Welcome to our beginner-friendly Machine Learning Hackathon! This challenge is specially designed for first-time participants, offering a safe and supportive environment where you can learn, collaborate, ask questions and grow with fellow newcomers.
Each year, thousands of fires blaze across the African continent. Some are natural occurrences, part of a ‘fire cycle’ that can actually benefit some dryland ecosystems. Many are started intentionally, used to clear land or to prepare fields for planting. And some are wildfires, which can rage over large areas and cause huge amounts of damage. Whatever the cause, fires pour vast amounts of CO2 into the atmosphere, along with smoke that degrades air quality for those living downwind.
Figuring out the dynamics that influence where and when these fires occur can help us to understand their effects better. And predicting how these dynamics will play out in the future, under different climatic conditions, could prove extremely useful.
The objective of this challenge is to create a machine-learning model capable of predicting the burned area in different locations from 2014 to 2016.
We’ve aggregated data on burned areas across Zimbabwe for each month since 2001. You’ll be given the burn area data up to the end of 2013, along with some additional information (such as rainfall, temperature, land cover etc) that extends into the test period.
We’ve split up Zimbabwe into 533 equal areas, centred around the locations provided in the lat/lon columns. The target variable, burn_area, is the percentage of the area that has been burned in a given month. Due to the way it’s measured, there may be some overlap of burned areas for two successive months, and so the total burned area over a time period isn’t necessarily equal to the sum of the ‘burn_area’ figures for all months. You are not permitted to use external data in this competition.
The evaluation metric for this competition is Root Mean Squared Error
You need to predict the proportion of the burned area per area square, with values of 0 to 1.
The IDs take the form of [area ID]\_yyyy-mm-dd. There are 3821 area squares each with a unique ID ranging from 0 to 3820.
Your submission file should look like this (numbers to show format only)
ID burn_area
0_2014-01-01 0.5624
1_2014-01-01 0.7654
2_2014-01-01 0.1134
3_2014-01-01 0.9751
Climate: All columns climate_[variable] derived from https://developers.google.com/earth-engine/datasets/catalog/IDAHO_EPSCOR_TERRACLIMATE - see the image bands listed for more information. Land Cover: The percentage of the area falling into each of the LC_Type4 land cover types https://developers.google.com/earth-engine/datasets/catalog/MODIS_006_MCD12Q1 - see for details. UN-Adjusted Population Density (estimated number of persons per square kilometre) from https://developers.google.com/earth-engine/datasets/catalog/CIESIN_GPWv411_GPW_UNWPP-Adjusted_Population_Density Mean elevation in meters, from https://developers.google.com/earth-engine/datasets/catalog/USGS_GMTED2010 Monthly Precipitation Estimates - https://developers.google.com/earth-engine/datasets/catalog/TRMM_3B43V7
By Zindi Africa