Will Coughlin

Predicting Power Grid Abnormalities

Project Repository | Report (PDF)

This project for Georgia Tech’s Data Mining course involved the prediction of abnormal grid conditon occurrences using public archive data from ISO-NE. I trained and evaluated several binary classification models, producing an XGBoost model and a Support Vector Machine model that perform reasonably well in predicting these occurrences.

This project encompasses a front-to-back data mining project, including

  1. Web Scraping
  2. Data Cleaning
  3. Exploratory Data Analysis
  4. Modeling
  5. Evaluation