GreenWise

Inspiration

While online shopping is convenient, buying things at the click of a button does not allow much time for customers to consider their environmental impact. We wanted to create an easy, convenient way for people to calculate the carbon emissions generated from shopping.

What it does

GreenWise allows online shoppers to calculate the carbon emissions of their online shopping in real time. When the user visits a product page, the extension scrapes relevant data about the product and sends it to the backend model which predicts the carbon emissions from that product.

How we built it

We first trained and fine-tuned a Random Forest Regressor on the Carbon Catalogue public dataset of various consumer products and carbon emissions
We then made a simple Flask backend that takes in the name of the product, the country of manufacturing, and the year of production to make a prediction about the kilograms of carbon emissions
We created a webscraper using Beautiful Soup to extract relevant data from the Amazon page
We incorporated both the webscraper and Flask app into a chrome extension that makes it easy for the user to calculate carbon offsets while on the site
We created a beautiful, user-friendly interface for the popup using Figma

Challenges we ran into

Creating the basic Flask was a challenge because we had to learn how to take new data through the frontend and pass it to the model to make predictions
Integration between the Chrome Extension and the backend posed a challenge becuase the backend was running on Python Flask while the extension was coded in JavaScript
Extracting reliable data from Amazon using web scraping because Amazon's page structure made if difficult to consistently extract the right data

Accomplishments that we're proud of:

Built a fully functional chrome extension that integrates machine learning and web scraping to make accurate predictions
Overcame the challenges of integrating a web scraper into our chrome extension that consistently fetches the correct data
Designed an engaging, easy-to use interface and brand logo

What we learned

-The challenges of integrating different parts of an app such as the webscraper, model, and user interface

How to connect a chrome extension with a Flask backend
The impact of different aspects of a product such as country of origin, manufacterer, and type of product on the carbon emissions generated

What's next for GreenWise

Improving the carbon estimates by training the model on new data and scraping more useful data from the Amazon site like packaging, materials, and type of shipping
Allowing users to compare different products in terms of carbon impact and also expanding to other online shopping sites like Walmart.com and Target.com
Making goal-setting features that allow users to set long-term carbon goals for their online shopping