2021) Karaca, Serhat; Ser, Gazel; Ulker, Hasan; Yilmaz, Onur; Cakmakci, Cihan; Ata, Nezih; Saribey, Muammer
This study was conducted to determine temperament and maternal care classes, CYP17 gene polymor-phisms, and cortisol status of a group of Norduz ewes. Additionally, the relationships between these fac-tors, their effects on lambs survival and growth characteristics were investigated. A total of 72 ewes were subjected to arena and isolation box tests to determine their temperament classes. The distance the ewe retreated from the lamb during tagging was utilized to assess ewe maternal care classes. Ewes were genotyped for CYP17 628 SNP. Blood samples were collected between lambing and weaning to measure cortisol concentrations of ewes. The survival and growth data for the lambs of these ewes were recorded until weaning. Two temperament (low reactive and high reactive) and 2 maternal care classes (strong and moderate) were present. Two CYP 17 SNP 628 alleles (A and G) and 3 genotypes (AA, AG, and GG) were identified. For the distribution of CYP17 genotype frequencies within the population, the temperament or maternal care classes were not different (P > 0.05). The temperament of ewes did not affect mater-nal care in our flock. Positive partial correlations between cortisol and total cross, distance to stimulus, and sniffing were determined (P < 0.05). CYP17 genotype was an effective factor for some behavioral traits (i.e., total cross) and cortisol release (P < 0.001). Neither temperament (P = 0.636) nor maternal care classes (P = 0.985) differed with cortisol concentrations. CYP17 genotype, cortisol status, tempera-ment or maternal care class of ewes did not affect birth or weaning weight of their lambs (P > 0.05). Although there was no significant relationship among CYP17 genotype, maternal care, and lamb survival and growth, the effect of CYP17 genotype on some behavioral traits and cortisol release was found to be significant. Further studies with larger animal populations are required to elucidate the possible use of CYP17 SNP 628 genotypes as genetic markers for temperament classification. (c) 2021 Elsevier Inc. All rights reserved.
Changes in the Current Patterns of Beef Consumption and Consumer Behavior Trends-Cross Study Brazil-Spain
(Mdpi, 2023) Magalhaes, Danielle Rodrigues; Cakmakci, Cihan; Campo, Maria del Mar; Cakmakci, Yusuf; Makishi, Fausto; Silva, Vivian Lara dos Santos; Trindade, Marco Antonio
This cross-cultural study aimed to determine the main factors behind potential changes in eating habits by analyzing changes in the patterns of beef consumption currently observed in Brazil, Spain, and Turkey. To achieve this aim, 412 regular beef consumers from Brazil, 407 from Spain, and 424 from Turkey answered a self-administered questionnaire. The study surveyed the effects of economic factors, switching from beef to other sources of protein, aspects of credence, health-related concerns, the influence of lifestyle on beef consumption patterns, and purchasing decision factors. The most important factors that changed consumer behavior and resulted in a decrease in consumption, mostly among Brazilian and Turkish consumers, were the economics and accessibility of the products. Beef was replaced by other alternative sources of protein that were likewise derived from animals. The consumers whose purchasing intentions were most significantly influenced by credence factors (e.g., indiscriminate use of agricultural products, substandard animal welfare requirements, among others) were Brazilian and Turkish and, to a lesser degree, Spanish consumers. Lifestyle factors (e.g., consumption of out-of-home meals, available time to cook, among others) were demonstrated to alter consumption patterns and therefore must be carefully considered by the industry, taking into account cultural differences and consumer needs. The population under investigation considered that eating beef had no impact on their health.
