Normalized data-driven weather forecasting using hybrid machine learning algorithm with extractive features

Accurate weather forecasting is essential for various real-world applications, yet traditional Numerical Weather Prediction (NWP) models are computationally intensive and often struggle with large-scale data. To address these limitations, this paper proposes a Normalized Data-driven Weather Forecasting using Hybrid Deep Learning (NDWFHDL) model. The framework comprises four stages such as data acquisition, preprocessing, feature extraction, and forecasting. Weather data collected from Internet of Things (IoT) sensors are preprocessed using a Modified Tanh (MTanh) normalization technique to improve data quality. Feature extraction combines Regularized Hybrid Deng Entropy (RHDE), statistical features, and raw data representations. Finally, an Improved ShuffleNet (IShuffleNet) integrated with a Bidirectional Long Short-Term Memory (BiLSTM) network captures spatial and temporal dependencies for accurate forecasting. Experimental results demonstrate superior performance over Bi-LSTM, LinkNet, EfficientNet, and DCNN, achieving a maximum specificity of 95.2% and precision of 96.4% using 90% of the training data, confirming the effectiveness and robustness of the proposed approach.