With the growing trend of developing and applying innovative bio-based materials in the construction industry, evaluating their performance across various building typologies and climatic conditions is crucial for their wider adoption. Traditional approaches to building energy simulation often require a large number of scenarios, leading to an exponential increase in simulations that demand significant computational resources and time. These challenges can be mitigated through data-driven methods. This study presents a predictive modelling framework applied to the Kádár-kocka, one of the most representative residential building types in Hungary’s building stock. Multiple regression algorithms were implemented for predictive modelling to enable rapid estimation of key performance indicators related to energy demand and thermal comfort. The models’ robustness was evaluated through a comprehensive set of evaluation metrics, including mean squared error (MSE), mean absolute error (MAE), root mean squared error (RMSE) and coefficient of determination (R2). Using state-of-the-art simulation tools such as EnergyPlus and Grasshopper, a diverse range of renovation scenarios was developed, serving as data sources for predictive model development. The simulations considered a range of factors, including building envelope characteristics, shading, orientation and occupant behaviour. The building envelope scenarios were based on the Existing State, Usual Refurbishment and Advanced Refurbishment options provided by the Tabula Webtool, alongside solutions developed as part of the BIO4EEB project. The results highlight the scalability of predictive modelling in guiding sustainable refurbishment strategies.