I built and trained a dense neural network(in train_v3.ipynb) on 15K animal profiles, each with 23 features, to predict adoption speed with an accuracy of 86%. We hope this model can help shelters improve pet profiles for faster adoptions and provide a numeric evaluation while developing online profiles for pets.
This is part of the final project for the class AI for Social Good in CMU.
Pytorch
A paper I wrote with my team as part of this project
A poster I made with my team as part of this project - we won the most popular award during poster session
A presentation I had done with my team as part of this project