Projects

Customer Churn Predictor

Developed a customer churn prediction system for a fictional telecommunication use case.
Trained and evaluated various machine learning models, including Logistic Regression, SVC, Random Forest Classifier, Decision Tree Classifier, XGBoost Classifier, and LightGBM Classifier.
Performed feature selection using Recursive Feature Elimination and tuned hyperparameters for the best performing model, achieving an accuracy score of 81.1% and an F1 score of 80.6% with the Logistic Regression model.
Saved the best model using joblib and used it to make predictions in a web application.
Built an interactive web application using Streamlit and deployed it on Streamlit Cloud, allowing users to input customer data and receive churn predictions.
Gained experience in data science, machine learning, and web development, and created a tool that can help businesses predict customer churn.

Developed an automatic speech recognition application using OpenAI's Whisper model that generates subtitles and transcriptions for YouTube videos and video files, improving accessibility and user experience.
The application performs 2 tasks:
- Detects the language, transcribes the input video in its original language.
- Detects the language, translates it into English and then transcribes.
Utilized ffmpeg to generate subtitled videos, enhancing user experience and accessibility.
Built a user-friendly interface using the Streamlit library to provide an intuitive experience for users.
Hosted the application on HuggingFace Spaces to reach a wider audience.
Gained hands-on experience in speech recognition, natural language processing, and video processing, while creating a powerful tool that improves accessibility for video content.

Developed a TV series recommender system that uses a K-Nearest Neighbors model to make recommendations based on user preferences.
Scraped the IMDb website using the BeautifulSoup library to create a dataset of TV series, which was saved as a CSV file.
Built an interactive web application using the Streamlit library to provide users with an easy-to-use interface for inputting their preferences and receiving recommendations.
Utilized the Euclidean distance metric in the K-Nearest Neighbors model to find TV series that are most similar to a user's input.
Incorporated the Youtube Data API v3 to dynamically retrieve and display trailers of recommended TV series within the web application.
Deployed the web application on Heroku, making it easily accessible to users online.

Developed a movie recommender system using the IMDB 5000 dataset, which includes data on movie titles, release dates, cast, directors, storylines, ratings, and trailers.
Utilized the K-Nearest Neighbors algorithm to recommend movies to users based on their preferences.
Dynamically retrieved trailers for recommended movies using the Youtube Data API v3, adding an interactive and engaging component to the web app.
Built the web application using the Streamlit library, providing users with an intuitive and user-friendly interface for entering their preferences and receiving recommendations.
Deployed the web application on Heroku, making it easily accessible to users online.

Developed a news summarizer application that provides users with a quick and easy way to get an overview of the latest news stories.
Built functions using the Beautiful Soup library to scrape news articles from Google News.
Implemented a categorization system that allows users to choose between trending news, favorite topics, or to search for a specific topic of interest.
Used the Newspaper3k library to fetch news, news posters, and automatically generate summaries of each news article.
Built the web application using the Streamlit library, creating a user-friendly interface for users to input their preferences and view the summarized news stories.
Deployed the web application on Streamlit Cloud, allowing users to access the application from anywhere with an internet connection.