Scholarlink

The sustainable supply of Tikog grass is essential for the continuous production of native handicrafts, ensuring artisans can meet customer demand while maintaining efficient resource management. However, as the Philippines is highly prone to typhoons, the availability of Tikog grass can be unpredictable, posing challenges to production. To address this issue, an accurate forecasting model is crucial to prevent shortages and optimize supply planning. This study employs a hybrid LSTM-XGBoost model to predict the required amount of Tikog grass needed for production. By combining Long Short-Term Memory (LSTM), which effectively captures sequential patterns in time-series data, with Extreme Gradient Boosting (XGBoost), which enhances predictive accuracy, the model provides a reliable approach to demand forecasting. The results demonstrate that LSTM-XGBoost outperforms traditional LSTM models, achieving a high predictive accuracy with an R2 value of 97.04%. Moreover, key error metrics indicate improved performance, with Mean Squared Error (MSE) at 0.00698, Mean Absolute Error (MAE) at 0.029454, and Root Mean Squared Error (RMSE) at 0.083544. These findings highlight the model's ability to enhance decision-making in Tikog production, ensuring sufficient raw material availability while minimizing waste. By integrating machine learning into traditional handicraft industries, this research provides a valuable tool for weavers, enabling them to adapt to fluctuating supply conditions and maintain the sustainability of Tikog weaving. Future research can further optimize the model by incorporating external factors such as weather patterns, economic indicators, and market trends to improve predictive accuracy and support long-term industry resilience.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Mitigating Overfitting in Tikog Grass Prediction: An Enhanced LSTM-XGBoost Model for Sustainable Handicraft Production

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Devine Grace D. Funcion,
  • Marvee Cheska Natividad

摘要

The sustainable supply of Tikog grass is essential for the continuous production of native handicrafts, ensuring artisans can meet customer demand while maintaining efficient resource management. However, as the Philippines is highly prone to typhoons, the availability of Tikog grass can be unpredictable, posing challenges to production. To address this issue, an accurate forecasting model is crucial to prevent shortages and optimize supply planning. This study employs a hybrid LSTM-XGBoost model to predict the required amount of Tikog grass needed for production. By combining Long Short-Term Memory (LSTM), which effectively captures sequential patterns in time-series data, with Extreme Gradient Boosting (XGBoost), which enhances predictive accuracy, the model provides a reliable approach to demand forecasting. The results demonstrate that LSTM-XGBoost outperforms traditional LSTM models, achieving a high predictive accuracy with an R2 value of 97.04%. Moreover, key error metrics indicate improved performance, with Mean Squared Error (MSE) at 0.00698, Mean Absolute Error (MAE) at 0.029454, and Root Mean Squared Error (RMSE) at 0.083544. These findings highlight the model's ability to enhance decision-making in Tikog production, ensuring sufficient raw material availability while minimizing waste. By integrating machine learning into traditional handicraft industries, this research provides a valuable tool for weavers, enabling them to adapt to fluctuating supply conditions and maintain the sustainability of Tikog weaving. Future research can further optimize the model by incorporating external factors such as weather patterns, economic indicators, and market trends to improve predictive accuracy and support long-term industry resilience.