kixng02/fleisKappa--201760630-M.S.-NKOSI

This is a public repository for my Postgraduate Diploma in ICT mini disertation tittled: A model for addressing AI algorithms biasness through indigenous South African knowledge systems. **Overview** This Streamlit application performs Fleiss Kappa analysis to measure inter-rater agreement among multiple AI models (ChatGPT, Gemini, Mistral AI) on indigenous plant classification tasks. The application provides statistical analysis and visualizations to assess the consistency and reliability of AI-generated content concerning Indigenous Knowledge Systems. **Features** - Statistical Analysis: Calculates Fleiss' Kappa coefficient with p-values and confidence intervals Multiple Data Sources - Pre-loaded research study data (20 indigenous plants) - User-provided custom classification data Visualizations - Agreement patterns heatmap - Detailed classification heatmap - Results bar charts Data Management - Input validation and duplicate prevention - Session state management - CSV export functionality **Dependencies** # Core dependencies pip install streamlit pip install pandas pip install numpy # Visualization and statistical dependencies pip install matplotlib pip install seaborn pip install scipy # Optional: For enhanced data handling pip install openpyxl # For Excel file support Installation & Execution 1. Installation & Execution pip install -r requirements.txt 2. Run the application streamlit run fleiss_kappa_app.py 3. Access the application - Open your web browser to localhost - The application will automatically load with study data analysis **Usage Instructions** 1. Study Data Analysis i. Default view shows original research data (20 indigenous plants) ii. Click "Run Fleiss Kappa Analysis on Study Data" to generate results iii. View comprehensive statistics including - Fleiss' Kappa coefficient - P-value and statistical significance - Observed vs. expected agreement rates - Interpretation based on Landis & Koch scale 2. User Data Input i. Switch to "Insert Your Own Data" in the navigation sidebar ii. Add plant classifications - Enter unique plant name (required) - Select classification from each AI model (Med …