Determination of Consumer Perceptions of Eco-Friendly Food Products Using Unsupervised Machine Learning
(Univ Namik Kemal, 2024) Cakmakci, Yusuf; Hurma, Harun; Cakmakci, Cihan
This study aims to comparatively determine the consumer perception of food products marketed under ecologically friendly concepts (organic food, good agriculture, and natural production) and food sold directly by farmers, conventional food, and farmer cooperative branded food. For this purpose, a face-to-face survey was conducted with 171 identified consumers. R program was used to perform all of the analyses. Machine learning methods such as Logistic Regression (LR), Correspondence Analysis (CA), and Principal Component Analysis (PCA) are used for determining consumer perception from obtained data. Descriptive statistics results showed that 51.5 percent of those polled were male and 48.5 percent were female. It found that the mean age of the consumers was joined to the survey was 36.4. According to the LR findings, consumer socioeconomic characteristics have a considerable impact on the purchase of various foods (such as organic labeled foods, good agricultural practices foods, producer cooperative branded foods, etc.). It has been discovered as the result of the PCA, people perceived organic branded food and good agricultural practices foods, which are healthier, more flavorful, and more trustworthy than other food. however, it has been discovered that they believe the costs of these types of food are expensive and that they are difficult to obtain. On the other hand, they perceive the pricing of farmer cooperative branded foods and food sold directly by the farmer to be lower. Furthermore, it was observed in CA findings that there was a correlation between these various food groups and purchase locations. While products sold directly by farmers are mostly purchased from public markets, they prefer grocery stores and supermarkets when purchasing foods with good agricultural practices and natural labeled (from the markets). When seen from this perspective, it is possible to conclude that ecologically friendly food and other food products are regarded differently by customers based on product characteristics. The use of marketing techniques that create a positive perspective in terms of affordability and accessibility and the development of policies and production techniques that boost consumers' current perceptions of these items are considered will promote the consumption of these products.
Discovering the Hidden Personality of Lambs: Harnessing the Power of Deep Convolutional Neural Networks (Dcnns) To Predict Temperament From Facial Images
(Elsevier, 2023) Cakmakci, Cihan; Magalhaes, Danielle Rodrigues; Pacor, Vitor Ramos; Silva de Almeida, Douglas Henrique; Cakmakci, Yusuf; Dalga, Selma; Titto, Cristiane Goncalves
The objective of this study was to define a more practical and reliable alternative to manual temperament classification methods that rely on the behavioral responses of animals individually subjected to various tests. Specifically, this study evaluated the correlation between facial image information and temperament based on deep convolutional neural networks (DCNNs) to predict the temperament of lambs based on their facial images. In the first phase, the lambs were categorized as to their temperament based on data acquired from a behavioral test to establish a ground truth for the temperament of the lambs. This enabled us to train (70%), validate (20%), and test (10%) deep-learning models in the second phase based on facial images and the corresponding temperament labels derived from the behavioral test. The performance of a custom deep convolutional neural network (C-DCNN) was compared to that of pre-trained VGG19 and Xception models for image classification. The Xception model achieved a training accuracy of 81%, which indicated that it learned well the underlying patterns in the data; however, lower validation (0.75) and test (0.58) accuracies indicate that it overfit the training data and did not generalize well to new samples. The VGG19 model, produced lower training (0.59), validation (0.46), and test (0.34) accuracies, which indicated that it did not learn the underlying patterns in the data as well as the Xception model. Furthermore, its precision (0.47), recall (0.42), and F1 score (0.41) indicated that the model performed poorly in identifying the classes correctly. The C-DCNN produced a moderate accuracy of 60%, which indicated that the model was able to predict the temperament traits of lambs with an accuracy of 60%, which was better than random guessing (33% accuracy), and demonstrated the potential of this approach in assessing temperament. The C-DCNN precision (0.69), recall (0.61) and F1 score (0.63) indicated that it had a moderate ability to correctly identify positive cases; however, the small size of the original dataset remains a limitation of the study because it might have caused the suboptimal performance of the models. To validate this approach, further research is needed based on a larger and more diverse dataset. We will continue to investigate the potential of deep learning and computer vision to predict animal personality traits from facial images based on large, diverse datasets, which might lead to more efficient and objective methods for assessing animal temperament and improving animal welfare.
