Fire Prediction Project Summary

Using Worldclim climate data and historical fire data from Data.gov, we used 4 different algorithms to predict the likliehood of future fires. Our models were trained using historical climatic data, which were then applied to Worldclim predictive datasets (2040, 2070) with 2 levels of climate change severity (driven by future air pollution levels).

Heroku: https://ucbx-fire-prediction-2019.herokuapp.com/

Extraction, Transform

• In R we extracted climate change data from:

https://catalog.data.gov/dataset/combined-wildfire-dataset-for-the-united-states-and-certain-territories-1870-2015

http://www.worldclim.org/version1

• Exported as CSV and read into python and loaded into Pandas DataFrames for the creation of our models

Machine Learning

Four Prediction Models:

1. Neural Network
2. Random Forest
3. KNN
4. Logistic Regression

Load

• Loaded our dataframe into MongoDB with PyMongo
• Hosted MongoDB in external server

Flask App

• Created our flask app using Flask-PyMongo

Web Template

• Used Flask’s {% %} notation to extend a layout.html file, to keep consistent navbar

Charting

• Leaflet map with Patrick Wied's heatmap plugin showing relative fire likliehood with options to adjust the year, degree of climate change, heatmap sensitivity, and algorithm used
• Collection of charts created using Tableau