<p>This research aims to develop an intelligent model to measure the sustainability of fashion trends using a large eco-friendly fashion dataset. As the interest in the environmental sustainability of fashion continues to rise, our framework blends sophisticated machine-learning schemes to quantify sustainability along various dimensions. We employ two robust schemes, the Gradient Boosting Classifier (GBC) and the Light Gradient Boosting Classifier (LGBC), and enhance their performance using the latest optimization algorithms, the Horned Lizard Optimization Algorithm and the Henry Gas Solubility Optimization, to optimize the choice of the best hyperparameters to guarantee higher predictive accuracy and robust generalization. The novelty of our study is the dual purpose of Feature Importance analysis and the application of Receiver Operating Characteristic (ROC) analysis. The most significant predictors that can help us achieve sustainable fashion are offered by Feature Importance, and the tool that we need to decide the most suitable classification thresholds is offered by ROC analysis. Our experiment results prove that the hybrid schemes are better than the baseline schemes. LGHL variant, such as it is, is outstanding with an accuracy of 0.973, precision of 0.974, recall of 0.973, and F1-score of 0.973, where minimal variance occurs between the training and test sets. Such quantifiable findings highlight the effectiveness of our method and the ability to inform sustainable fashion through a data-driven analysis of eco-friendly fashion trends.</p>