ChimbuezeDavid/NairaPulse-AI

A web-based simulator for forecasting price dynamics for inflation-driven consumer goods in Nigeria using a stacked ensemble model # Consumer Goods Price Dynamics A comprehensive analytics and forecasting tool for monitoring and predicting consumer goods price dynamics. This project uses machine learning models (SARIMAX, Random Forest, XGBoost, and LSTM) to analyze CPI data and forecast future trends across various categories like Food, Health, Transport, etc. ## Features - **Exploratory Data Analysis (EDA):** Visualize CPI trends and distribution across different categories. - **Forecasting:** Predict future price indices using advanced time series and machine learning models. - **Model Comparison:** Compare predictions from multiple models (SARIMAX, RF, LSTM, XGBoost Stacked). - **Interactive Dashboard:** Built with Streamlit for an easy-to-use user experience. - **Progressive Web App (PWA):** Fully configured with a service worker (`sw.js`), web app manifest (`manifest.json`), and app icons for offline caching and installation support. - **Clean Embedded & Standalone UI:** Streamlined interface with Streamlit watermarks, footer, and toolbar chrome hidden natively. The sidebar navigation remains fully responsive and accessible. ## Installation 1. Clone this repository: ```bash git clone cd "Consumer Goods Price Dynamics" ``` 2. Create a virtual environment: ```bash python -m venv venv source venv/bin/activate # On Windows: venv\Scripts\activate ``` 3. Install dependencies: ```bash pip install -r requirements.txt ``` ## Usage Run the Streamlit app: ```bash streamlit run app.py ``` ## Progressive Web App & Embedded View Details ### 1. PWA Configurations Static file serving is enabled in the configuration file .streamlit/config.toml: ```toml [server] enableStaticServing = true ``` This serves `manifest.json` and `sw.js` under the path `/app/static/` so the browser can discover and install the application as a standalone PWA. ### 2. Standalone URL & IFrame Embedding * **Direct Access:** Opening the app directly (e.g. `localhost`) redirects automatically to the `?embed=tru …