Sentiment Analyzer

This project implements a sentiment analysis system using classical machine learning techniques in Python. It classifies tweets as positive or negative by applying text preprocessing, feature extraction with TF-IDF, and a Linear Support Vector Classifier.

---

Overview

The code performs the following operations:

• Loads a tweet dataset from a CSV file
• Cleans and preprocesses each tweet by:
  • removing links, mentions, hashtags, punctuation
  • converting text to lowercase
  • removing stopwords
  • lemmatizing words
• Vectorizes the cleaned text using TF-IDF with n-grams up to trigrams
• Splits the dataset into training and testing sets (80%/20%)
• Trains a calibrated LinearSVC classifier for binary sentiment prediction
• Evaluates the model with a classification report and accuracy score
• Provides a function to predict sentiment probabilities for new text inputs

---

Approach

• Machine Learning Model: Linear Support Vector Classifier (LinearSVC) with probability calibration
• Feature Extraction: TF-IDF vectorizer (unigrams, bigrams, trigrams)
• Preprocessing Steps:
  • Lowercasing
  • Removing mentions, hashtags, URLs
  • Removing non-alphabetic characters
  • Removing stopwords
  • Lemmatization
• Train/Test Split: 80% training data, 20% testing data
• Evaluation: Classification report and accuracy score

---

Dataset

The project expects a CSV dataset named dataSet.csv with the following columns:

Column	Description
target	Sentiment label (0 = negative, 4 = positive)
id
date
query
user
text	The tweet text
Note: Download the Sentiment140 dataset from Kaggle and rename it to dataSet.csv before using it.

---

How to Run

1. Clone the project:

   git clone https://github.com/hafbldprin/SentimentAnalyzer.git

2. Navigate to the project folder:

   cd SentimentAnalyzer

3. a Run MakeFile:

   make

4. b Set up a virtual environment:

   python -m venv venv

5. Activate the environment:

   • macOS/Linux:

     source venv/bin/activate

   • Windows:

     venv\Scripts\activate

6. Install dependencies:

   pip install pandas scikit-learn nltk

7. Place your CSV dataset in this folder and rename it to dataSet.csv.

8. Train the model:

   python train.py

9. Run the sentiment analyzer:

   python analyzer.py

---

Repository Contents

• load.py
  Loads the CSV dataset, applies filtering and cleaning, and trains the model

• analyzer.py
  Runs the trained model and performs sentiment analysis on new input

• preprocess.py Helper function for cleaing data set

• dataSet.csv Helper function for cleaing data set

• requirements.txt List of dependencies (Packages used by this project)

• Makeup
  Auto-run script to train and run the model

• README.md
  Description of the project

---