Predicting power outage using weather forecast data
Our model predicts the possibility of a power outage based on the current weather conditions in a region. Using the historical weather records from Climate Data Online and the Electrical Disturbances OE-417 Data from the Department of Energy, it is able to predict accurately most of the power outages in a region.
For a proof of concept, we have focused on Texas, Houston and Michigan, Ann Arbor. We also tested our model's potential for transfer learning by testing it on a neighboring state of Michigan, Indiana.
You can find our final report and presentation in the root directory. THe Weather Data Analysis contains all our code for the model.
This project was a part of CS542: Boston University, taught by Prof. Sang Chin in Spring 2018.