"Drilling" To The Future

• 

  Comparison between our Neural Network's predictions and the actual peak output rates

• 

• 

• 

Inspiration

This project is a submission for the Rice Datathon2024's Chevron Track, where we are asked to predict the peak output rate of oil wells using data about their location, proppant use, fluid use, and length.

What it does

Our predictions are based on a Linear Model (preliminarily) and a Neural Network that was trained on the provided data.

How we built it

We principally used Scikit-Learn and Tensorflow-Keras to create our models, along with Numpy, Pandas, and Matplotlib (Python).

Challenges we ran into

Our linear model had difficulty, so we moved on to a neural network.

Accomplishments that we're proud of

Our best RMSE was about 100

What we learned

We've learned the benefits of using a neural network for modeling complex relationships in a dataset.

Built With

• keras
• pandas
• python
• scikit-learn
• tensorflow