/Toxic_Comments_Kaggle

Primary LanguageJupyter Notebook

Toxic_Comments_Kaggle

  • Input

    • glove.840B.300d.txt: GloVe (Global Vectors for Word Representation), obtainable from GloVe
    • train.csv: training data
    • test.csv: test data
    • test_labels.csv: test labels for toxic comments toxicity
    • sample_submission.csv: sample submission form for Kaggle score. Used to generate submission.csv and submission_LSTM_1DCNN.csv

  • Code

    • EDA.ipynb: notebook with data exploration, corpus cleaning, feature engineering and Logistic Regression with baseline Naive-bayes SVM classifier model
    • main.ipynb: Simple logistic Regression model
    • LSTM_1DCNN.py: Toxic Comment classification using LSTM and 1D-CNN model
    • xgb-baseline-with-feature-engineering: Multi-label classification using XGBoost and feature engineering. Notebook runs on Kaggle here