Skip to content
No description, website, or topics provided.
Branch: master
Clone or download
Latest commit a304cbf Apr 6, 2019
Type Name Latest commit message Commit time
Failed to load latest commit information.
Procfile prep for heroku Apr 6, 2019 Update Apr 6, 2019 prep for heroku Apr 6, 2019
requirements.txt prep for heroku Apr 6, 2019
runtime.txt prep for heroku Apr 6, 2019

Fire Prediction Project Summary

Using Worldclim climate data and historical fire data from, 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).


Extraction, Transform

  • In R we extracted climate change data from:

  • 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


  • 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


  • 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
You can’t perform that action at this time.