Skip to content

Python code using supervised learning to prioritize shelter animals by their probability of getting adopted to maximize the rate of adoption.

Notifications You must be signed in to change notification settings

jieliang/prioritizing_shelter_animals_for_adoption

Repository files navigation

Project Description Slides

  • prioritizing_shelter_animals_for_adoption_slides.pdf

Data cleaning and transformation

  • clean_animal_services_data.ipynb : cleans animal services data
  • clean_animal_services_data_save_to_database.ipynb : saves cleaned animal services data as a table named animal_services in aws postgres database
  • clean_economic_index_data.ipynb : cleans economic data and saves it as a table named economic_index in aws postgres database
  • clean_expenditures_data.ipynb : cleans animal services department expenditure data and saves it as a table named expenses in aws postgres database

Modelling

  • model_selection.ipynb : joins tables in database as clean data for modelling; tries different models and selects winner based on metrics
  • model_evaluation.ipynb : plots and compares roc curves of finalists and winner from model selection

Data directory(not included in submission because size exceeds git hub limit)

  • downloaded csv files from data sources
  • saved pickle files

Flask directory

  • files for creating web based app that takes in user input and returns prediction using selected model

Helper file

  • generate_template_for_flask_api.ipynb : generates a template filled with some default values for api.py in flask directory to use