NLP with Disaster Tweets
Predict which Tweets are about real disasters and which ones are not.
Alexander Bricken
Currect Submission Accuracy and Position on Leaderboard: 84.063%, position #71 (although #52 if you subtract cheaters).
Project structure:
├── README.md <- The top-level README for developers using this project.
├── data
│ ├── raw <- The raw data
│ ├── submissions <- The final data to be submitted
│
│
├── requirements.txt <- Requirements for this project.
│
├── utils.py <- Utility functions for project.
├── tweet-scraping.py <- Tweet scraping for more data.
│
├── notebooks <- Jupyter notebooks for this project.
│ ├── nlp_disaster_tweets <- The main Jupyter notebook
│
├── data-dictionary.txt <- Data dictionaries, manuals, and all other explanatory materials.
Data
Raw data source: https://www.kaggle.com/c/nlp-getting-started/overview
Using The Project
Check in the notebooks folder to see the associated exploratory analysis.
If you want to play with it, simply type git clone https://github.com/Briiick/NLP-disaster-tweets.git in your terminal.
References
Natural Language Processing with Disaster Tweets
NLP with Disaster Tweets: EDA, cleaning and BERT
Basics of using pre-trained GloVe Vectors