Prediction of fresh herbage yield using data mining techniques with limited plant quality parameters

Area of Science:

• Agricultural Science
• Data Mining
• Forage Production

Background:

• Sorghum-Sudangrass hybrids are vital forage crops in arid and semi-arid regions.
• Optimizing fresh herbage yield is crucial for livestock sustainability.
• Understanding plant traits and their relationship with yield is essential for effective management.

Purpose of the Study:

• To assess fresh herbage yield, fertilizer effects, and plant characteristics of Sorghum-Sudangrass.
• To evaluate the predictive performance of data mining techniques for herbage yield.
• To identify the most suitable model for forecasting fresh herbage production in arid environments.

Main Methods:

• Data mining techniques including CHAID, CART, MARS, and Bagging MARS were employed.
• Plant traits and yield data were collected from Konya and Sanliurfa in 2021-2022.
• Model performance was evaluated using R², adjusted R², RMSE, MAPE, SD ratio, and AIC.

Main Results:

• The MARS algorithm demonstrated superior predictive accuracy.
• MARS achieved the highest R² (0.993) and adjusted R² (0.989).
• MARS recorded the lowest RMSE (246), MAPE (1.926), SD ratio (0.085), and AIC (845).

Conclusions:

• The MARS algorithm is the most effective model for characterizing Sorghum-Sudangrass fresh herbage yield.
• MARS provides a robust alternative to other data mining methods for yield forecasting.
• Accurate yield prediction supports optimized agricultural practices in arid regions.