i72bad/Forest-Fire-Prediction-EDA

Predicting the occurrence of forest fires in Algeria using environmental data, Logistic Regression, and Decision Tree classifiers. # Algerian Forest Fires Prediction ## Project Overview Forest fires pose a significant environmental and economic threat in Algeria, where hot and dry summer conditions create ideal environments for fires to spread. This project utilizes environmental and meteorological data to develop a binary classification framework to predict the occurrence of forest fires. Accurate prediction models hold significant importance for ecological preservation, resource deployment, and early-warning systems. To address this challenge, this project tests two supervised learning models: * **Decision Tree Classifier:** Chosen for its interpretability and ability to handle non-linear patterns. * **Logistic Regression:** Used as a robust and widely-used baseline for binary classification problems. ## The Dataset The model is built on the Algerian Forest Fires Dataset, which contains 243 instances collected between June and September from 2002 to 2013. The data was collected from two high-risk regions: Bejaia and Sidi Bel-Abbes. ### Features The dataset records various environmental and meteorological conditions, classifying each record as either "fire" or "not fire". The key features include: * **Day & Month:** Categorical data representing the date of the record. * **Temp:** Maximum temperature at noon in Celsius (ranging from 22 to 42). * **RH:** Relative Humidity percentage (ranging from 21 to 90). * **Ws:** Wind speed in km/h (ranging from 6 to 29). * **Rain:** Total daily rainfall in mm (ranging from 0 to 16.8). * **FWI System Indexes:** * Fine Fuel Moisture Code (FFMC) * Duff Moisture Code (DMC) * Drought Code (DC) * Initial Spread Index (ISI) * Buildup Index (BUI) * Fire Weather Index (FWI) * **Target:** Binary classification indicating "Fire" or "not Fire". ## Exploratory Data Analysis (EDA) The project includes a preliminary exploratory analysis to understand the data distribution. Visualizations include: * A pie chart detailing the class distribution between fire and non-fi …