Does Coping Style Affect Behavioral Responses and Growth Performance of Lambs Weaned at Different Ages
(Elsevier Science inc, 2021) Cakmakci, Cihan; Karaca, Serhat; Maria, Gustavo A.
The objective of this study was to determine the relationships between coping style (CS), and physiological, behavioral characteristics and growth performance of lambs weaned at different ages. Male Karakas lambs (n = 75, mean age = 30 d) obtained from a breeder farm were subjected to a 3-minute open field test. Of the 75 lambs classified by hierarchical clustering on principal components' scores on open field test behaviors, 40 were assigned to 1 of 2 groups as proactive (Pro, n = 20) or reactive (Rea, n = 20). Pro (n = 10) and Rea (n = 10) lambs were selected randomly and weaned at 45 d (D45). The remaining 20 lambs (10 Pro, 10 Rea) were weaned at 75 d (D75). Lambs from Pro and Rea groups (Pro n = 5, Rea n = 5) were randomly assigned to each pen (n = 10). Blood samples were taken from lambs just before and 20 minutes after the start of open field test and at weaning (on 0, 1, 3, 5, and 7 d after weaning) to measure serum cortisol, glucose, lactate dehydrogenase, creatine kinase, total protein, and triglyceride concentrations. There was no significant effect of CS on blood parameters measured both in open field test and that measured at weaning (P > 0.05). CS had significant effect on live weight (LW) and average daily weight gain (ADWG) (P < 0.05). Reactive lambs had greater LW and ADWG than Pro lambs (P < 0.05). Weaning age (WA) had significant effect on LW and ADWG (P < 0.05). The ADWG were greater in D75 lambs than in D45 lambs (P < 0.05). CS did not have significant effect on behavioral variables measured at weaning, except for success index values. Rea lamb had greater index values than Pro lambs (P < 0.05). WA had significant effect on behavioral variables measured at weaning and those variables were higher in D75 lambs than in D45 lambs (P < 0.05). Partial correlations between open field test behaviors and those measured at weaning were not significant (P > 0.05). There was a significant positive partial correlation between aggressive behaviors and glucose (P < 0.05). The strongest partial correlations in the network were between creatine kinase and lactate dehydrogenase (r = + 0.78, P <0.05), and stereotypic behaviors and aggressive behaviors (r = + 0.67, P <0.05). Although the behavioral responses measured in the open field test are less time consuming than are those measured in ethological field studies, they do not reflect adequately the social behavior of lambs after weaning; however, WA had significant effects on social and stereotypical behaviors; therefore, to ensure social cohesion among lambs, it is important to identify the appropriate age for weaning. (C) 2021 Elsevier Inc. All rights reserved.
Does the Length of Time Dried Distillers' Grain With Solubles Substitution for Soybean Meal Affect Physiological Indicators and Meat Quality in Finishing Lambs
(Wiley, 2021) Karaca, Serhat; Erdogan, Sibel; Guney, Mehtap; Cakmakci, Cihan; Saribey, Muammer; Kor, Askin; Ulker, Hasan
This study examined how inclusion of dried distillers' grains with solubles (DDGS) in finishing lamb diets for different periods affects some physiological indicators and meat quality. A total of 40 male lambs were divided into four groups according to feeding regimen during the 120-day finishing period as follows: C120: no DDGS included in diet for 120 days; D120: DDGS included in diet for 120 days; D75: no DDGS included in diet for 45 days + DDGS included in diet for 75 days; D45: no DDGS included in diet for 75 days + DDGS included in diet for 45 days. Dietary inclusion rate of DDGS was 27.5%. Fattening performance and rumen parameters were not affected by treatment. Feeding regimens had no significant effect on meat quality except the instrumental tenderness and juiciness score. DDGS fed lambs had higher level of total trans fatty acids and n-6/n-3 ratio in meat. 10t-C18:1, 11t-C18:1 and 9c,11t conjugated linoleic acid contents of adipose tissue were higher in all the DDGS groups compared with C120 lambs (p < .001). These results suggest that soybean meal can be replaced with corn DDGS in lamb diets for up to 120 days during the finishing period with no adverse effects on some physiological response feedlot performance and meat quality.
