r-bitbot/GAM-Prediction-in-Somalia

# Global Acute Malnutrition (GAM) Prediction in Somalia This project aims to predict the Global Acute Malnutrition (GAM) prevalence in Somalia using machine learning methods. Leveraging data from Somalia Food Security and Nutrition Analysis Unit (FSNAU) for district level forecasting of global acute malnutrition. The goal is to provide insights for actionable intervention and prepare for rises in GAM. ## Objectives ## To Run Locally The project is designed for local Jupyter Notebook executiom. To run it locally: ``` git clone github.com cd GAM-Prediction-in-Somalia jupyter notebook ``` Ensure you have all the libraries installed (via pip or conda). ## Tools and Libraries Used * Python 3.7 and above * Jupyter Notebook * pandas: For data manipulation and cleaning * numpy: For numerical operations * scikit-learn: For machine learning (Random Forest, Gradient Boosting) * matplotlib and seaborn: For data visualization * requests and urllib: For data fetching from FSNAU APIs ## Methodology ### Data Collection From FSNAU, the datasets of GAM, rainfall, NDVI, incidents, fatalities, and crops were manually taken from the duration of January 2017 to December 2019. ### Data Preprocessing Missing data was handled and datasets were cleaned up using Pandas. Normalization and feature engineering were also leveraged. ### Exploratory Data Analysis Trends were visualized using predictors such as monthly and annual factors of rainfall, NDVI, incidents, fatalities, and crop production. Analysis of each district and region was done by leveraging graphs drawn against monthly and annual GAM to see which factors affected GAM values. ### Feature Selection Environmental factors like rainfall, NDVI, and crop production was taken into account. Conflicts like incidents and fatalities were included. Previous months' malnutrition trends were also noted, along with lagged variables of all of the above. ### Models The XGBoost model was trained …