Developed and deployed a sentiment analysis model using a comprehensive approach to data preprocessing, feature engineering, and model evaluation. Leveraged the Multinomial Naive Bayes algorithm for its superior accuracy, recall, and precision in classifying sentiment.
Algorithms Implemented: Multinomial Naive Bayes, Gaussian Naive Bayes, Logistic Regression (One vs All), SVM (Sigmoid), Random Forest (Gini, Entropy), Bag of Words, TF-IDF, Word Embedding, Python (NLTK, scikit-learn), Streamlit
Cleaned and preprocessed the dataset using techniques such as stemming and lemmatization to enhance data quality.
Implemented feature engineering methods including Bag of Words, TF-IDF, and Word Embedding to transform text data into meaningful features.
Evaluated multiple machine learning models (Gaussian Naive Bayes, Logistic Regression One vs All, SVM Sigmoid, Random Forest Gini, Random Forest Entropy) to identify the best-performing algorithm.
Selected Multinomial Naive Bayes based on its superior performance in terms of accuracy, recall, and precision.
Deployed the trained model using Streamlit, creating an interactive web application for real-time sentiment analysis.
Ensured scalability and robustness of the deployed application to handle real-time data.
Achieved high accuracy, recall, and precision with the Multinomial Naive Bayes model, demonstrating its effectiveness for sentiment analysis.Enhanced the preprocessing pipeline to ensure the dataset was clean and well-prepared for model training.
Delivered a robust and scalable sentiment analysis tool through Streamlit, providing valuable insights into customer sentiment via an interactive web interface.
This project showcases my expertise in machine learning, data preprocessing, feature engineering, model evaluation, and deploying machine learning solutions with Streamlit for practical applications.