🚴‍♂️ Demand Forecasting - Bike Rentals: The Streamlit App

This project is a Streamlit web app designed for visualizing and forecasting the bike rental demand in Washington, DC. The app provides an interactive interface for exploring bike rental data and weather trends and enables users to analyze and visualize bike-sharing trends.

🏠 Home Page

The home page provides an overview of the project, featuring an introduction to the app’s purpose and a fun bike GIF. The sidebar contains information about the app, allowing users to understand the project's context.

📊 Features

1. Bike Rental Data

• Time Span Selection: Users can select a time frame (daily, weekly, monthly) to visualize average bike rentals by station.
• Interactive Map: Using the pydeck library, the app displays an interactive map showing the average bike rentals per station.
  • The map includes options for zooming and panning, with markers colored and elevated based on the number of bikes rented.
• Best/Worst Performing Stations: Users can filter to see the best or worst-performing stations based on bike rental demand.
• Bar Charts: Data is visualized using Plotly bar charts, allowing users to explore bike rental trends at different stations.
• Progressive Line Chart: The app includes a dynamic line chart showing the total bike rentals over time, with real-time progress updates using a progress bar.

2. Weather Data

• Average Temperature By Month: Visualizes the average observed temperature across different months using Plotly line charts.
• Average Rainfall By Month: Similar to the temperature, this feature provides insights into monthly rainfall patterns, helping users understand how weather affects bike demand.

🔧 How It Works

1. Data Input: Users can select various options, such as the time frame or the number of stations to visualize.
2. Dynamic Visualizations: Depending on the user input, the app dynamically generates maps, bar charts, and line charts for an interactive data analysis experience.
3. Pydeck & Plotly Integration: The app leverages pydeck for rendering 3D maps and Plotly for creating aesthetically pleasing and interactive charts.

📂 Project Structure

demand_forecasting_streamlit_xgboost/
│
├── assets/                      # Contains images, GIFs, or other assets for the app
│   └── bike_2_giphy.gif          # Example GIF used in the app
│
├── data/                        # Data files used for visualizations and modeling
│   ├── combined_data.csv
│   ├── daily_avg_by_station.csv
│   ├── daily.csv
│   ├── monthly_avg_by_station.csv
│   ├── monthly.csv
│   ├── weekly_avg_by_station.csv
│   └── weekly.csv
│
├── models/                      # Directory for machine learning models
│   └── xgb_gs_best_estimator.pkl # Trained XGBoost model for forecasting
│
├── pages/                       # Streamlit page scripts for different parts of the app
│   ├── 2_📊_Visualizations.py    # Page for visualizing bike rental and weather data
│   └── 3_🚴_Forecast.py          # Page for visualizing forecasted bike demand using ML models
│
├── 1_🏠_Home.py                 # Home page of the Streamlit app
├── README.md                    # Documentation for the project
├── requirements.txt             # Dependencies for the project

📊 Libraries Used

• Streamlit: For creating the web app interface.
• Pydeck: For visualizing bike station data on interactive maps.
• Plotly: For creating bar and line charts.
• Pandas: For handling and analyzing data.
• Base64: For encoding media like GIFs.

⚙️ Installation

1. Clone the repository:

   git clone <repository-url>

2. Install the required dependencies:

   pip install -r requirements.txt

3. Run the Streamlit app:

   streamlit run 1_🏠_Home.py

📈 Data Sources

The bike rental and weather data used in the app are stored in the data/ folder and are dynamically loaded based on the user’s selection.

🌟 Future Enhancements

• Predictive Models: Incorporate machine learning models to forecast future bike rental demand based on historical data.
• More Weather Factors: Analyze the impact of other weather conditions such as wind speed or humidity on bike rental demand.

👨‍💻 Authors

• Hammad Faheem - Developer and Data Scientist

📜 License

This project is licensed under the MIT License - see the LICENSE file for details.

---

Let me know if you'd like any changes or additions!