Our friend and colleague, Andras Zsom, recently posted an excellent write up documenting some of his work in a Kaggle competition. The aim of the competition was to build a classifier that most accurately predicted the outcome of animals arriving at an animal shelter (e.g,. adopted, transferred, euthanized). The data came from an animal shelter in Austin, TX.
Having an accurate classifier of animal outcomes has interesting and important implications. Most notably, it could help shelters allocate their resources so as to minimize the need to euthanize animals. I like to see work like this. That is, work that uses data science and machine learning in support of some social good.
In his post Andras details the entire data-analytic pipeline: from data cleaning and feature engineering all the way to the various machine learning algorithms that were tried. Andras also includes some excellent visualizations. You can find his code here.
You should go read it, and then go adopt a puppy.