Projects
This section highlights some of the key projects I have worked on, showcasing my expertise in data analysis, visualization, and advanced analytics. Each project reflects my ability to transform complex data into actionable insights using tools like Power BI, Python, and SQL. Explore how I’ve applied data-driven strategies to solve real-world challenges and drive meaningful results.
Mortage Trading Analysis in
Power BI
Harnessed Power BI to analyze and manage mortgage data for capital market trades. Cleaned and transformed loan data to ensure accuracy, calculating key financial metrics such as Loan to Value Ratio (60.44%) and Debt to Income Ratio (31.70%).
In bid analysis, identified top bids (e.g., 108.68) and benchmarked them against UMBS bond prices, revealing that 27.33% of loans were viable trading candidates. Calculated a $15.14M trade premium and achieved a 7.073% Loan Profit Margin, exceeding the 5% target.
Leveraged Power BI’s Key Influencers visual to uncover critical factors affecting loan prices, such as Loan Amount and FICO Score. Designed interactive dashboards summarizing trade execution and portfolio performance, enabling data-driven decision-making.
Outcome: Enhanced skills in data transformation, DAX calculations, and financial benchmarking while delivering actionable insights for strategic mortgage trading.
HR Analytics Dashboard in Power BI
Developed a comprehensive HR analytics dashboard in Power BI to evaluate employee demographics, performance metrics, and attrition rates, facilitating data-driven decisions for HR departments.
Uncovered critical HR insights by analyzing average salaries and attrition rates; revealed a mean wage of $58,000 for diverse ethnic groups informing, fair pay assessments and identified that frequent travelers had a 24.9% attrition rate while employees with five years of tenure had a 17.4% attrition rate, enabling targeted retention strategies.
Enhanced performance tracking and demographic analysis through interactive dashboards utilizing advanced DAX functions (e.g., USERELATIONSHIP()), providing actionable insights into individual and departmental performance metrics and contributing to a 15% reduction in employee turnover.
Ecommerce Analysis in Power BI
Conducted an analysis for Whiskique, an online pet supply company, focusing on customer behavior, sales, and shipping metrics. Cleaned and structured data to reveal insights like California leading in customers (598) and Florida in East Coast sales.
Analyzed Customer Lifetime Value (LTV), with Nevada averaging $972.97, and identified Memory Foam Pet Beds as the second-best seller at $127,642. Built profit metrics including Cost of Goods Sold (COGS) and profit margins discovering a 27.50% overall profit margin. Highlighted $10,904 in shipping savings for Dog and Puppy Pads through optimized quantities, reducing total shipping costs by $118.19K.
Utilized market basket analysis to recommend cross-sell opportunities, such as promoting Memory Foam Pet Beds. Designed interactive dashboards showcasing sales trends, shipping metrics, and cross-sell strategies, providing actionable insights for decision-makers.
Outcome: Improved decision-making by uncovering key sales trends, optimizing shipping costs, and identifying high-margin products, thereby enhancing profitability and operational efficiency.
Recipe Popularity Prediction Model, Tasty Bytes Platform
Executed data preprocessing and exploratory analysis on a culinary dataset, employing statistical and machine learning techniques to identify key drivers of website traffic. Developed a logistic regression model that distinguished high from low traffic recipes with 77.4% accuracy and an AUC of 0.87, enhancing the platform's content strategy. Guided recipe selection with a feature importance analysis to boost user engagement and website traffic. Outlined strategic recommendations for the model's deployment and ongoing enhancement to align with shifting culinary trends and consumer preferences.
Achieved a significant increase in platform traffic and user engagement through strategic analytics and model application.
Enhanced content strategy with actionable insights derived from advanced data analysis, resulting in sustained user interest and interaction.
DetectAI: Web-Based AI text Authenticity Checker
Orchestrated the development of a system to detect AI-generated text within 50,000 chatbot interactions, integrating OpenAI’s dataset to enhance data security and minimise exposure risks. Advanced text classification accuracy by 40% through the creation of machine learning models utilising Naive Bayes, SVM, and XGBoost techniques.
Successfully reduced exposure to AI-generated text by 89%, safeguarding the integrity of communications.
Pioneered a suite of robust machine learning models, setting a new standard for AI text detection efficacy.
BeanSorter: Automated Dry Bean Seed Classification System
Spearheaded the implementation of a machine learning algorithm for classifying seven distinct types of dry bean seeds via image analysis, which halved classification errors. Enhanced the UCI Dry Bean dataset through meticulous feature extraction and normalisation, readying it for advanced predictive modelling.
Attained a 95% accuracy rate in seed classification, significantly outperforming previous models and optimising operational decision-making.
