Sentiment Analysis Using LSTM

Overview

This project focuses on sentiment analysis utilizing Long Short-Term Memory (LSTM) networks. Sentiment analysis involves determining the sentiment expressed in textual data, categorizing it as positive, negative, or neutral. LSTM, a type of Recurrent Neural Network (RNN), is particularly effective for this task due to its ability to capture long-term dependencies in sequential data.

The data for this project was retrieved from user reviews of the DANA App on the Google Play Store. These reviews were used to train and evaluate the model.

---

Key Features

• Data Collection from Play Store: Extracted reviews from the DANA App on Google Play Store for analysis.
• Data Preprocessing: Cleaned and tokenized text data to prepare it for model training.
• LSTM Model Implementation: Utilized LSTM networks for effective sentiment classification.
• Performance Metrics: Included accuracy evaluation to measure model effectiveness.
• Visualization: Provided visual insights into data distribution and model performance.

---

Key Steps

1. Data Collection:

   • Gathered user reviews from the DANA App on the Google Play Store.
   • Labeled reviews with corresponding sentiment categories (positive, negative, neutral).

2. Data Preprocessing:

   • Cleaned raw text to remove noise.
   • Tokenized and transformed the text into numerical format using embedding techniques.

3. Model Building:

   • Built an LSTM-based neural network using TensorFlow/Keras.
   • Designed layers with appropriate activation functions and optimizers.

4. Model Training:

   • Trained the model using the labeled dataset.
   • Monitored accuracy and loss during training.

5. Evaluation and Testing:

   • Tested the model on unseen data to assess its generalization capability.
   • Achieved an accuracy of 85% on the test set.

---

Libraries Used

The project leverages the following Python libraries:

• TensorFlow: For building and training the LSTM model.
• Keras: A high-level API within TensorFlow for constructing neural networks.
• NumPy: For numerical operations and handling arrays.
• Pandas: For data manipulation and analysis.
• Scikit-learn: For data preprocessing, splitting, and evaluation metrics.
• Matplotlib: For creating visualizations of data and model performance.
• Seaborn: For advanced visualization and insights.

---

Results

The LSTM model successfully achieved an accuracy of 82%, demonstrating its capability to classify sentiments effectively. The model performed particularly well on longer reviews, where context played a significant role in sentiment determination.

---

Project Limitations

• Data Limitations:

  • The dataset is limited to reviews from the DANA App, which may not generalize well to other applications or domains.
  • Imbalanced classes in the dataset could affect the model's ability to predict less-represented sentiments.

• Computational Resources:

  • Training LSTM networks is resource-intensive, which may limit scalability and experimentation with larger datasets.

• Domain Generalization:

  • The model may require retraining or fine-tuning to work effectively with reviews from other apps or domains.

---