🦠 Covid-19 Cases Prediction – Machine Learning Project

Author: Leon Motaung

Technologies Used: Python, Pandas, NumPy, Matplotlib, Plotly Express, Facebook Prophet, Scikit-learn

🔍 Objective

The objective of this project is to analyze the global spread of Covid-19 and predict the number of confirmed cases for the next 30 days using time-series forecasting. The project combines data analysis, visualization, and machine learning to understand trends and future patterns of the pandemic.

🚀 Steps I Took

• Created a Python virtual environment:

  python -m venv wenv

• Installed required libraries:

  pip install pandas numpy matplotlib plotly fbprophet scikit-learn

• Loaded global Covid-19 confirmed cases and deaths datasets.
• Cleaned and aggregated country-level data.
• Prepared time-series data for global daily cases and deaths.
• Visualized worldwide spread and daily trends.
• Applied Facebook Prophet to forecast cases for the next 30 days.

📊 Data Visualizations

🌍 Worldwide Spread of Covid-19

A geographical choropleth map was created to visualize the global distribution of Covid-19 cases across countries.

📈 Daily Global Covid-19 Cases

This visualization shows daily confirmed Covid-19 cases worldwide along with a 5-day moving average to highlight trends.

⚰️ Daily Global Covid-19 Deaths

This chart displays daily reported Covid-19 deaths globally, helping to understand mortality trends over time.

🤖 Covid-19 Cases Prediction (Next 30 Days)

Using the Facebook Prophet time-series forecasting model, the system predicts global Covid-19 cases for the next 30 days. The model captures weekly seasonality and provides confidence intervals for future predictions.

🧠 What I Learned

• How to clean and merge large real-world datasets.
• How to visualize global data using geographical maps.
• How to analyze time-series trends with moving averages.
• How to apply Facebook Prophet for forecasting.
• How to evaluate model performance using the R² score.

📌 Key Insights from the Analysis

📊 Global Trends Are Clearly Time-Dependent

Covid-19 cases and deaths follow strong time-based patterns, making time-series forecasting an effective approach.

📈 Moving Averages Smooth Daily Noise

Applying a 5-day moving average helps reduce daily fluctuations and reveals the true trend of infections and deaths.

🔮 Forecasting Provides Decision Support

The 30-day forecast offers valuable insight for planning and awareness, showing both expected case numbers and uncertainty bounds.

📂 Project Structure

• data/ – Covid-19 confirmed cases and deaths datasets
• scripts/ – Python scripts for analysis and prediction
• visuals/ – Generated plots and maps
• cases1.png – Daily global cases visualization
• cases2.png – Daily global deaths visualization
• cases3.png – 30-day Covid-19 cases forecast

🔥 This project strengthened my skills in data analysis, visualization, and time-series machine learning using real-world global health data.