Abstract :

Regional Weather Forecasting Is Critical For Agriculture, Disaster Management, And Environmental Planning. Traditional Numerical Weather Prediction (NWP) Models Require High Computational Resources And Often Fail For Localized Predictions In Geographically Diverse Regions. This Paper Presents An Intelligent Regional Weather Prediction System Leveraging A Hybrid Micro–Macro Data Fusion Strategy Combined With A Random Forest (RF) Ensemble Regression Algorithm. Historical Meteorological Observations—including Temperature, Humidity, Rainfall, Wind Speed, Dew Point, And Atmospheric Pressure—from IoT Weather Stations And ERA5 Reanalysis Grids Are Preprocessed Through A Structured Pipeline Comprising Seasonal Imputation, Winsorization, Min-Max Normalization, And Multi-criteria Feature Selection (Pearson Correlation, RF Gini Importance, And Recursive Feature Elimination). The Proposed RF Model Is Benchmarked Against Linear Regression, Decision Tree, And Gradient Boosting On Six Years Of Andhra Pradesh Meteorological Data. Experimental Results Demonstrate That RF Achieves 73.94% Prediction Accuracy With RMSE=3.87 And MAE=2.94 (R²=0.76), Outperforming All Baselines. Feature Importance Analysis Identifies Temperature History, Relative Humidity, And Atmospheric Pressure As The Dominant Predictors. The Framework Is Scalable, Interpretable, And Suitable For Real Time Deployment In Agriculture Advisory, Disaster Warning, And Smart City Applications.

Published:

01-5-2026

Issue:

Vol. 26 No. 5 (2026)

Page Nos:

23-29

Section:

Articles

License:

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.

How to Cite