Inventorize

Inspiration

In the restaurant industry, one of the biggest challenges is predicting the right amount of food to prepare each day. Overestimation leads to waste, while underestimation can result in lost sales and disappointed customers. This issue inspired us to develop Inventorize, a solution that not only forecasts demand accurately but also enhances inventory management and reduces food waste, contributing to a more sustainable environment.

What it does

Inventorize uses advanced machine learning models to provide accurate daily demand forecasts for restaurants and food suppliers. The application helps users manage inventory more effectively by predicting the quantity of food items that need to be prepared, thus optimizing the supply chain, reducing food waste, and ensuring that businesses can meet customer demand efficiently.

How we built it

We built Inventorize using a multi-tiered technology stack:

Frontend: Developed using Next.js and TypeScript to provide an intuitive and responsive user experience. Backend: Implemented in Flask to manage requests and serve predictions from our machine learning models. Machine Learning: Utilized Random Forest for regression tasks to predict demand and explored the Prophet model for its robustness in handling seasonal variations in time series data. Database: Chose Supabase for its real-time data management and ease of integration with web technologies.

Challenges we ran into

Finding a reliable and right dataset to work with. Integrating complex machine learning models with a real-time web application. Ensuring the accuracy of predictions with varying data quality required sophisticated data preprocessing and feature engineering.

Accomplishments that we're proud of

Creating a user-friendly web interface and a working machine learning model. Significantly reducing potential food waste for users by providing more accurate demand predictions. Creating a scalable solution that can adapt to different scales of operation, from small cafes to large restaurants.

What we learned

Throughout this project, we deepened our understanding of machine learning applications in real-world scenarios. We explored how different models handle various types of data and learned to integrate these models into a functional web application. This project was particularly enlightening in demonstrating the impact of data-driven decisions in operational efficiency.

What's next for Inventorize

Looking ahead, we plan to: Integrate machine learning model into our frontend interface Enhance model accuracy by incorporating more granular data and exploring additional predictive factors. Expand the range of features in the application, such as automatic ordering systems based on the predictions. Explore partnerships with food supply chain stakeholders to further integrate and streamline operations.

Built With

• next.js
• react
• scikit-learn
• supabase