Forest Fire Predictor

This webapp can be used to predict the amount of area burnt by a fire by selecting a point on a map.

Built Using

• Flask API - hosted on Heroku
• Keras with a Tensorflow backend for predicting - computed on the cloud and sklearn for pre-processing
• JS and HTML for the websites
• Openweathermap API for the weather data
• Pandas, Numpy, Dill for csv reading, linear algebra and saving objects, respectively.
• Integrated and hosted on GitHub Pages

How it works

• When the user clicks the button on the web page, the Heroku API is sent the latitude and longitude
  of the place selected on the map.
• The python script then fetches weather data from the openweathermap API.
• This data is used to calculate the various Fire Weather Indices.
• The FWIs and the raw data from the API are fed into the Deep Regression model which
  outputs

Requirements

• keras, tensorflow, sklearn for machine learning
• flask, flask-cors for the API
• numpy for linear algebra
• dill for object saving
• pandas for reading the training data in the form of *.csv