datasponge

Inspiration

Machine learning takes time to learn and can be hard to implement. We wanted to make it so that anybody can perform this up and coming technology.

What it does

Datasponge acts as a medium for machine learning with minimal human assistance, taking in user defined data sets and providing an output model.

How we built it

Using Tensorflow as the backend machine learning library we created a flexible program to take user input and change the structure to fit the data.

Challenges we ran into

A big challenge was getting our code to work with multiple different types of data sets, for example different data sets had different numbers of attributes and outputs, or differing data types. It was also hard creating a visually pleasing website that properly defines what our project does, as this was the very first website we attempted to make from scratch

Accomplishments that we're proud of

Creating a visually pleasing website from scratch, which included becoming familiarized with different forms of visual creation. We also accessed 1 million data points in under 5 seconds.

What we learned

How to use HTML, CSS, JavaScript, and Jquery.
Creating a CSS file without using the Bootstrap framework.
How to manipulate tensorflow to handle different data situations

What's next for datasponge

Due to time constraints, we only made datasponge compatible with classification machine learning. In the future we could add support for all other models.