WilliamZLee/Algerian-Forest-Fire-Analysis

R Course Design # Algerian-Forest-Fire-Analysis R Course Design Executive Summary Purpose: The purpose of this report is to identify factors that can predict the occurrence of forest fires by conducting an analysis of the various weather conditions. It also recommends measures to combat them. The analysis is conducted by using different testing methods and models. Key Descriptive Statistics: The data contains 15 features, including 3 classification factors: month, region and classes. Numeric variables consist of 4 weather variables and the other 6 variables calculated from weather variables. To avoid overfitting, we only take fine fuel moisture code (FFMC), duff moisture code (DMC), drought code (DC) and initial spread index (ISI) and the 4 key weather variables to use in further model analysis. Main Result and Inferences: With outcomes from multiple machine learning models, the prediction outcome from the random forest model is best fit. It indicated that the velocity of wind spread (Ws), relative humidity (RH), Drought Code (DC), Initial Spread Index (ISI) are significantly contributing to the occurrence of forest fires in the two distinct regions of Algeria. Recommendations and conclusion: By gathering insights of outcomes from models we have undergone, we recommend Algerian authorities to enhance their weather monitoring capabilities by deploying more sensors and forecasting devices, particularly during the summer months. This would enable better tracking of temperature, wind, relative humidity, and precipitation, crucial for fire risk assessment. Secondly, we propose utilizing this weather data to calculate the Fire Weather Index (FWI), which quantifies fire intensity potential. When FWI reaches the hazardous level of 5, prompt action should be taken, including deploying additional fire control and rescue forces to prepare for and mitigate potential disasters.