The main aim of the present work was to determine potential factors that influence average fattening
final live weight (AFFLW) per enterprise for crossbred cattle reared in beef cattle enterprises in North
East Anatolia region (comprising Erzurum, Igdir, Kars and Agri provinces) of Turkey. For this goal,
Multivariate Adaptive Regression Splines (MARS), a-non parametric regression technique, was used to
develop a respectable prediction equation that can denote interaction effects of influential predictors in
the definition of the influential factors on AFFLW as an output variable for male and female crossbred
cattle. Several predictors in the current survey were province (Erzurum, Igdir, Kars and Agri), farmer age,
educational level (illiterate, primary school, secondary school, high school, and college), social security
(available and unavailable), husbandry experience of farmer (year), farmer`s irrigated, dry and pasturage
land, live weight before fattening, and fattening period. Predictive accuracy of MARS algorithm was
evaluated using coefficient of determination (R2), Standard Deviation Ratio (SDRATIO), Generalized
Cross Validation (GCV) and Pearson`s correlation (r) between actual and predicted AFFLW. Pearson`s
correlation coefficients between actual and predicted AFFLW for male and female crossbred cattle were
very strongly estimated (r~1.00, P<2.2e-16). Results of MARS, giving a very good fit for 2nd interaction
order, denoted that almost all of the variability in AFFLW per enterprise was explained based on the
lowest GCV. In this study, social factors (farmer’s age, experience, province, educational level, social
security status, and aim in performing animal production), economic factors (dry land, irrigated land, and
pasturage land of the farmer) and biological factors (first live weight before fattening, and fattening period
of the beef cattle) were found as influential factors on fattening final live weight in the crossbred beef
cattle. The achieved results suggested that interaction effects of influential predictors entered into MARS
prediction equation could change AFFLW per enterprise.