Placed Top 10 at HackMIT 2021

easyML: ML For All

We want ML to become more accessible for everyone and to develop technologies that help people feel comfortable using ML. Our project makes it easy for users to understand and work with machine learning and data. A user can upload a csv dataset to our website and our service will automatically train a gradient boosted decision tree model on their data. The user can then query this model to generate new predictions. By abstracting away the complexities of downloading software libraries and managing compute hardware, we enable users to focus entirely on their data and the application of their models.

Demo: easy-ml.xyz

1. Acquire a csv file of the data you want to train on (e.g. heart_disease.csv)
2. Upload csv file to home page of easy-ml.xyz
3. Select the target column that you want to predict for in the csv file (e.g. diagnostic in heart_disease.csv)
4. A gradient boosted decision tree will be trained on the data in our backend EC2 instance
5. Finally, a page pops up where you can input new data to generate a prediction (e.g. is this person likely to be diagnosed with heart disease?)

Motivation

Over the past several years, our team has seen certain communities across the world blast off with robust access to information and tech while others wallow behind. The has resulted in distrust and conflict: while certain communities tend to make claims backed academic knowledge, other communities lack the technical knowledge to verify those claims, creating a general information divide in the world. Take COVID-19 for example—well-educated communities were well-equipped to make and understand analytically-driven public health guidelines, while those without access to education and technology were not as well equipped with the tools or knowledge necessary to verify the reasoning behind guidelines. Such a disparity has resulted in a lack of trust between communities.

Our project easyML attempts to solve this communal disparity. Machine learning has allowed many communities to make many fascinating advancements, but at the same time, its complexity has left many communities in the dust. We seek to make machine learning extremely easy for anyone to use.

With our website, all you need to do is upload a data set, and easyML will guide you through the rest. No parameter tuning, no model selection, no complexity at all. Through Tangram, we’ve engineered a tool that we hope has lowered the barrier for those seeking to understand the information of the world.