Projects
(All projects are located in the github link)
AWS Face ReKognition and Face_recognition package Evaluation
-
Project Overview: The project aimed to assess the efficacy of two facial recognition systems, Amazon ReKognition and face_recognition package. In this context, we gave three images of MSBA students in a classroom. Then, we tested the systems to identify the individuals in the images using a collection of MSBA student headshots
-
Evaluation Focus: The project used critical performance metrics, namely accuracy, precision, recall, and confusion matrix to comprehensively gauge the capabilities of Amazon ReKognition and the face_recognition package. By prioritizing these metrics, the assessment provided an understanding of the systems' ability to identify my peers while ensuring a comprehensive and informed evaluation
-
Recommendation: After comprehensively evaluating facial recognition systems, we recommend Amazon ReKognition for student attendance verification. However, the face_recognition package showcased superior performance in specific areas like user-friendliness and a slightly higher percentage of correct matches. Amazon ReKognition's advanced capabilities, real-time efficiency, and cloud-based infrastructure offer a more robust solution for accurately identifying employees upon their return to work
Fraud Detection with Machine Learning
-
Project Overview: Analyzed a dataset of 125,000 transactions, performed exploratory data analysis, and implemented advanced machine learning techniques such as Logistic Regression, Random Forest, and Gradient Boosting Machine
-
Implemented data preprocessing techniques, including handling missing values through data pipelines and addressing class imbalance through the Synthetic Minority Oversampling Technique (SMOTE), resulting in a refined model that effectively detected fraudulent transactions while maintaining a balance between precision and recall
Music Recommendation System
-
Project Overview: Guided an in-depth exploratory analysis to find user preferences and behaviors, leading the implementation of recommendation systems using Singular Value Decomposition
-
Utilized Root Mean Squared Error (RMSE) and Mean Absolute Error (MAE) as benchmark evaluation metrics
Data-Driven Insights to Enhance Organic Product Sales Strategy
-
Project Overview: Analyzed over 22,000 observations of supermarket loyalty program data to model and predict customer behaviors regarding organic product purchases
-
Developed multiple predictive models, providing actionable insights and recommendations for increasing organic product sales, and created a PowerPoint for supermarket management, summarizing key insights, customer profiles, and strategic recommendations to boost organic product sales effectively
Airbnb Market Analysis & Modeling
-
Project Overview: Initiated a data-driven analysis of nearly 6000 Airbnb properties in Austin, TX, actively employing techniques such as linear regression, logistic regression, and K-means clustering to unveil key factors influencing nightly rental prices and bookings
-
Recommendations focused on optimizing host response rates, highlighting property amenities, and encouraging positive guest feedback to enhance overall booking performance
Graduate Consulting Project with Renowned Clothing Company
Project Overview: Conducting an in-depth analysis of multi-year product sales data for a renowned clothing company to forecast sales trends, optimize inventory management, and strategically determine order times for core products
Skills Utilized: Data Exploration, Data Visualization, Data Modeling, Data Pipelines, Strategy, Forecasting
