Fire prediction using Machine Learning Algorithms based on the confusion matrix

Résumé

Abstract In an earlier article, we outlined the process of developing a Machine Learning project that is often complex to establish, and that the problem must be broken down into several stages to facilitate its resolution [1]. We were able to identify 5 steps that we think are the most important to tackle such a project. These 5 steps are : the definition of the problem, the preparation of the data, the choice of the right algorithms, the optimization of the results and the presentation of the final results. In this manuscript, we will propose the application of different ways of evaluating classification models through an algorithm that predicts whether there is a fire in a given location or not. We are conscious that this problem is difficult to solve, especially when we have to predict the latter when there is none, do not predict when there is really one, or not predict when there is none. The method of this study will show how to choose the right algorithm and how to Evaluate it. The experiment shows promising results obtained thanks to the classification model algorithm and confusion matrix which provide fire detection accuracy around 92.71%.

Citer ce document

Accès au document

Texte intégral en lecture en ligne, réservé aux abonnés SPHAERO et aux membres de l'institution. Se connecter

Auteur(s)

Statistiques

Consultations : 1

Téléchargements : 0