With the vision of new urbanization and the continuous advancement of the renewal and renovation of existing urban areas and building energy efficiency improvement projects, large-scale energy-saving renovation of building clusters has become an important content of eco-friendly and low-carbon as well as an important means to improve the energy efficiency level of the region. This study proposes a building energy consumption information model adapted to data-driven analysis and a predictive stochastic model with an improved Monte Carlo method. Focusing on the calibrated estimation method based on Bayesian inference model, the estimation of the parameters to be estimated is generated by setting the prior distribution of the parameters to be estimated, deriving the posterior distribution, and simulating the Gibbs sampling based on the MCMC algorithm. The Monte Carlo method is applied to simulate and analyze the whole life cycle cost of the building complex, and the predictive stochastic model is also verified by the energy-saving renovation of the building complex. From the experimental results, it can be obtained that the total cost of the traditional house is 595,582 yuan, while the total cost of the eco-house is 564,251 yuan, i.e., compared with the traditional house, the eco-house can save more money in the process of using the house. In the validation comparison between the output of the prediction model and the actual data before and after remodeling, the average energy savings before and after remodeling both deviate by 22.22%, and the deviation of the energy saving rate is larger than the deviation of the total energy consumption. In summary, this paper proposes an improved predictive stochastic model of Monte Carlo method to provide theoretical guidance for the prediction, decision-making and optimization of energy-saving retrofit of building clusters.