Deep learning assisted data visualisation framework for real time data interpretation and predictive analytics

Abstract

Rapid industrialization and urbanization have significantly increased the importance of air quality prediction for environmental monitoring, public health protection, and intelligent urban management. However, conventional forecasting methods often struggle to capture the complex spatial and temporal characteristics of environmental data, resulting in limited predictive accuracy. This study proposes a deep learning-assisted data visualization framework for real-time air quality interpretation and predictive analytics by integrating Convolutional Neural Networks (CNN) and Long Short-Term Memory (LSTM) networks. The Beijing Multi-Site Air Quality Dataset was utilized, followed by data cleaning, missing value imputation, Min–Max normalization, feature representation, and dataset partitioning in a 70:15:15 ratio for training, validation, and testing, respectively. The CNN extracts high-level spatial features, while the LSTM captures temporal dependencies to enhance forecasting performance. Experimental results demonstrate that the proposed hybrid CNN–LSTM model outperforms the standalone CNN and LSTM models, achieving an MSE of 11.38, MAE of 2.41, RMSE of 3.37, and an R² value of 0.972. Furthermore, the model achieved an MAE of 2.18, RMSE of 3.05, and an R² of 0.981 for 1-hour-ahead forecasting, demonstrating its effectiveness for accurate real-time air quality prediction and intelligent visualization-based decision support.