We are a diverse group with computer science, statistics and mechanical engineering majors. The ConocoPhillips challenge seemed to be a unique conglomeration of all of our team members interests and efforts.
What it does
Our model predicts failures related to above and below ground downhole pressure and equipment.
How we built it
We used python to create our model testing and creating it using logistic regression and random forests.
Challenges we ran into
The n/a values in the data set were difficult to navigate also the processing time using random selection for selecting hyperparameters was a challenge.
Accomplishments that we're proud of
Creating a model with 99.37% predicability as beginners using machine learning was a feat for our team, as we on a constant learning curve.
What we learned
We got to learn about machine learning, at least a basic introduction.
What's next for ConocoPhillips Datathon2019
If we were given variables it would be easier to understand the relationship and correlation between the failures and strive to decrease profit loss.