In this project, we aim to develop a predictive model using machine learning classification techniques to assess whether individuals seeking loans pose a risk of defaulting or not. Leveraging a dataset containing a variety of independent variables such as checking account balance, credit history, purpose, and loan amount, our objective is to create a robust model that accurately categorizes loan applicants as either defaulters or non-defaulters.
This project is supervised, as we are using labeled data (defaulters vs. non-defaulters) to train the predictive model. The model will learn from historical data to make predictions on new loan applicants.Skills and libraries include decision tree modeling, ensemble methods (bagging and boosting), exploratory data analysis (EDA), and feature importance assessment using GINI impurity. Python libraries such as scikit-learn for machine learning and matplotlib/seaborn for data visualization are utilized.
The primary goal of this project is to utilize machine learning techniques, specifically neural networks, for cancer prediction based on gene expression data. By analyzing patterns in gene expression profiles across various tumor types (BRCA, KIRC, COAD, LUAD, PRAD), the project aims to develop a predictive model that accurately classifies patients into their respective cancer types. The focus is on enhancing diagnostic capabilities and exploring potential biomarkers that can aid in early detection and treatment planning.
Skills and tools employed include deep learning with neural networks, Python programming for data preprocessing and model development, TensorFlow or PyTorch for building and training neural networks, scikit-learn for data manipulation and evaluation metrics, and data visualization libraries like matplotlib and seaborn for result interpretation.
Analyzing subscription churn rates for Codeflix, a streaming service, using advanced SQL techniques to provide insights into user retention. The goal is to identify patterns in user cancellation behavior over time and differences between user groups to assist Codeflix in improving subscriber retention strategies.
SQL techniques such as aggregates, unions, temporary tables, cross joins, and case statements are employed to extract meaningful insights from the data. Advanced SQL techniques for data manipulation and analysis, including aggregates, unions, temporary tables, cross joins, case statements, and aliasing.
This project showcases the development of a Streamlit application that integrates conversational AI to analyze health and wellness data from Oura and WHOOP devices. The app provides users with intuitive, data-driven insights into their activity, sleep, and recovery metrics, emphasizing user engagement and personalized health tracking."
Developed expertise in Streamlit for creating interactive web applications, implemented conversational AI for user engagement, conducted data analysis and visualization using Python libraries like Pandas and Matplotlib, integrated APIs for real-time data retrieval, and managed back-end development with a focus on efficient data pipelines and seamless user experience.
TalkPDF is a Streamlit-based web application that enables users to interact with their PDF documents in a conversational manner. With TalkPDF, you can upload a PDF file, extract its content, and ask questions about it, leveraging the power of OpenAI's GPT models. The goal of this project is to bridge the gap in science communication.
- Streamlit for building an intuitive and interactive web interface, pdfplumber for extracting text from PDF documents, OpenAI's GPT API for conversational AI integration and Python for back-end development and data handling.
There is also an FPL project in the Streamlit folder where I integrated AI to analyze FPL data.