toxic-comment-classification

Toxic Comment Classification Challenge The task was to classify online comments into 6 categories: toxic, severve_toxic, obscene, threat, insult, identity_hate.

Summary of approach

Private score: 0.9826 / Public score: 0.9833

Data

FastText pre-trained embedding vector is used on this project.
Please download word2vec binary file from fasttext site And copy word2vec binary file to the data folder.

└── data
    └── crawl-300d-2M.vec
    └── sample_submission.csv
    └── test.vec
    └── train.vec

Usage

Training locally

$ python main.py --output_file_path submission_result.csv

Requirements

• python 3
• keras
• numpy
• matplotlib
• tensorflow
• pandas
• sklearn