Random Acts of Pizza
Projects | | Links: Report | Github
This classifier predicts if a user will get a free pizza on reddit.com/r/Random_Acts_Of_Pizza/.
Random Acts of Pizza is a project in which we trained multiple classification models to predict if a user gets a free pizza from the Reddit random acts of pizza community. This work was the final project of the course Applied Machine Learning. Authors: Carolina Arriaga, Kanika Mahajan.
Overview
We used data from the Kaggle competition Random Acts of Pizza to train multiple classifiers to predict a binary class. We used feature engineering by collecting user metadata and adding text-based features. We also applied dimensionality reduction using SVD. We explored four models: Logistic regression, Random Forests (Ada and XGB), Multilayer Neural Network, and Dense Neural Network. Finally, we proceeded with the hyperparameter tuning. We were able to predict whether a user would get a pizza or not.
