Manishrajmss13/Regression_project

A predictive machine learning model to forecast the Algerian Forest Fire FWI using Python, Scikit-learn, and Statsmodels. Includes complete data cleaning and EDA. # Algerian Forest Fire Prediction using Regression *A complete data science project in Python to predict the Fire Weather Index (FWI) using Scikit-learn and Statsmodels, with a focus on rigorous data cleaning and feature selection.* ## Project Goal The objective of this project was to develop a regression model capable of predicting the Fire Weather Index (FWI) using meteorological data from two regions in Algeria. This notebook walks through the entire process, from initial data wrangling and cleaning to model training and evaluation. ## Dataset Insights ### Data Source: Algerian Forest Fires Dataset from the UCI Machine learning Repository. ### Content: The dataset comprises 244 records of daily weather observations from the Bejaia and Sidi Bel-Abbes regions during the 2012 fire season (June-September). ### Challenge: The raw data was particularly challenging, arriving as a single file with embedded headers, inconsistent data types, missing values, and formatting issues like extra whitespace in column names and categorical values. ## My Approach: A Step-by-Step Breakdown This project followed a structured machine learning workflow to ensure robust and reliable results. ### 1. Data Cleaning & Preprocessing This was the most intensive part of the project. Key steps included: Structuring the Data: I first parsed the single file and programmatically split it into two distinct datasets, one for each Algerian region. Tidying Up: I corrected column names by stripping extra whitespace and dropped irrelevant header rows that were mixed in with the data. Handling Missing & Inconsistent Data: After identifying nulls, I used .dropna() for removal. I also cleaned up the categorical Classes feature (e.g., "fire " vs. "fire") and encoded it into a binary (1/0) format. Correcting Data Types: I converted all feature columns from their initial object type to the correct numerical int or float types to prepare for analysis. ### 2. Exploratory Data Analysis (EDA) …