Projects
Each project highlights my skills and expertise in handling complex data sets, deriving insights, and providing valuable recommendations. Feel free to explore and learn more about my work.
Amazon-Themed Sales Dashboard Using PowerBI
Data Visualization
Project Overview: Designed an Amazon-themed dashboard to provide granular sales insights, driving strategic decision-making.
-
Business Impact: Increased regional targeting efficiency by 15% through comprehensive, real-time sales analysis.
-
Data Integration: Combined data from multiple sources with 100% accuracy, utilizing data cleaning techniques to ensure reliability.
-
Key Performance Metrics: Devised custom KPIs and calculated metrics using DAX, illuminating high-performing regions for executive decisions.
-
Brand-Centric Design: Styled the dashboard with Amazon branding, enhancing user engagement and improving accessibility for stakeholders.
Office of IT Project on Ticket Analysis (UTA OIT)
Data Analysis
Project Overview: Analyzed ticket patterns in UTA OIT’s ServiceNow system to optimize support workflows and reduce “bouncing tickets.”
-
Efficiency Gains: Enhanced resolution times by 20% for recurring issues through in-depth ticket trend analysis.
-
Workflow Optimization: Collaborated with IT teams, achieving a 12% increase in client satisfaction by streamlining escalation paths.
-
Predictive Analytics: Implemented predictive modeling to proactively assign high-latency tickets, reducing average resolution time.
-
Data Visualization: Developed visual reports that reduced unnecessary ticket escalations by 15%, fostering a more efficient support process.
Predicting Car Price and Classifying by Price Range Using Vehicle Features
Regression and Classification
Project Overview: Created regression and classification models to predict car prices and classify vehicles by price range, aiding strategic marketing.
-
Prediction Accuracy: Achieved 87% accuracy in price predictions, helping sales teams with market positioning strategies.
-
Targeted Marketing: Improved price range classification accuracy by 22%, supporting tailored marketing approaches.
-
Feature Engineering: Utilized Python and scikit-learn for robust feature selection, increasing model precision by 18%.
-
Insight Delivery: Visualized factors influencing price, providing actionable insights to the sales and marketing teams.
User Behavior Analysis for Shared Mobility Services
Data Visualization
Project Overview: Conducted in-depth user behavior analysis for shared mobility services to enhance customer experience and optimize service distribution.
-
Resource Optimization: Improved service allocation by 25% based on data-driven insights into peak usage times and popular routes.
-
Promotional Efficiency: Increased targeting accuracy by 18% through customer segmentation and tailored promotions.
-
Real-Time Monitoring: Built Tableau dashboards for tracking user metrics, enabling agile decision-making.
-
Retention Strategy: Identified key retention factors, reducing customer churn by 15% through data-backed service enhancements.
Predicting Cancer Diagnosis Using Tumor Characteristics
Predictive Analysis
Project Overview: Developed a risk scoring model to predict breast cancer diagnosis, aiding early detection and treatment planning.
-
Clinical Accuracy: Achieved 92% accuracy in predicting malignancy, supporting early diagnosis and personalized care.
-
Feature Engineering: Improved model precision by 19% through clinical feature selection, enhancing diagnostic reliability.
-
Model Evaluation: Used ROC and AUC metrics to validate performance, optimizing model selection for clinical application.
-
Visualization for Impact: Designed risk score dashboards, empowering medical professionals with data-driven insights for patient care.
Intelligent Facial Recognition Using Deep Learning
Deep Learning
Project Overview: Built a facial recognition model to enhance security through real-time identification and authentication.
-
Recognition Accuracy: Achieved 95% accuracy, using CNN-based architecture for robust facial recognition.
-
Data Optimization: Employed image normalization and augmentation, enhancing model performance by 20%.
-
Technical Tools: Leveraged Python, TensorFlow, and OpenCV to streamline model training and facial feature extraction.
-
False Positive Reduction: Established validation frameworks to reduce false positives by 15%, ensuring reliable recognition.
Toxic Comment Classifier Using LSTM
Deep Learning
Project Overview: Created an LSTM model to detect toxic comments, supporting automated moderation and promoting a positive online community.
-
Model Accuracy: Reached 90% accuracy in classifying toxicity, using extensive NLP preprocessing techniques.
-
Training Process: Employed Keras and TensorFlow, optimizing hyperparameters to reduce validation loss by 15%.
-
Real-Time Monitoring: Developed dashboards to track toxicity levels, enabling swift moderation responses.
-
Community Impact: Streamlined content moderation efforts, enhancing community standards through automated detection.
