fcole90 / kesko-food-waste-hackathon

K-market app for reducing food waste - Hackathon Junction 2019

fresKo

Hackathon Junction 2019

K-group app for reducing food waste by optimising consumer consumption and market store optimisation.

Find a pdf presentation here.

Team

• Fabio Colella
• Mamdouh El Nakeeb
• Abhishek Singh
• Wissa Azmy
• Agisilaos Kounelis

Kesko, kustomers, kestävä

The apps aims at food sustainability by reducing food waste. This is done in a double way:

• directing customers to buy expire-soon food;
• gathering customers data habits for K-markets to better optimise future store updates.

How does it work?

The app is a webapp that suggests users the best K-markets to go to buy the products they desire. The user will provide a position and a grocery list. The app will hence find the closest K-market that has the greatest amount of expire-soon food among the items in the user's grocery list, and taking into account the selling probability of an item given the popularity of the item and the shop.

Frontend

Simple mobile-first webapp, with few screens. The workflow is the following:

1. User selects grocery list (input 1)
2. User receives a list of ranked K-market according to optimisation (output 1)
3. User selects a market (input 2)
4. User receives a link to map (output 2) (map can be any in Google Maps, OpenStreetMap, Reittiopass)

K-market advantage

We can gather data on user consumptions habits from this simple app, which K-group can then use for improving the store selection.

Why the user wants to use it?

Target users care for environment and want to reduce food waste. In addition users might want to use the app to try to save money on close-to-expiry products.

Backend

Implemented in Django using the Restful APIs plugin. It also interacts with a simpple optimisation algorithm for returning the markets ranking list.

Expiry date workaround

We have no expiry date API available, so we estimate the amount of close-expiry items. We know that K-group has such API internally, so it can be implemented later.

Future ideas (aka Wishlist)

• Return a group of K-markets to cover all food items (like if K1 doesn't have fish, also optimise for a K2)
• Gamify the app to make users compete on sustinability
• Group items brandless (like just milk, then autoselect a brand which many items on close expiry)
• Suggest accurate quantities based on recipe portions amount

Initialisation

Before running the server run the script /optimisation/data_download_api.py which is needed for data retrivial and augmentation, and fast analysis.

Frontend / Backend

Frontend and backend are separate. The frontend can be deployed on something that supports React.js, like Netlify, while the backend is a Django app which is used to retrieve APIs, which can be hosted for example on python anywhere.

Note python anywhere works with proxies and may not work with K-group APIs.