mirfan57/algerian_forest_fire

# Algerian Forest Fire Prediction **User Interface** ## Heroku App Link # Demo user-images.githubuserconte… ## A brief explanation of the project's objectives. Forest Fire Prediction is a Supervised Machine learning problem statements. Using Regression and Classification Algorithm, Regression and Classification Model is build that detected future fires based on certain Weather report. ## Libraries Implemented **Data Pre-Processing** - Numpy, Pandas, Matplotlib, Seaborn **Feature Selection** - Variance Inflation Factor **Model Building** - Sklearn, statsmodels **Hyperparameter Tuning** - Randomized SearchCV, Grid SearchCV **Model Selection** - Repeated Stratified KFold ## About the Dataset **Algerian Forest Fires Dataset** I used a UCI dataset on forest fires in Algeria. The **Bejaia and Sidi Bel-abbes** areas of Algeria are represented by the dataset's observations and records on forest fires are loaded. This dataset's time period runs from **June 2012 to September 2012**. In this case, we investigated if a few machine learning algorithms might accurately predict forest fires in certain locations using specific weather information. **Dataset taken from:** Link ***Data Set Information:*** - The dataset contains 244 occurrences that aggregate data from two locations of Algeria: the **Sidi Bel-abbes** region in northwest Algeria and the **Bejaia** region in northeast Algeria. - 122 instances for each region. - The data collected in a span of June 2012 to September 2012. - The dataset includes a total of 12 attributes including 11 feature attributes and 1 output attribute depending on the task performed. - The 244 instances have been classified into **fire** (138 classes) and **not fire** (106 classes) classes. **Attribute Information:** **1. Date :** (DD/MM/YYYY) Day, month ('june' to 'september'), year (2012) **Weather data observations:** **2. Temperature :** …