/Binary-Classification-for-Sentiment-Analysis

Binary Classification for Sentiment Analysis

Primary LanguageJupyter Notebook

Binary-Classification-for-Sentiment-Analysis

This repository contains different Binary Text Classification Techniques for Sentiment Analysis (work during my internship period)

  • Using WordNet Lemmatizer
  • Using Spacy Lemmatizer Without Pretrained Word Embeddings
  • Using Spacy Lemmatizer With Pretrained Word Embeddings

Data

It uses Amazon reviews for sentiment analysis dataset from Kaggle.

Layers

For Machine Learning part, it uses different layers like LSTM, Embedding, Dense, SpatialDropout1D, Dropout, Flatten and GlobalMaxPooling1D.

Results

WordNet: Best result is 0.77 using Sequential model
Spacy Without Pretrained Embeddings: Best result is 0.81 using Sequential model with LSTM layer.
Spacy With Word2Vec Pretrained Embeddings: Best result is 0.95 using Sequential model with LSTM layer.