This study developed a housing price prediction model that integrates tabular data and satellite imagery to enhance prediction accuracy and model interpretability. A dataset named “Taiwan Housing Integrated Dataset (THID)” was constructed, comprising tabular data from Taiwan’s real estate registration website and high-resolution satellite imagery from the Google API. In the model design, machine learning techniques were applied to analyze features and predict housing prices based on tabular data. Simultaneously, deep learning models were employed to process high-resolution satellite imagery to capture geographical environmental features. Furthermore, a linear regression model was used to integrate predictions from both tabular data and satellite imagery. To improve model transparency, interpretability techniques were introduced to analyze the decision-making process of the model. Lastly, GPT technology was utilized to transform interpretability analysis results into easily understandable text. Key contributions of this study include 1) the integration of tabular data and satellite imagery for housing price prediction, leveraging a linear regression model to combine predictions from different data sources; 2) emphasizing model interpretability to enhance transparency in the prediction process, and 3) using interpretability techniques to generate explanatory text for housing price predictions.