Editorial: Behavior and Heat Stress
(Frontiers Media Sa, 2023) Titto, Cristiane Goncalves; Henrique, Fabio Luis; Pantoja, Messy Hannear de Andrade; Cakmakci, Cihan; Silva, Priscila dos Santos
Effect of Dried Corn Distillers' Grains With Solubles and Soybean Meal Supplements on Physiological Indicators and Reproductive Performance of Ewes
(Taylor & Francis Ltd, 2018) Erdogan, Sibel; Karaca, Serhat; Guney, Mehtap; Kor, Askin; Cakmakci, Cihan; Saribey, Muammer
This study aimed to evaluate the effects of adding dried distillers' grains with solubles (DDGS) to the nutritional regimens of ewes at different time periods on reproductive traits, serum hormones and serum metabolites. In Experiment 1, 100 ewes were divided into 4 groups (n=25) according to diet. The four diets were grass hay (H) alone (group H), H with barley (group H+B), H with DDGS for the whole feeding period (27d) (group H+DDGS) or H and DDGS for 5d (d 8-12 of oestrus cycle) (group H+DDGS5). Serum progesterone concentrations were affected by oestrus cycle (p<0.001), but not by dietary treatments. However, feeding H+DDGS caused significant increases in serum insulin, leptin and growth hormone concentrations (p<0.05). In Experiment 2, 30 ewes were divided into 2 groups (n=15), receiving DDGS or soybean meal (SBM) during the prepartum period. Diets had no significant effect on weights of dams or lambs at birth; however, the weaning weights of lambs born from ewes of group DDGS were significantly higher (p<0.05). Moreover, serum glucose, blood urea nitrogen, triglyceride and beta hydroxybutyric acid (BHBA) concentrations during the prepartum period were affected by dietary treatments and by time (p<0.001). During the prepartum period, BHBA concentrations of ewes fed SBM were higher, while BHBA levels decreased during the last weeks of pregnancy regardless of diet. Lamb serum immunoglobulin G concentrations increased from 1h to 24h after birth (p<0.001). Colostrum of ewes fed SBM had higher fat-free dry matter (DM) and protein contents in comparison to colostrum of ewes fed DDGS (p<0.05). In conclusion, DDGS can be included as protein source in pregnancy rations up to 15% of DM to obtain reproductive performance outcomes equal to or exceeding those obtained with SBM.
Estimating the Repeatability of Behavioral Traits in Norduz Sheep Subjected To an Arena Test
(Elsevier, 2022) Cakmakci, Cihan
The objective of this study was to estimate the repeatability of behavioral traits in sheep subjected to a 5-min arena test. Two separate tests were conducted over two years to assess the among-individual consistency of behavioral responses in 68 nonpregnant Norduz ewes (2-5 years of age). During each arena test, five behavioral variables were measured: distance to group sheep (DTG), locomotor activity (LOC), zone boundaries crossed (CRS), vocalization (VOC), escape attempts (ESC). A linear mixed model was used to predict DTG and LOC, while a GLMM with a Poisson distribution was used to predict VOC, CRS, and ESC. Age and year were used as fixed factors in both models. The effect of the year was statistically significant on DTG, LOC, and CRS (P < 0.05). The correlations between traits changed over time. In the first year, DTG had no correlation with other variables. However, DTG had a significant positive correlation with LOC (0.54) and CRS (0.64) in the second year. The repeatability of behavioral variables was estimated as linear mixed-effects models (LMMs) and generalized linear mixed-effects models (GLMMs). To estimate adjusted repeatability, the ewes' age and the year of testing were included as fixed effects. The ID of ewes was included as a random effect in all models. The effect of age was statistically significant on LOC and CRS (P < 0.05). The repeatability of behavioral variables varied from the lowest for the ESC (R = 0.10; 95 % CI [0.000, 0.389], P = 0.319) to the highest for the VOC (R = 0.38; 95 % CI [0.054, 0.614], P = 0.006). These results indicate the consistency of among-individual variation in sheep be-haviors that was measured in the arena test.
