nlp_toxic_comments

Sentiment analysis: Toxic comments identification

Jupyter NoteBook viewer

This project shows:

• ability to write structured code in Python.
• ability to use existing utilities (libraries) for processing
• preprocessing skills
• Text preprocessing:
  • lemmatization
  • working with regular expressions
  • text conversion tf-idf
• using Machine Learning models

The project includes:

1. Working with NLP:
   • Pre-processing, text transformation for Machine Learning models
2. Working with Machine Learning models:
   • Logistic Regression.
   • Decision Tree
   • Random Forest
   • XGBoost
   • LightGBM

Data description

Provided data - comments with markup about the toxicity of edits.

Column	Description	Column type
text	Comment	features
toxic	Indicator whether comment is toxic or not	target

Task

The online store is launching a new service. The store needs a tool that will detect toxic comments and send them to be edited or viewed. Build a model that will classify comments into positive and negative.

Task details

The customer is concerned about:

• The value of the F1 metric must be at least 75.

Libraries used

pandas numpy matplotlib seaborn scipy nltk xgboost lightgbm time re