HermelaDev/African-HealthBurden-Time-Analysis-Forcasting-2003-2023

# African-HealthBurden-Analysis-2003-2023 ## 1. Introduction The aim of this project is to analyze temporal patterns in health outcomes across African countries using the Global Burden of Disease (GBD 2023) dataset. The study focuses on two key population-level measures: - DALYs (Disability-Adjusted Life Years) - Deaths The analysis includes country-level aggregation, age-structured analysis, cause-level decomposition, and exploratory modelling to understand long-term trends, relationships between indicators, and potential forecasting applications. ## 2. Data Preparation and Cleaning ### 2.1 Loading and standardizing the main dataset The main GBD dataset (“IHME-GBD_2023_DATA…csv”) was loaded and inspected. Measure names were standardized to ensure consistency (e.g., mapping "DALYs (Disability-Adjusted Life Years)" → "DALYs"). ### 2.2 Pivoting into country-year format To enable time-series modelling, the dataset was reshaped into a wide format with columns: - val_DALYs, val_Deaths - upper_DALYs, upper_Deaths - lower_DALYs, lower_Deaths This was achieved using a multi-index pivot: year × location × sex × age × cause × metric. ### 2.3 Aggregation to country totals To obtain a clean country-level time-series, values were summed across: - all sexes - all age groups all causes This produced a dataset (simple_df_total) with yearly totals for each country from 2003-2023. ## 3. Country-Level Time Series Patterns ### 3.1 Total deaths over time A multi-country line chart revealed that: - Some countries have steadily increasing deaths over time. - Others exhibit declines or oscillations depending on conflict, disease outbreaks, or healthcare improvements. - The annual frequency (dtick=1) confirms the dataset is annual - consistent with ARIMA-type modelling. ### 3.2 DALYs time trend Similarly, DALYs show long-term increasing or decreasing patterns depending on country. These series often contain: - monotonic trends - structural shifts - non-stationary behaviour …