Submit or Track your Manuscript LOG-IN

Comparing Predictive Performances of MARS and CHAID Algorithms for Defining Factors Affecting Final Fattening Live Weight in Cultural Beef Cattle Enterprises

Comparing Predictive Performances of MARS and CHAID Algorithms for Defining Factors Affecting Final Fattening Live Weight in Cultural Beef Cattle Enterprises

Adem Aksoy1, Yakup Erdal Ertürk2, Ecevit Eyduran3 and Mohammad Masood Tariq4,*

1Department of Agricultural Economics, Agricultural Faculty, Ataturk University, Erzurum, Turkey
2Department of Agricultural Economics, Agricultural Faculty, Iğdır University, Iğdır, Turkey
3Department of Business Administration, Faculty of Faculty of Economics and Administrative Sciences, Iğdır University, Iğdır, Turkey
4Centre of Advanced Studies in Vaccinology and Biotechnology, University of Balochistan, Quetta, Pakistan

*      Corresponding author: tariqkianiraja@hotmail.com

 

ABSTRACT

This study was conducted to define vital factors on final fattening live weight (FFW) on cultural beef cattle enterprises from Eastern region of Turkey. Predictive performances of Multivariate Adaptive Regression Splines (MARS) and Chi-Square Interaction Detector (CHAID) were evaluated comparatively in the definition of significant factors and interaction effects between the factors. Before the definition process, the data on socio-economic (age, province, educational level, experience, social security, lands and the reason at ranching of the animal breeders) and biological factors (sex, first live weight before fattening and fattening period of the beef cattle) were recorded from the related beef cattle enterprises. For the statistical evaluation of MARS algorithm, the package “earth” of the R software was employed based on the smallest GCV value. In the CHAID algorithm, minimum enterprise numbers in parent and child nodes were set at 4 and 2 for ensuring strong predictive accuracy with the Bonferroni adjustment. MARS algorithm gave a very good performance in the prediction of final fattening weight according to goodness of fit criteria i.e. R2 (0.983) and SDRATIO (0.114). Very strongly significant Pearson correlation coefficient (r=0.992) between observed and predicted FFW values in the MARS were found for the cultural beef cattle enterprises, respectively (P<0.01). The respective model evaluation criteria for CHAID algorithm were estimated as 0.671 R2 and 0.574 SDRATIO. Whereas, the respective correlation coefficient for CHAID algorithm was 0.819 (P<0.01). MARS outperformed CHAID algorithm in predictive quality. In the CHAID algorithm, the first live weight, farmer’s age, pasture land, SOCSEC, fattening period and sex of the beef cattle were found for FFW as the influential predictors, whereas main and interaction effects of all the predictors handled here were found significant in the MARS. In conclusion, the results represented that MARS may submit meaningful hints to enterprises in the description of noticeable factors on FFW for further studies to be conducted under similar conditions.
 

To share on other social networks, click on P-share. What are these?

Pakistan Journal of Zoology

October

Vol. 50, Iss. 5, Pages 1601-1998

Featuring

Click here for more

Subscribe Today

Receive free updates on new articles, opportunities and benefits


Subscribe Unsubscribe