Shelter Animals outcomes predicted by machine learning

Predicted Outcomes

Adoption Died Euthanasia Return to owner Transfer to outside org
65.4% chance 0.0% chance 0.3% chance 25.1% chance 9.3% chance



Select factors to include in outcome prediction

Intake factors

Animal type

Name at intake


Time and Date of Intake

(between 10/1/2013 and 3/1/2016)


Outcome factors

Project Description

This machine learning model predicts outcomes for animals delivered to a municipal shelter based upon information collected at intake and when they leave.

The concept and data for this project comes from a Kaggle competition. The original data source is the Austin Animal Center.

The highest scoring (i.e., lowest log-loss) model uses a cross-validated gradient tree boosting classifier.

Features used include: age upon outcome, gender upon outcome, animal type (dog or cat), named or unnamed, date and time of intake, purebred or mix.

Project Details