abhichris63/Algerian_Forest_Fires_Prediction

Forest Fire Prediction Machine Learning Project Homepage Project is Created by Abhishek B CLICK HERE TO START AZURE APP * This is Forest Fire Prediction Project using **Data Science** and **Machine Learning**. * Source Dataset is stored in **MongoDB**. * Applied EDA (Exploratory Data Analysis), Feature Engineering, Feature Selection. * Model Building. * Deployment **Flask app** in **Microsoft Azure**. Dataset * Dataset from UCI Repository Algerian Forest Fire Dataset. * From this dataset (features with weather indexes) we will try to predict future fires by using machine learning algorithms in these regions. * It is always a good pratice to understand dataset first. * The dataset includes 244 instances and it is divided into 2 groups namely Bejaia region located in northeast of Algeria & Sidi-Bel-abbes region located in northwest of Algeria. * In this dataset the time period is from June 2012 to September 2012. Loading Dataset & inserting into MongoDB Database * Using Pandas Library the Dataset(CSV) file is loaded in Jupyter notebook as DataFrame. * DataFrame is converted to dictionary and inserted into MongoDB database. * 'pymongo' library is used to connect MongoDB Atlas. * 'client['Database_name]' is used to create database. * 'Database_name['Collections]' is used to create Collection/Tables. * 'collection.insert_many' (Tables in MongoDB are called **Collections**), records are called documents in MongoDB. EDA * Exploratory Data Analysis helps to analysis the dataset using pandas, numpy, matplotlib & seaborn. * EDA extracts insights from the dataset & provides the path for future forest fire predictions. * EDA helps to find feature importance which are more contributed in predicting Forest Fire. Feature Engineering * Feature Engineering is a process of selecting & transforming features into suitable format for machine learning model training. * Label Encoding technique is used in th …