sai-manas/FWI_Predictor_ML

Fire Weather Index (FWI) - Web App Predictor: Algerian Forest Fires dataset. Using Ridge, Lasso, Linear Regression, and ElasticNet models. Deployed as a Flask app on AWS Elastic Beanstalk. Explore the prediction insights for fire risk assessment. # Fire Weather Index (FWI) Web App Predictor: Algerian Forest Fires dataset ## Overview FWI predictor web application predicts the Fire Weather Index (FWI) for Algerian forest fires using various regression models, including Ridge Regression, Lasso Regression, Linear Regression, and ElasticNet. The model selection process involved thorough comparison and cross-validation, ultimately determining Ridge Regression as the optimal choice due to its superior accuracy. The model is trained on a dataset encompassing two regions of Algeria—Bejaia in the northeast and Sidi Bel-abbes in the northwest. The dataset spans from June to September 2012 and includes various weather-related attributes such as temperature, relative humidity, wind speed, rain, and components of the FWI system. ## Jupyter Notebook The detailed analysis and code implementation are available in the Jupyter notebook. ## Screen Recording of FWI predictor application deployed on aws github.com ## Dataset Information - **Instances:** 244 (122 for each region) - **Attributes:** 11 weather-related attributes and 1 output attribute (FWI) ## Technologies Used - **Programming Language:** Python - **Libraries:** Pandas, NumPy, Seaborn, Matplotlib, Scikit-learn , Pickle, Warnings - **Web Framework:** Flask - **Frontend:** HTML, CSS - **Deployment:** AWS Elastic Beanstalk, AWS CodePipeline ## Data Cleaning and Preprocessing The dataset underwent rigorous cleaning, addressing missing values, mispaced entries, and data type conversions. Regions were added as a new column, and unnecessary features were dropped. Categorical values were cleaned, and the dataset was saved as a cleaned CSV file. ## Exploratory Data Analysis (EDA) EDA was performed to analyze the distribution of fires across different months and regions. August emerged as the month with the highest number of fires in both regions. Visualization tools like bo …