SATISFACTION PREDICTION OF URBAN WATERFRONT SPACE LANDSCAPES USING CNNS

Urban waterfront spaces, as core carriers integrating the natural and human environment, possess both ecological regulation and public activity functions. Their quality directly impacts user satisfaction. However, current construction suffers from problems such as incoordination with the urban environment and insufficient ecological protection. Traditional satisfaction evaluation methods based on questionnaires and expert ratings are highly subjective and inefficient. This study aims to construct an efficient and accurate prediction model for urban waterfront space landscape satisfaction, providing technical support for spatial planning optimization. Through a multi-source data fusion strategy, landscape images, environmental parameters, user questionnaires, and behavioral data were collected to construct a comprehensive dataset. A prediction model was designed based on a convolutional neural network (CNN), optimizing key parameters such as kernel size and learning rate. Performance was compared with support vector machine (SVM) and random forest (RF) models. The results show that the constructed CNN model achieves a prediction accuracy of 87.3%, a coefficient of determination R2 of 0.891, and a root mean square error (RMSE) of 0.324, significantly outperforming traditional machine learning models. This study introduces deep learning technology into the field of urban waterfront space evaluation, improving the efficiency and objectivity of satisfaction assessment and providing new technical paths and decision-making basis for the sustainable development of urban waterfront spaces and urban renewal.