A Machine Learning Approach to Forest Fire Prediction Through Environment Parameters

Forest fire is serious problem throughout the world and environment factors are one among the many of the influencing factors to cause forest fire. This paper is focused on predicting forest fire through environment parameters. Every year mass destruction of flora and fauna takes place due to forest fire as it destroys huge area of forest. Artificial intelligence helps to predict the future and using it in predicting forest fires can successfully save wildlife and resources. Oxygen, temperature and humidity are 3 factors on which fire essentially depends. This system predicts the forest fire based on given environment factors such as oxygen, humidity and temperature. The key algorithm used for prediction is logistic regression, support vector machine and multiple linear regression. Accuracy, precision, recall and f-measure are calculated and compared between these models. The system is implemented with given algorithm using only three environment parameters and the proposed method can give robust outcome with small set of data sets.