Inheriting the Sins of Their Fathers: Boar Life Experiences Can Shape the Emotional Responses of Their Offspring
(Frontiers Media Sa, 2023) Sabei, Leandro; Sarmiento, Marisol Parada; Bernardino, Thiago; Cakmakci, Cihan; Farias, Sharacely de Souza; Sato, Denis; Zanella, Adroaldo Jose
IntroductionThe welfare of breeding boars is often overlooked, resulting in limited scientific data to foster discussion of the topic. We aimed to investigate the effect of different boar housing conditions on their offspring's emotionality. MethodsEighteen boars were housed in three different conditions: crates (C; n = 6), pens (P; n = 6), or enriched pens (E; n = 6). Boars were distributed by semen quality (SQ; high, medium, or low). Three semen pools were used to inseminate 13 gilts housed in outdoor paddocks. At 25 days of age, 138 suckling piglets were subjected to open field (OF), novel object (NO), and elevated plus-maze (EPM) tests. Saliva was collected before and after the OF and NO tests to measure cortisol concentrations. At the end of the experiment, hair samples were collected for DNA paternity tests. Piglets were classified based on their behavioral responses using hierarchical cluster analysis of the principal components extracted from factor analysis of mixed data. The variables were reduced to seven principal components (dimensions, Dims), which explained 73% of the total variation, and were analyzed using linear mixed models. The models included each Dim as a dependent variable, paternal treatment and body weight (BW) as fixed effects, and paternal SQ as a random effect. Kruskal-Wallis and Wilcoxon rank-sum tests were used to compare the cortisol concentration ratios (before and after the OF and NO tests) between groups. ResultsThere was an effect of treatment on Dim 3 (EPM; activity/fear), with higher values in C piglets than E piglets (p = 0.047). Although C piglets had significantly higher values than P piglets in Dim 4 (EPM; anxiety; p = 0.029) and Dim 6 (NO; inactivity far from the object/exploration; p < 0.0001), the effect of the paternal treatment x BW interaction was significant in both dimensions (p < 0.05). The cortisol ratio in E piglets was greater than that in P and C piglets (p < 0.05). Discussion and conclusionOur findings indicate that boar breeding environments affect the stress response and emotionality (anxiety, fear, and exploration) of their offspring.
Number of Consumers and Days of Display Necessary for the Assessment of Meat Colour Acceptability
(Elsevier Science Bv, 2019) Cortez Passetti, Rodrigo Augusto; Celia Resconi, Virginia; Cakmakci, Cihan; del Mar Campo, Maria; Karsten Kirinus, Jackeline; Gomes Passetti, Ludmila Couto; Sanudo, Carlos
Visual assessment is regarded as the gold standard to evaluate meat colour shelf-life, but it is costly and time consuming. To address this issue, this paper aims to evaluate the number of consumers and days of display that are necessaries in order to assess the colour shelf-life of meat, presented with different methods, all using images. Photographs of thirty-six lamb steaks were taken just after cutting (day 0) and on each of the following days until the 14th day of display under standardized conditions. Images were presented in three different manners: 1) with days of display and animals in random order (Random); 2) days of display in sequential and animals in random order (Sequential); and, 3) days of display and animals in sequential order (Animal); they were presented to 211 consumers who evaluated visual acceptability on a 9-point scale. At day zero, visual acceptability scores were the highest in Animal, followed by Sequential, and then by the Random (P <.05) method. Scores decreased over time for all methods tested (P <.05). The Random method presented the highest standard deviation; however, an increase in standard deviation among consumers along days of display was observed for all methods tested (P <.05). Shelf-life determined by regression varied according to the method of presentation (7.83, 7.00 and 7.54 days for Random, Sequential and Animal, respectively). A minimum number of 4 day points before and 4 day points after neutral scores had been reached (scores = 5.0) were necessary in order to obtain a robust model. The minimum number of required consumers (alpha = 0.05; d = 0.1 and beta = 0.2 or 0.1) varied according to methodology: it was 81 to 109 consumers for Random, 69 to 92 for Sequential, and 55 to 74 for Animal. Our study indicates that an optimal number of days and evaluators can be calculated depending on the manner of sample presentation. These findings should be taken into account in further studies that aim to balance data reliability with the cost involved in meat colour analyses.
