Livability Assessment of Urban Residential Communities Based on Sentiment Analysis and Genetic Algorithm‐Improved Backpropagation Neural Network (GA‐BP) Model

Assessing the livability of residential communities is essential for human‐centric urban development. However, livability assessment suffers challenges related to weight determination and the inapplicability of machine learning owing to the lack of training samples with reliable livability values. Here, we propose a livability assessment model based on sentiment analysis and genetic algorithm‐improved backpropagation neural network (GA‐BP) for urban residential communities. A livability assessment index system for urban residential communities was developed based on six parameters: living cost, safety, environment, comfort, convenience, and transportation. The SHapley Additive exPlanations (SHAP) method was used to examine the impact of each index on livability. The case study results in Nanjing indicate that the proposed model scientifically analyzes livability, aligning with those of the official government livability rating standard. Moreover, the model effectively captures the complex relationships between livability and its contributing factors, demonstrating high practical relevance and application potential.