Predicting the Presence of Amphibians Near Road Construction Sites Using Emerging Machine Learning Algorithms

Abstract

The construction of dense road networks exerts a drastic influence on the persistence of amphibian species inhabiting the adjacent areas. Preventing any arising conflicts between nature conservation and urbanization is vital. We suggest an efficient system to predict the existence of amphibians in the vicinity while constructing roads and planned infrastructure projects. This model uses the XGBoost framework. Moreover, we implement various classification techniques such as XGBClassifier with GridSearchCV and without GridSearchCV, Naive Bayes Classifier, Decision Tree, KNN Classifier, SVM, and RidgeClassifier and compare their performances. Comparative review of these classifiers shows that XGBClassifier with GridSearchCV outperforms the other classification algorithms with high accuracy. The factors thus identified should be taken into account for sustainable urban planning.