Srujana767/Currency_Watch

Currency Watch: Analysis of African currencies against USD # Currency Watch ## Analysis of African Currencies Against USD Currency Watch is a data analysis project that studies the movement of four African currencies against the US Dollar from 2023 to 2025. ### Currencies - USD/ZAR – South African Rand - USD/EGP – Egyptian Pound - USD/NGN – Nigerian Naira - USD/KES – Kenyan Shilling ## Focus Question Which currency shows a steady long-term slide versus which shows sudden shocks? What is the practical difference for a business holding that currency? ## Project Objectives - Clean and harmonize raw exchange-rate data - Handle missing values, duplicates, and outliers - Calculate daily returns - Calculate rolling volatility - Compare currency performance - Visualize currency trends - Display percentage changes on a geographical map - Build an interactive Streamlit dashboard ## Data Cleaning The dataset contained mixed date formats, inconsistent currency-pair names, missing values, duplicate records, and extreme outliers. The following steps were performed: - Standardized currency pair names - Converted dates into a consistent datetime format - Removed duplicate records - Converted closing prices to numeric values - Detected extreme values using the IQR method - Handled missing and invalid closing prices - Merged currency data with annual inflation and GDP growth data ## Key Findings - USD/NGN showed the highest volatility among the four currencies. - USD/KES showed the lowest volatility. - USD/NGN experienced the largest overall depreciation against the US Dollar. - USD/EGP showed significant sudden exchange-rate movements. - USD/ZAR showed moderate movement compared with NGN and EGP. ## Technologies Used - Python - Pandas - NumPy - Matplotlib - Folium - Streamlit - Jupyter Notebook ## Project Files - `currency_analysis.ipynb` – Data cleaning and analysis - `app.py` – Streamlit dashboard - `currency_set.csv` – Raw currency dataset - `web_set.csv` – Macroeconomic dataset - `screenshots/` – Project screenshots # …