Skip to content

Latest commit

 

History

History
28 lines (18 loc) · 898 Bytes

README.md

File metadata and controls

28 lines (18 loc) · 898 Bytes

kaggle-nlp-disasters

https://www.kaggle.com/c/nlp-getting-started

tweet example

You can imagine how an NLP classifier would mistakenly label this tweet as one referring to a real emergency, since it uses the word ablaze.

Competition Description

Agencies and organizations will mine twitter, sometimes for emergency response.

The Kaggle challenge is as follows:

Train an NLP classifier that can do well-enough at detecting real emergencies from a set of tweets.

Data

  • sample_submission.csv: Used for formatting your output in a way that Kaggle can score you
  • train.csv: Data you will use for training and validation
  • test.csv: Hold-out set

Here is an example of training data:

train dataframe image

training data quality