Sheep's Coping Style Can Be Identified by Unsupervised Machine Learning From Unlabeled Data
(Elsevier, 2022) Cakmakci, Cihan
The objective of this study was to define coping style of sheep by using unsupervised machine learning approaches. A total of 105 Norduz sheep (age 3-5 years) were subjected to a 5-minute arena test. Agglomerative Hierarchical Clustering (HCA) was performed on scores of selected principal components retained from Principal Components Analysis (PCA) on arena behaviors to identify sheep coping style. Initially, the variables retained for the PCA were determined with Bartlett's test for sphericity and Kaiser-Meyer-Olkin (KMO) measure of sample adequacy. Seven behavioral variables with KMO values greater than 0.5 were used for final PCA: the average distance to group sheep (DTG), the average distance to stimulus (DTS), the duration of locomotion (LOC), the total number of zone boundaries crossed during the test (CRS), the total number of times that tested sheep sniffed stimulus (NSS), latency to the first sniff the stimulus (LSS), and subjective scores (SCR) scored by an observer on a scale from 1 to 5 (1: extremely calm, 5: extremely restless). The first two components, which were the only ones with an eigenvalue greater than one, accounted for 70.32% of the total variation and were used for clustering analysis. Clustering tendency showed that the scores for the first two components were suitable for clustering (Hopkins' H = 0.852). Several cluster validity indexes were used to obtain aggregated results to determine the most appropriate clustering method and number of clusters. Five different clustering methods: k-means and hierarchical clustering with Ward, average, single and complete linkage were compared. Bootstrap resampling was used to evaluate the stability of a given cluster using the Jaccard coefficient. The clustering method and number of clusters corresponding to the highest rank aggregation score from the bootstrap resampling indicate that the hierarchical clustering method with average linkage and 5 clusters is the most suggested clustering method. However, Ward's algorithm identified the strongest clustering structure for hierarchical clustering, as it had the highest agglomerative coefficient value (0.98). When both Jaccard and aggregation scores are considered together, Ward's method with 3 clusters was selected as the most appropriate method. Sheep were classified into three coping styles (CS) based on HCA results as reactive (Cluster 1, n = 71), intermediate (Cluster 2, n = 22) or proactive (Cluster 3, n = 12). Coping style had significant effect on behavioral variables, DTG, DTS, LOC, CRS and NSS (P < 0.05). The individuals that have proactive coping style had the highest mean values for the variables DTG, DTS and LOC and SCR (P < 0.0001). This indicates that proactive sheep are more active then reactive sheep. The CRS, LOC and NSS mean values were higher for intermediate sheep compared to reactive sheep (P < 0.05). The NSS values were higher for intermediate sheep compare to proactive sheep (P < 0.0001). The findings of the current study show that distinct coping styles in sheep may be identified based on behaviors recorded in an arena test. The findings also revealed that sheep's coping style can be objectively identified by unsupervised machine learning from unlabeled behavioral data.
