Gastronalytics

Inspiration:

The inspiration for Gastronalytics stems from the challenges faced by restaurants in optimizing their operations, understanding customer behavior, and adapting to external factors like weather and holidays. With TouchBistro's rich dataset of bill-level transactions and venue details, we saw an opportunity to create a data-driven solution that empowers restaurateurs to make smarter decisions. From staffing recommendations to dynamic pricing strategies, Gastronalytics aims to provide actionable insights that enhance operational efficiency, improve customer satisfaction, and drive revenue growth.

What it Does:

Gastronalytics is a Restaurant Insights Web App designed to help restaurateurs streamline operations and make data-driven decisions. Key features include:

Dynamic Opening Hour Optimization : Suggests optimal opening and closing hours based on peak order data and seasonal trends.

Smart Menu Pricing : Analyzes regional sales data to recommend competitive price ranges for menu items.

Peak Hour Staffing Recommendations : Identifies busy periods and suggests optimal staffing levels to improve service efficiency.

Weather-Based Demand Forecasting : Predicts sales fluctuations based on weather conditions (e.g., increased delivery orders during rain).

Customer Spend Insights : Examines how dine-in vs. takeout impacts revenue and tips, helping restaurants prioritize strategies.

AI-Generated “Golden Hour” Deals : Identifies slow periods and suggests limited-time discounts to boost traffic.

*Waiter Performance Tracker *: Ranks waiters based on tip percentages and bill sizes to optimize staff training and motivation.

Inflation Impact Analysis : Evaluates how inflation affects restaurant pricing and customer spending patterns.

By integrating advanced analytics, machine learning, and external data sources (e.g., weather APIs, holiday calendars), Gastronalytics provides a comprehensive view of restaurant performance and external influences.

How We Built It

Data Preparation: Cleaned and preprocessed TouchBistro’s bill-level dataset using Modin for efficient data handling. Aggregated transaction timestamps, order types, and payment details to derive actionable insights.

Feature Engineering :

Created derived features such as hour_of_day, season, and tip_percentage.

Segmented data by city, concept, and venue type to uncover granular trends.

Machine Learning Models : Built time-series forecasting models (e.g., ARIMA, Prophet) to predict daily/weekly sales. Used clustering algorithms to identify peak periods and slow hours.

External Data Integration : Fetched real-time weather data using OpenWeatherMap API. Integrated public holiday datasets to analyze event-based sales trends.

Visualization :

Developed interactive dashboards using Plotly and Matplotlib to visualize insights. Included heatmaps, bar charts, and trend lines for intuitive data exploration. Backend and Deployment : Built the backend using Flask for API integration and data processing. Deployed the app on Heroku for easy access and scalability.

Challenges We Ran Into

Handling Large Datasets : Processing large datasets efficiently required leveraging Modin and Ray for parallel computation. Temporary files created during failed installations consumed significant disk space, which was resolved by cleaning up /tmp and cache directories. Integrating External APIs : Fetching real-time weather and holiday data posed challenges in terms of API rate limits and data consistency. Implemented caching mechanisms to minimize redundant API calls. Package Configuration Issues : During dependency installation, some packages (e.g., pyarrow) raised warnings about missing configurations. Addressed these by explicitly specifying package paths and using find_namespace_packages.

Accomplishments That We're Proud Of:

Comprehensive Feature Set : Successfully implemented a wide range of features, from operational optimization to external factor analysis, providing a holistic solution for restaurateurs. Scalable Architecture : Designed the app to handle large datasets efficiently using Modin and Ray , ensuring smooth performance even with extensive data. Real-Time Insights : Integrated real-time weather and holiday data to provide dynamic, actionable insights that adapt to changing conditions. User-Friendly Interface : Created intuitive dashboards and visualizations that make complex data accessible and actionable for non-technical users.

What I Learned

Data Preprocessing : Cleaning and preprocessing large datasets is a critical step that requires careful planning and efficient tools like Modin . Dependency Management : Installing complex libraries like pyarrow can be challenging, but understanding underlying dependencies (e.g., cmake) is key to resolving issues. API Integration : Working with external APIs taught us the importance of caching and handling rate limits to ensure consistent data availability. Parallel Computing : Leveraging Ray for parallel computation significantly improved processing speed and scalability. Problem-Solving Under Pressure : The hackathon environment pushed us to think creatively and solve problems quickly, enhancing our ability to deliver under tight deadlines.

What's Next for Gastronalytics

Enhanced Machine Learning Models : Incorporate more advanced models (e.g., deep learning) for demand forecasting and anomaly detection. Mobile App Development : Develop a mobile-friendly version of the app for restaurateurs to access insights on the go. Integration with POS Systems : Directly integrate Gastronalytics with TouchBistro’s POS system for real-time data updates and seamless operation. Customer Behavior Analysis : Expand insights to include customer segmentation, loyalty program effectiveness, and personalized marketing strategies. Global Expansion : Extend the app’s capabilities to support international markets by incorporating region-specific data and trends.

Sustainability Features : Add features to help restaurants reduce food waste and optimize inventory management for sustainable operations.

Community Feedback : Gather feedback from restaurateurs to refine existing features and identify new opportunities for improvement.

Gastronalytics has the potential to revolutionize the restaurant industry by empowering businesses with data-driven insights. With continued development and refinement, we aim to make it an indispensable tool for restaurateurs worldwide.

Built With

• javascript
• json
• matplotlib
• pandas
• plotly
• python
• streamlit