DATA PROJECTS
Student Performance Prediction for Academic Success
Conducted an in-depth analysis of student performance data to identify socio-economic factors contributing to academic success. Utilized data preprocessing techniques, and explored relationships between student attributes (study time, parental education, school support) with final grades through data visualizations. Applied machine learning algorithms including Linear Regression, Random Forest, and Gradient Boosting to predict student grades, with Random Forest achieving an R² score of 91.7%. Generated feature importance insights, and visualized key predictors affecting student outcomes.
Business Intelligence Project for an E-commerce Venture : To Understand Customer footfall
This project aims to leverage data-driven insights to make informed marketing and product decisions through a centralized data warehouse to integrate data from various sources, understand customer behavior, and improve retention and repeat purchases.
Heart Failure Prediction
This study aims to predict the accuracy of demographic and clinical factors like gender, age, and various test results leading to heart disease using various machine-learning models achieving a prediction accuracy of 91.9%.
Grocery-Store-Sales-Time-Series-Forecasting
The "Grocery Store Sales Time Series Forecasting" project focuses on predicting future sales using historical sales data. After performing exploratory data analysis and data cleaning, Fourier terms and the XGBoost model has been used. The results indicate that the model can be reliably used for inventory management and strategic planning. The project successfully forecasts future sales, providing insights that can help improve inventory management and decision-making in the grocery store.
Forecasting Enrollment Trends And Predicting Factors that Influence University Admission
This project aimed to explore how demographics, citizenship, and scholarship levels impact the conversion from inquiry to admission, enrollment, and beyond. Key factors affecting enrollment likelihood were identified and used Time Series Forecasting to predict departmental enrollments over the next two years.
In our analysis of advertising campaigns, we focused on optimizing the new-to-brand segment by leveraging exploratory data analysis and machine learning techniques. Using Tableau and Python, we identified key features influencing cost and impressions, including placement slots, targeting categories, months, and days of the week. Our goal was to propose an optimal bidding strategy to targeted advertisers, focusing on maximizing viewable impressions and reach. By identifying the most significant factors, we aimed to reduce missed opportunities and pinpoint areas for improvement and growth within the campaign strategy.
