Inspiration We were inspired by recent forest fires in the west coast to experiment with the feasibility of predicting forest fires so authorities may take the necessary preventative measures.
What it does Plots 120,000 historical data points in the last five years of forest fires in the states of Washington, California and Oregon to reveal trends. Used clustering model to detect spots that have high risk to forest fire
How we built it We utilized open historical data of past fires in Washington, Oregon, and California and aimed to show trends with Plotly in Python. The data is from the following sources: USDA Forest Service, Data.gov, Data.gov
Challenges KMean is not optimal for geospatial data clustering; Used DBSCAN. Size of the data set is large. tuning machine learning parameters for each of the 12 months' data is time-consuming. JS syntax was challenging to work with
What we learned Gaby learned how to plot simple graphs from CSV data with Plotly