Austin Animal Shelter Outcomes
Deanna Aburaad and Amanuel Alemu
[email protected]
[email protected]
EECS 349 Machine Learning
Northwestern University
[email protected]
[email protected]
EECS 349 Machine Learning
Northwestern University
Our task is to predict whether or not a shelter animal will be adopted based on its characteristics. Every year, 7.3 million dogs and cats enter shelters. 2.6 million of these are euthanized, and 2.7 million are adopted. Being able to predict an animal’s outcome upon its arrival at the shelter allows the shelter to better allocate its resources and make adjustments to increase each animal’s chances of adoption.
Using an animal’s attributes (named/unnamed, animal type, spayed/neutered, sex, age, breed, and color), we can predict its outcome based on the classification tree we created. We trained it with data from Austin Animal Shelter in Texas, the largest no-kill shelter in the US. The only animals that are euthanized there are those that are suffering or are aggressive, so there is bound to be bias towards adoption, as most other shelters euthanize animals because of overcrowding. Thus, our model is best suited for no-kill shelters.