Skip to content

Twitter has become an important communication channel in times of emergency. The ubiquitousness of smartphones enables people to announce an emergency they’re observing in real-time. Because of this, more agencies are interested in programatically monitoring Twitter (i.e. disaster relief organizations and news agencies). But, it’s not always cle…

Notifications You must be signed in to change notification settings

anikch/nlp-predict-tweets-about-real-disasters

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

8 Commits
 
 
 
 
 
 

Repository files navigation

nlp-predict-tweets-about-real-disasters

Problem Statement:

Twitter has become an important communication channel in times of emergency. The ubiquitousness of smartphones enables people to announce an emergency they’re observing in real-time. Because of this, more agencies are interested in programatically monitoring Twitter (i.e. disaster relief organizations and news agencies). But, it’s not always clear whether a person’s words are actually announcing a disaster. In this competition, you’re challenged to build a machine learning model that predicts which Tweets are about real disasters and which one’s aren’t. You’ll have access to a dataset of 10,000 tweets that were hand classified.

Competition details and dataset can be found in kaggle: https://www.kaggle.com/c/nlp-getting-started

Approach Taken:

  1. Performed text pre-processing and replaced constactions (e.g. wouldn't to would not) in dataset.
  2. Used BERT pre-trainted model bert-base-uncased with maxlength 512.
  3. Identified optimal learning rate (3e-5) and fine-tuned using one cycle policy and the optimal learning rate.
  4. Evaluated on Test dataset. Confusion matrix is as below:

image

  1. Performed prediction on final dataset and submitted. Got Public Score: 0.81428

About

Twitter has become an important communication channel in times of emergency. The ubiquitousness of smartphones enables people to announce an emergency they’re observing in real-time. Because of this, more agencies are interested in programatically monitoring Twitter (i.e. disaster relief organizations and news agencies). But, it’s not always cle…

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published