Austin Green Energy Predictor

Overview

This project is a culmination of a 5-week capstone project for the UT Austin Data Analytics & Visualization Bootcamp. The Austin Energy Predictor predicts renewable energy output in Megawatt-Hours (MWh) generated from a wind and solar energy farm in Texas. The model is trained on time and weather factors, such as temperature, wind speed and cloud coverage. The purpose was to forecast power generation to get a better understanding of renewable energy as a mainstream power source for a healthier planet.

Results

Using data from Austin Energy and historical weather data, we were able to train the model to predict energy output to more than 85% accuracy. The solar model is the most accurate and does a particularly good job at predicting when during the day the farm starts and stops producing.

Responsibilities

Project Manager - Coordinated with team members to establish goals and timelines, communicated with Austin Energy to increase project effectiveness, ensured the project was completed on time.
Database Administrator - Established and maintained NoSQL database and validated database access.
Machine Learning Engineer - Created and trained the Tensorflow deep learning neural network along with a multiple variable linear regression model.

Technologies Used