rumble_banner_2.png
MIDS Capstone Project Spring 2019

Rumble: Mountain bike predictive analytics from your smartphone

Rumble leverages machine learning to deliver insights to citizen mechanics on the health of their bikes components enabling predictive maintenance.  Rumble applies advanced techniques, such as fast-fourier-transform signal-processing, to vibration data captured on a users own cell phone, before passing it to SVM based machine learning models.  The results are then made available in a consumable and easy to use format, understandable to our users.

The application collects messy, native, real-world data and uses data science techniques to understand it, process it, build models from it, and deploy those models in a production pipeline that let's anyone derive real, actionable insights from their own vehicles. And even though this model is focused on mountain bikes, the framework and techniques can be extended to many applications across a multitude of industries.

Last updated: October 1, 2019