Social Networks of Pregnant Gilts During Outdoor Feeding and the Effects on Their Offspring
(Elsevier, 2024) Sabei, Leandro; Sarmiento, Marisol Parada; Cakmakci, Cihan; Farias, Sharacely de Souza; Bernardino, Thiago; Poletto, Rosangela; Zanella, Adroaldo Jose
Social relationships are important aspects of the behavioural biology of pigs and can be affected by the type of housing pigs are kept in. Exploring agonistic interactions and affiliative behaviours can reveal effects on adult pigs and their descendants. This research investigated the social dynamics among gilts throughout pregnancy during collective feeding in an outdoor housing system and the effects of these dynamics on the stability of the group and the offspring of the group members. For this study, 15 gilts were oestrus synchronised and artificially inseminated with three different semen pools. The paddocks where the gilts were housed contained a mud pool, natural tree shade, and two nipple drinkers. Two daily meals (similar to 2.5 kg/day/gilt) were provided to the group on the floor. Feeding behaviour was recorded for three continuous days (20 min in the morning and again in the afternoon) every gestational (a total of 42 h of video recording). Saliva samples were collected at 6:00 a.m. and 6:00 p.m. on these same days. A trained observer evaluated all the videos using Boris software to determine agonistic and affiliative behaviours. After farrowing, the piglet data collected included sex, mortality, and body weight (BW) at 10, 25, 29, and 36 d of age. Elo scores were calculated to quantify the relative hierarchy among the gilts based on the behaviour assessment. Generalised linear mixed models (GLMMs) were used for data analyses, and the significance of fixed effects was determined at p < 0.05. The tendency of the gilts to feed together decreased as gestation progressed. Heavier gilts had lower salivary cortisol concentrations, and higher gilt rank was correlated with higher morning salivary cortisol concentrations and heavier piglets. Maternal modulation of offspring performance in pigs warrant further investigation. Assessment of gilts' social feeding behaviours is relevant for considering pigs' behaviour in genetic selection and improving commercial facilities and management practices to improve animal welfare.
Use of Machine Learning Approaches for Body Weight Prediction in Peruvian Corriedale Sheep
(Elsevier, 2024) Canaza-Cayo, Ali William; Churata Huacani, Roxana; Cakmakci, Cihan; Rodriguez-Huanca, Francisco Halley; Filzo, Julio Silvio de Sousa Bueno; Fernandes, Tales Jesus; De La Cruz, Yhan Carlos Rojas
The goal of this study was to predict the body weight of Corriedale ewes using machine learning (ML) algorithms. Fourteen body measurements (BM) and six different machine learning models were used. Body weight (BW) and BM: wither height (WH), rump height (RH), thoracic perimeter (TP), abdominal perimeter (AP), foreshank length (FSL), fore-shank width (FSW), fore-shank perimeter (FSP), tail width (TW), tail perimeter (TPe), hip width (HW), loin width (LWi), shoulder width (SW), forelimb length (FL), and body length (BL), were collected from 100 Corriedale ewes between 1.5 and 2 years old from the Illpa Experimental Centre of the National University of Altiplano in Peru. The machine learning algorithms used to estimate body weight were Support Vector Machines for Regression (SVMR), Classification and Regression Trees (CART), Random Forest (RF), Model Average Neural Networks (MANN), Multivariate Adaptive Regression Splines (MARS) and eXtreme Gradient Boosting (XGBoost). The performance of the models was evaluated by the coefficient of determination (R2), root mean square error (RMSE), mean absolute error (MAE), and mean absolute percentage error (MAPE). Highly correlated predictors (r >= 075) were removed from the dataset. The remaining predictors were then subjected to variable selection procedures using the Boruta algorithm. Boruta results confirmed the importance of TP, LWi, BL, FSL, SW and HW as predictors of ewe weight. The ML models were then trained on those selected predictors. RF had the highest R2 values and lowest values of MAE, RMSE, and MAPE. In conclusion, the RF algorithm can be recommended for accurately estimating BW from body measurements of Corriedale sheep.