APPLICATION OF MACHINE-LEARNING-BASED HYBRID ALGORITHM FOR PRODUCTION FORECAST IN TEXTILE COMPANY
Résumé
The increasing consumer demand for sustainable textiles has led companies to invest in organic and recycled materials while adopting green manufacturing processes to reduce their environmental footprint. The textile industry is exploring innovative approaches such as upcycling and green chemistry to regenerate textile waste fibers sustainably. And these efforts are aimed at transforming the fashion and textile industry towards a more sustainable and eco-friendly future. To achieve this the textile industries, have to leverage on the technological advancements available which one of it is, various applications of ML models for a more accurate and efficient results. This study focused on the application of a hybrid machine learning model which is the combination of Seasonal Auto-Regressive Integrated Moving Average (SARIMA) model and Long Short-Term Memory (LSTM) networks for Accurate Production Forecasting in Textile Industry. Research studies reviewed gave important insights into the study. This study adopted a quantitative research approach, using a five-year historical production data from a textile company to train and evaluate the hybrid ML model. The research design includes data collection, preprocessing, model training, evaluation, and comparative analysis. The research analysis presents a well-performing predictive model, as evidenced by the high R-squared value of 0.87, relatively low root mean squared error (RMSE) of 1.23, low mean squared error (MSE) of 1.51, and low mean absolute error (MAE) of 0.98. However, the residual time series and the predicted and actual values plots suggest possible degradation in the model's performance over time.
| Origine | Publication financée par une institution |
|---|---|
| Licence |