zaghloul2012/Predicting-Forest-Fire-in-Algeria

In my project, I aim to develop a more accurate system for forest fire prediction by testing and comparing various classification models # Predicting Forest Fire in Algeria ## Introduction Forest fire is a critical concern over the world, each year millions of hectares are destroyed all around the word. Algeria is one of the countries affected by this phenomenon, mainly in summerForest fires cause economic and ecological damage, as well as human life threat. Predicting forest fires is crucial to mitigate this threat. You can read more through the paper puplished in 2020 on Predicting Forest Fire in Algeria Using Data Mining Techniques: Case Study of the Decision Tree Algorithm ## Motivation In my project, I aim to develop a more accurate system for forest fire prediction by testing and comparing various classification models. In addition to the decision tree-based system proposed in existing literature, I will assess the performance of logistic regression, k-nearest neighbors, support vector machine, kernel SVM, Naive Bayes, random forest classification, and other classification models. To identify the most significant features for predicting forest fires, I will also conduct tests based on the relationship between the independent variables (features) and the dependent variable (target). Specifically, I will utilize box plots to identify the variables that may be the best predictors and also conduct the Wald significance test. ## Dataset The dataset includes 244 instances that regroup a data of two regions of Algeria,namely the Bejaia region located in the northeast of Algeria and the Sidi Bel-abbes region located in the northwest of Algeria. 122 instances for each region. Data can be downloaded from here ## Model Development To select potential predictor features for our target variable, we first examined the relationship between each independent variable and the target variable using boxplots. We looked for significant variations in mean, median, and distribution between the two categories. Next, we developed different classification models to assess their suitability for our case. These mo …