Authors:
Cucu Ika Agustyaningrum
1
;
Haryani Haryani
1
;
Taufik Baidawi
1
;
Wahyudin Wahyudin
1
;
Siti Marlina
2
;
Artika Surniandari
1
and
Sucitra Sahara
1
Affiliations:
1
Fakultas Teknologi dan Informasi, Universitas Bina Sarana Informatika, Jakarta, Indonesia
;
2
Fakultas Teknologi Informasi, Universitas Nusa Mandiri, Jakarta, Indonesia
Keyword(s):
Algorithm, Conventional Machine Learning, Forest Fire, Method, Python.
Abstract:
A forest fire is a situation in which a forest is consumed by fire, damaging the forest’s products and causing harm to the environment and the economy. Finding out how frequently forest fires occur is the aim of forest fire prediction. The process of analyzing the data is therefore carried out using traditional machine learning techniques utilizing the Random Forest, Decision Tree, Logistic Regression, Nave Bayes, and Multilayer Per-ceptron methods. Knowing the accuracy and F1 score values allows for a comparison of this method using the Python programming language. The test results showed that the multilayer peceptron approach outperformed the Random Forest, Decision Tree, Logistic Regression, and Nave Bayes methods, with accuracy values of 86.70% and 87.93%, respectively, with a hidden layer size of 32.32. When compared to the other approaches investigated, the value of the multilayer perceptron method is quite prominent. This research can help determine the probability of forest f
ires.
(More)