NLP-disaster-tweet-prediction

This project has been realized by me (Rebecca Di Francesco) and my collegue Stefanija Galevska for the course "Machine Learning" of the Master's degree in Data science at University of Padova. With this project we participated to the challenge on kaggle: https://www.kaggle.com/c/nlp-getting-started. The aim of this challenge was to predict which tweets were describing a disaster so it was a binary classification problem. The main focus was on preprocessing the textual data with NLP techniques; after that we implemented and compared different ML techniques: Logistic Regression, Support Vector Machine, Random forest and Neural networks. For each of this techniques we chose the hyperparameters based on a validation set.