Problem Statement

Determining the degree of damage that is done to buildings post an earthquake can help identify safe and unsafe buildings, thus avoiding death and injuries resulting from aftershocks. Leveraging the power of machine learning is one viable option that can potentially prevent massive loss of lives while simultaneously making rescue efforts easy and efficient.

In this challenge we provide you with the before and after details of nearly one million buildings after an earthquake. The damage to a building is categorized in five grades. Each grade depicts the extent of damage done to a building post an earthquake.

Given building details, your task is to build a model that can predict the extent of damage that has been done to a building after an earthquake.

Data Description:

You’re give four files: train.csv, test.csv, Building_Ownership_Use.csv and Building_Structure.csv.

Details of the train.csv file:

Variable	Description
area_assesed	Indicates the nature of the damage assessment in terms of the areas of the building that were assessed
building_id	A unique ID that identifies every individual building
damage_grade	Damage grade assigned to the building after assessment (Target Variable)
district_id	District where the building is located
has_geotechnical_risk	Indicates if building has geotechnical risks
has_geotechnical_risk_fault_crack	Indicates if building has geotechnical risks related to fault cracking
has_geotechnical_risk_flood	Indicates if building has geotechnical risks related to flood
has_geotechnical_risk_land_settlement	Indicates if building has geotechnical risks related to land settlement
has_geotechnical_risk_landslide	Indicates if building has geotechnical risks related to landslide
has_geotechnical_risk_liquefaction	Indicates if building has geotechnical risks related to liquefaction
has_geotechnical_risk_other	Indicates if building has any other geotechnical risks
has_geotechnical_risk_rock_fall	Indicates if building has geotechnical risks related to rock fall
has_repair_started	Indicates if the repair work had started
vdcmun_id	Municipality where the building is located

Details of the remaining files are described in the ReadMe file.

Download dataset

Submission

A participant has to submit a zip file containing your ‘building_id’ and predicted ‘damage_grade’ in a csv format. Check the sample submission file for format.

building_id, damage_grade
a3380c4f75, Grade 3
a338a4e653, Grade 1
a338a4e6b7, Grade 1
a33a6eaa3a, Grade 5
a33b073ff6, Grade 4

For challenge related queries, discussions and announcements join our Slack channel.

Evaluation Metric

The submissions will be evaluated based on F1 Score with ‘weighted’ average.