Hybrid Feature Selection and Ensemble Classification for Climate Change Indicators: A Machine Learning Approach

Authors

Firas Ali Hashim, Department of Computer Science, College of Science, Mustansiriyah University, Baghdad, IraqFollow
Yasmin Makki Mohialden, Department of Computer Science, College of Science, Mustansiriyah University, Baghdad, IraqFollow
Nadia Mahmood Hussien, Department of Computer Science, College of Science, Mustansiriyah University, Baghdad, IraqFollow

Abstract

Climate change poses a significant challenge across various domains, necessitating advanced analytical techniques for effective monitoring and decision-making. This paper introduces a hybrid machine learning pipeline that selects and classifies features of climate change indicators. A synthetic dataset, representing different climate indicators and their application fields, is generated and processed using a combination of SelectKBest and Recursive Feature Elimination with Cross-Validation (RFECV) for robust feature selection. An ensemble of random forest and gradient-boosting classifiers is employed to enhance prediction accuracy. The predicted methodology has attained the lower initial accuracy of classification of 0.4 with five chosen climate characteristics. Optimizing the size of the dataset, hyper parameter tuning, and relating features to seven features raised the accuracy to 0.7 indicating increased efficiency and stability of the model. Important climate indicators like sea level, CO₂ level, ocean temperature, extent of ice and the level of drought were found to be most effective features that lead to precise forecasts. The pipeline’s results, including feature importance visualization, underscore its potential for improving climate science, policy-making, public awareness, agriculture, insurance, and urban planning initiatives. This study contributes to the growing field of climate informatics, providing a scalable and efficient framework for analyzing complex climate datasets.

Recommended Citation

Hashim, Firas Ali; Mohialden, Yasmin Makki; and Hussien, Nadia Mahmood (2025) "Hybrid Feature Selection and Ensemble Classification for Climate Change Indicators: A Machine Learning Approach," Terra Joule Journal: Vol. 1: Iss. 2, Article 8.
DOI: https://doi.org/10.64071/3080-5724.1021