Exoplanet Detection

Inspiration

This was a space themed hackathon, so we decided that it was in our best interest to create a project that aligns with the them. Our team prior to the hackathon were discussing humanity and the space exploration with the recent success of Space X CEO Elon Musk's recent feat of contacting the space station. We were very intrigued with the ideas surrounding the ways humanity would spread to other planets and the use of Machine Learning to aid us.

What it does

Our program takes in star flux values. Then we use multiple data cleaning methods in order to ensure that our model makes accurate predictions about whether there are exoplanets.

How we built it

We first read the data and cleaned the data up using SMOTE and mean normalization. We also used a Gaussian Smoother on the dataset. We then ran the train dataset through decision tree algorithm then tested it using a test dataset. We then used flask and html to design a website for the program to run on.

Challenges we ran into

We ran into a few challenges along the way. First, we needed to find a data set that would be implemented into the program we plan to create. Finding a data set was extremely difficult, however, we found a huge data set with nearly 4000 columns and more than 5000 rows. Then, we figure out an algorithm that best fit out data set. Machine learning has a vast array of algorithms that can be implemented into the product we create. We first used a logistic regression, however, we were not able to accurately predict the results that the data had. We also tried a neural network but we couldn’t get it to work accurately. We also had trouble getting all the HTML files organized in replit using flask

Accomplishments that we're proud of

We are proud of figuring out how to use many new python packages that we never used before.

What we learned

We learned multiple new methods of making data more usable for machine learning We learned more about decision tree classifier and how to approach large data sets.

What's next for Exoplanet Detection

The next steps are to come up with a better machine learning algorithm and to also get the flask website to properly display results and matplotlib plots