Priyanshuiiitg/sentiment-analysis-svm

Sentiment Analysis with Support Vector Machine (SVM)

This project implements a sentiment analysis model using a Support Vector Machine (SVM) to classify movie reviews as positive or negative. The model is built using Python and various libraries, including Scikit-learn, Pandas, and Matplotlib.

Table of Contents

1. Introduction
2. Requirements
3. Data
4. Installation
5. Usage
6. Model Training
   • Preprocessing
   • Feature Extraction
   • Splitting the Data
   • Hyperparameter Tuning
   • Model Evaluation
7. Results
8. Conclusion
9. License

Introduction

Sentiment analysis is a Natural Language Processing (NLP) technique used to determine the emotional tone behind a body of text. In this project, we will classify movie reviews from the Rotten Tomatoes dataset into positive and negative categories using an SVM classifier.

Requirements

To run this project, you need to have Python installed along with the following libraries:

• pandas
• scikit-learn
• matplotlib
• seaborn

You can install the required libraries using pip:

pip install pandas scikit-learn matplotlib seaborn

Data

The dataset used in this project consists of movie reviews labeled as positive or negative. It includes two files:

• rt-polarity.pos: Contains positive movie reviews.
• rt-polarity.neg: Contains negative movie reviews.

Make sure to place these files in the same directory as the script.

Installation

1. Clone this repository:

   git clone https://github.com/Priyanshuiiitg/sentiment-analysis-svm.git
   cd sentiment-analysis-svm

2. Ensure you have all the required libraries installed (as mentioned above).

Usage

1. Place the rt-polarity.pos and rt-polarity.neg files in the project directory.

2. Run the Python script:

   python app.py

Model Training

Preprocessing

The script preprocesses the text data by:

• Converting text to lowercase.
• Removing special characters and digits.

Feature Extraction

The TfidfVectorizer is used to convert text data into numerical format, focusing on the most relevant features while removing common stop words.

Splitting the Data

The dataset is split into training, validation, and test sets:

• Training Set: 4000 positive and 4000 negative reviews.
• Validation Set: 500 positive and 500 negative reviews.
• Test Set: Remaining reviews.

Hyperparameter Tuning

GridSearchCV is used for hyperparameter tuning to find the best parameters for the SVM model. The parameters tuned include:

• C: Regularization parameter.
• kernel: Type of kernel (linear or RBF).
• gamma: Kernel coefficient.

Model Evaluation

The model's performance is evaluated using:

• Cross-validation scores.
• Accuracy on validation and test sets.
• Confusion matrix to visualize the model's predictions.

Results

The script will output the following metrics:

• True Positive (TP)
• True Negative (TN)
• False Positive (FP)
• False Negative (FN)
• Validation Accuracy
• Test Accuracy
• Confusion Matrix
• Precision
• Recall
• F1 Score
• Classification Report

The confusion matrix will also be displayed as a heatmap.

Conclusion

The SVM model provides a robust method for sentiment analysis. By tuning hyperparameters and evaluating the model, we achieve good performance in classifying movie reviews. This approach can be extended to other text classification tasks.

License

This project is licensed under the MIT License. Feel free to modify and distribute it as you see fit.

Feel free to adjust any sections according to your preferences or project specifics, especially the GitHub repository link. You can copy and paste this directly into your README file!