<p>This study experimentally investigates the thermoelectric performance of a series-connected photovoltaic-thermal (PVT) system employing a nanofluid-based direct absorption solar collector. The impact of ambient temperature, solar radiation, and working fluid (water and silver nanofluid) on key system performance metrics, such as generated electrical power and outlet fluid temperature, is assessed. The results reveal a significant enhancement in electrical power generation when using silver nanofluid. Specifically, a 500 ppm concentration of silver nanofluid yielded an 8.2% increase in maximum electrical power compared to using water alone. Furthermore, a 1000 ppm silver nanofluid resulted in a 7.8% and 3.5% improvement in average total thermal efficiency compared to 500 ppm nanofluid and water, respectively. The study also explored the application of machine learning models (ANN, CatBoost, XGBoost, and RF) to predict the system's electrical power, outlet temperature, thermal efficiency, and electrical efficiency. Model performance was evaluated using <i>R</i>², MSE, MAPE, MAE, and WI. The RF, CatBoost, and XGBoost models demonstrated acceptable accuracy, with XGBoost exhibiting the best performance for electrical power prediction (<i>R</i>² = 99.53%, MAPE = 0.0079). CatBoost excelled in predicting outlet temperature (<i>R</i>² = 99.65%, MAPE = 0.0046) and electrical efficiency (<i>R</i>² = 97.06). RF performed best with thermal efficiency prediction (<i>R</i>² = 93.32%, MAPE = 0.117). ANN consistently showed lower accuracy and higher error rates across all parameters.</p>