Project Sunburn

Solar Panel Feasibility Map - generated by Sunburn.py

Inspiration

Create a recommendation algorithm to predict the best locations for solar panel implementation using a number of weighted factors related to feasibility and addressing potential barriers to adoption of renewable energy sources.

What it does

Uses clustering to pinpoint locations within the US with high annual sunlight hours (technological feasibility), large carbon footprint replacement values(quantified environmental impact), and high income (financial feasibility).

How we built it

Combined two datasets: average incomes and solar info per zip code (US_Income_Kaggle and Google Project Sun). Used k-means clustering to identify locations with maximum income, carbon footprint, and sunlight hours.

Challenges we ran into

Choosing a machine-learning algorithm to best approach the problem. Initial implementation of the algorithm due to relative inexperience with data science (learner track).

Accomplishments that I'm proud of

Rendered 10 well-reasoned location recommendations substantiated with data and validated by zip code data.

What I learned

Basics of data frames, data cleaning, clustering, machine learning. Extensive experience with debugging and verification.

What's next for Project Sunburn

More accurate income info Include populations to further normalize input data

Built With

• google-maps
• pandas
• python
• scikit-learn