Tweet Sentiment Classifier with Machine Learning Models

This project aims to develop a sentiment classifier model using an equally distributed subset of kaggle’s Twitter Sentiment Analysis dataset. The objective is to classify each tweet into one of the sentiment classes using various machine learning models, and optimize their performance through hyperparameter tuning and evaluation techniques.

Project Overview

The key objectives were:

1. Load and Preprocess Data: We utilized Kaggle’s Twitter Sentiment Analysis Dataset, balanced the dataset, and applied text preprocessing techniques including lemmatization and special token replacement.
2. TF-IDF Vectorization: Conversion of text data into numerical features using TF-IDF, focusing on unigram and bigram features, in order to prepare the data for classification.
3. Classification Models: Implementation and evaluation of several classifiers, including:
   • Dummy Classifier (Baseline)
   • Naive Bayes
   • Logistic Regression
   • Multi-Layer Perceptron (MLP)
   • k-Nearest Neighbors (k-NN)
   • Support Vector Machines (SVM)
4. Hyperparameter Tuning: Utilized Grid Search to optimize the Logistic Regression model.
5. Evaluation Tools: Applied learning curves and precision-recall curves to evaluate model performance.

Dataset

The dataset used for this project is a Twitter Sentiment Analysis Dataset from Kaggle, which contains 100,000 entries. We balanced the dataset to include 5,000 instances each for positive and negative sentiments.

Preprocessing

The preprocessing steps included:

• Tokenization and Lemmatization: Standard NLP techniques to reduce the dimensionality of text data.
• Special Token Replacement: Replaced URLs, usernames, hashtags, stopwords, etc., with placeholders.
• TF-IDF Vectorization: Transformed the text data into numerical features using the TfidfVectorizer from scikit-learn.

Classification Models

Dummy Classifier

A baseline model that predicts the most frequent class. This model serves as a benchmark for comparison with more sophisticated classifiers and also helps in identifying class imbalances.

Naive Bayes

A probabilistic classifier based on applying Bayes' theorem. Used both a simple implementation and a version with cross-validation.

Logistic Regression

A linear model for binary classification. Used both simple Logistic Regression and a version with cross-validation, leading to slight improvements in F1 scores.

Multi-Layer Perceptron (MLP)

A neural network model that was evaluated for its ability to classify the sentiment.

k-Nearest Neighbors (k-NN)

A non-parametric classification algorithm based on majority voting among the k nearest data points.

Support Vector Machines (SVM)

A classifier that finds the hyperplane that best separates the classes in a high-dimensional space. Also applied Singular Value Decomposition (SVD) for dimensionality reduction.

Hyperparameter Tuning

For Logistic Regression, we conducted hyperparameter tuning using Grid Search, optimizing parameters like the solver, penalty, and the number of features in the TF-IDF vectorizer.

Evaluation

Learning Curves

Used to analyze how the model's performance improves with more data, helping to detect underfitting or overfitting.

Precision-Recall Curves

Evaluated the performance of models, particularly useful for imbalanced datasets. The area under the precision-recall curve (AUC-PR) was used as a summary metric.