This study presents a systematic machine learning approach to predict the thermal conductivity of water-based phase change materials (PCMs) enhanced with various carbon-based nanoparticles, including graphene nanoplatelets, multi-walled carbon nanotubes, activated carbon, graphitized mesoporous carbon, and natural graphite flakes. Four models—artificial neural network, extreme gradient boosting, categorical boosting, and generalized regression neural network—were developed using 329 datasets reported in the literature. An initial heatmap correlation analysis revealed a strong positive correlation (+ 0.602) between the thermal conductivity of the base PCM and that of the PCM nanocomposite, while temperature exhibited a moderately negative correlation (− 0.4162). Among the evaluated models, extreme gradient boosting achieved the highest prediction accuracy, with a coefficient of determination of 0.9961, a mean absolute error of 0.0369, and a root mean square error of 0.0609. The influence of dataset size on model performance indicated that larger datasets (> 60 samples) produced more consistent results, with ensemble models (extreme gradient boosting and categorical boosting) outperforming the generalized regression neural network and artificial neural network. Furthermore, the extreme gradient boosting-predicted values were compared with conventional correlations, which exhibited mean absolute errors ranging from 7.15 % to 35.02 %, whereas the extreme gradient boosting model demonstrated significantly improved predictive accuracy, with mean absolute errors between 0.051 % and 0.2691 %. Finally, a comparison with previously reported machine learning-based predictions of thermal conductivity for water nanofluids confirmed that the proposed extreme gradient boosting model provides superior reliability and robustness. These results highlight that ensemble-based algorithms can serve as powerful predictive tools for complex heat transfer systems and can assist in the design and optimization of nanostructured PCMs for thermal energy storage applications.