Body weight prediction using different data mining algorithms in Thalli sheep: A comparative study.

Area of Science:

• Animal Science
• Agricultural Engineering
• Biostatistics

Background:

• Thalli sheep are a primary breed in Pakistan.
• Effective prediction of live body weight (BW) using linear body measurements is crucial for Thalli sheep management.
• Current predictive models for Thalli sheep BW are insufficient.

Purpose of the Study:

• To compare the predictive accuracy of four algorithms: Chi-square automatic interaction detector (CHAID), exhaustive CHAID, artificial neural network, and classification and regression tree (CART).
• To establish an optimal algorithm for predicting live BW in female Pakistani Thalli sheep using linear body measurements.

Main Methods:

• Utilized 152 body weight records and nine linear body measurements (wither height, body length, head length, rump length, tail length, head width, rump width, heart girth, barrel depth).
• Evaluated predictive performance using coefficient of determination (R²), standard deviation ratio, and root-mean-square error (RMSE).

Main Results:

• CHAID demonstrated the highest predictive capability with an R² of 64.48% and the lowest RMSE of 2.61.
• Heart girth was the most significant predictor for live BW across all tested algorithms.
• Body length greater than 73.91 cm was associated with the heaviest average BW (41.12 kg).

Conclusions:

• The CHAID algorithm is the most suitable for predicting live BW in female Thalli sheep.
• The developed predictive models accurately estimate Thalli sheep BW, supporting management decisions.
• Accurate BW prediction facilitates setting animal standards, determining drug dosages, and calculating feed requirements.