phy-guanzh/Disasters_Classification

The project is about analyzing disaster data for a few countries in Africa and South America. The data files for 13 countries are collected from ‘desinventar.xn--net-to0a. The project goes through 5 stages: Data Cleaning and Feature Engineering, Data Visualization, Principles Component Analysis, Regression and Classification. # Disaster Classification Analysis This repository contains datasets and analysis scripts for exploring disasters across Africa and South America. It covers data cleaning, visualization, feature engineering, dimensionality reduction (PCA), and classification. --- ## Folder Structure ### Data Folders - **`original_data/`** Contains the originally downloaded datasets, divided into: - `Africa/` - `South_America/` - **`cleaned_data/`** Contains datasets after **Data Cleaning** and **Feature Engineering**. - **`pca_data/`** Contains datasets including PCA results and the datasets used for classification. --- ## Script Descriptions ### 1. `Africa_combined.Rmd` This script includes: - **Data Cleaning** - **Data Visualization** - **PCA Analysis** - **Regression Analysis**: - Using the **originally cleaned dataset**. - Using the **PCA-transformed dataset**. Applied to **African countries**. --- ### 2. `South_America_combined.Rmd` This script includes: - **Data Cleaning** - **Data Visualization** - **PCA Analysis** - **Regression Analysis**: - Using the **originally cleaned dataset**. - Using the **PCA-transformed dataset**. Applied to **South American countries**. --- ### 3. `Classification_binary.Rmd` This script performs **binary classification** to predict the **CONTINENT** where a disaster occurred (Africa or South America). --- ### 4. `Classification_multi.Rmd` This script performs **multi-class classification** to predict the **DISASTER CATEGORY**, which includes 17 different types of disasters. --- ## How to Run ### Prerequisites 1. Install **R** (version 2024.04.2+764 or later). 2. Install **RStudio** (recommended for running `.Rmd` files).