An Integrated Statistical and Machine Learning Framework for Predicting the Dry Sliding Tribological Behavior of CFRP and GFRP Composites Under Environmental Aging and Loading Conditions

ABSTRACT Accurate prediction of the tribological behavior of fiber‐reinforced polymer composites is essential for assessing their performance under different service conditions. This study presents an integrated statistical and machine learning framework for analyzing the dry sliding tribological behavior of carbon fiber reinforced polymer (CFRP) and glass fiber reinforced polymer (GFRP) composites under different environmental conditions, aging durations, and applied loads. A previously published dataset comprising 60 observations was analyzed using exploratory data analysis, four‐factor analysis of variance (ANOVA), and five machine learning algorithms. Model performance was evaluated through hyperparameter optimization, nested cross‐validation, repeated k ‐fold validation, and leave‐one‐out cross‐validation (LOOCV), while feature importance and SHAP analyses were used for model interpretation. Aging duration was identified as the dominant factor affecting specific wear rate (SWR), whereas coefficient of friction (COF) was primarily influenced by reinforcing fiber material. Gradient boosting provided the best SWR predictions, while linear and ridge regression produced the most reliable COF predictions. The agreement between ANOVA and interpretable machine learning analyses demonstrated physically meaningful predictive relationships. The proposed framework integrates statistical inference, predictive modeling, and interpretability to support tribological assessment and data‐driven engineering decisions.