What Is Our Carbon Footprint?
Roots
Summary
EUR 2000 million in Food waste is generated for AgroChain in the Netherlands (Source: Ministry of Agriculture, Nature and Food Quality)Restaurants account for 60-100 million of this waste.A typical restaurant wastes around 18% of the fresh food it purchases. This leads to an extremely high Carbon Footprint Increase.Most of this food waste is generated due to sub-optimal ordering practices. where the supply chain orders are placed using gut feeling and experience instead of actionable data.Roots is a platform that aims to reduce these inefficiencies using Machine Learning algorithms leading to more efficient ordering practices and thus, a reduction is Food Waste and Carbon Footprint.
How We Addressed This Challenge
We developed a platform which provides the restaurants the correct quantities of ingredients to order based on the past historical sales of restaurants combined with a number of factors like weather, season and general public sentiment. This will allow to reduce food waste compared to the current inefficient practice where food ingredients are ordered based on gut feeling and intuition.
How We Developed This Project
To do this we first break down, each of restaurant dishes, into ingredients
Then we use past sales data to predict future sales.
We take into account multiple parameters for prediction, these include, total sales per day, the time of the year and as well as the holidays.
The front end is built on React.js and backend is based on Node.js. Pytorch is used for building the ML model and Dataset from NASA and Upenn is used for validation.
Our test on restaurant dataset shows we are 86% accurate, which is 16% more than industry practice of ordering based on gut feeling and experience. This accuracy is expected to improve as our algorithm is always training.
This generates multiple benefits for various stakeholders involved.
The values for the stakeholders include; lower operating costs for the restaurant, a reduction in their food wastage, as well as reduced carbon footprint.
As our platform is used by multiple restaurants, we are able to leverage the power of group buying to order the supplies directly from the farmers. Thus, Bypassing the traditional entities in the supply chain
And this is how it looks: https://youtu.be/KKBRJDIhV4o
We plan to deploy this platform in the city of delft at small restaurants like Pasta2go and Doner kingdom.
A typical small restaurant like Pasta2Go spends around 3000 Euros every month in ordering fresh supplies from the distributors.
Using the rough figure of 18.5% of wasted fresh food, this leads to a savings of around 550 euros per month for the restaurant owners. Our platform takes a small cut from these monthly savings, roundabout 5%.
Our main revenue is generated from the restaurants who are using our platform.
We generate a monthly revenue based on a cut based pricing model.
Later on, With group buying in volume, small farmers can also act as key stakeholders. Thus, generating supplies directly to the restaurants.
We are changing the traditional behaviour of ordering supplies by restaurants. Our algorithm has increased accuracy by utilizing actionable data.
In higher stages of the supply chain, supermarkets already use an ERP system to order supplies from distributors.
Further, for group buying, our end users are volume buyers instead of individual households.
In the future, we also plan to show our demand insights for each crop to the farmers so they can plan their production accordingly.
How We Used Space Agency Data in This Project
We used the CSA in our idea in our plan for extending this, by adding extra features, for instance, visualization tools for both users and restaurants to better visualized the reduced carbon footprint and especially its effects.
Project Demo
Data & Resources
https://wca.wharton.upenn.edu/research/independent-purchasing-cooperative/