Ecal | Devpost

Inspiration

From the beginning, we wanted to tackle problems relevant to how companies can become more sustainable. We then realized that it is hard for small companies to make predictions about how expensive going sustainable is since such estimation mostly involves a dedicated analyst that might be expensive to hire. Therefore, we decided to help some companies to estimate how much it takes financially to become more sustainable. We also chose to target the e-commerce industry since it involves a lot of waste such as packaging and transportation.

What it does

We created an algorithm that automatically calculates the sustainability cost according to the size of a company. Built an interactive website to make our algorithm easy to use and display our working process and ideology.

How we built it

We first had a brainstorm to determine how to measure a company's environmental sustainability, so that we can quantify a company's investment in sustainability. Then, we determine how much money big eCommerce companies like Amazon and eBay spent on sustainability during the past few years. Next, we used companies' info such as total yearly income and total yearly expedenture as inputs and companies' investment in sustainability as the outcome to train a regression model that finds the relationships between these inputs. After testing and adjusting to make sure our model is reliable, we created a website that shows our project and supports our algorithm. Users can input their companies' data and get the result instantly.

Challenges we ran into

One of the most challenging parts of this project is to gather enough valid data to train our model. Companies' financial data is difficult to get, and getting a valid report on measurements like CO2 emission and electricity usage is even harder. Because we only had a limited amount of data, machine-learning algorithms like neural networks that we originally planned to use becomes impractical. We had to return to multi-input polynomial regression to get a reliable prediction.

Accomplishments that we're proud of

We are very proud of the work we did to gather and processed data, given the difficulty to find valid datasets. We are also proud to create a beautiful website that is very easy to interact with.

What we learned

We have learned that we should validate our data source first before setting out to work on a project. We should also be more versatile when we face critical problems like data shortages. We should think more creatively and embrace necessary changes when one method is already proven to be impractical.

What's next for Ecal

For the next steps, more data can be added to our data set used in the algorithm right now. After cooperating with real companies, we will have the access to more detailed and reliable data, and this will secure a more accurate prediction result for our customers. We can also add more dimension to our dataset if more time is given, which would also result in a more accurate prediction. Meanwhile, the website we are presenting right now is in a very early form and needed more refinement in the future to attract more customers to use our services.