Data Science Portfolio
Education
- Doctor of Philosophy in Chemical & Molecular Engineering at Stony Brook University (2018-2022)
Professional Certifications & Specializations
- Machine Learning Specialization by DeepLearning.AI (2023)
- Advanced-Data Analytics by Google (2023)
- Data Analytics by Google (2023)
- Business Intelligence by Google (2023)
Competency Skills
- Skillset: EDA - Data & Business Analysis - Modeling - Dashboarding & Reporting
- Programming Languages: Python and SQL
- Libraries: Numpy, Pandas, Matplotlib, Seaborn, and SciKit Learn
- Software Programs: Jupyter Notebooks, Excel, Tableau and Power BI
Projects
1. Predictive Machine Learning Analytics for Telecom Customer Churn
- Demonstrated expertise in business analytics, encompassing a range of objectives, such as understanding business cases, classifier model implementation, data handling, exploratory data analysis, and interactive data visualization.
- Proficiently prepared and employed data to develop and train logistic regression, support vector machine, random forest, k-nearest neighbor, and naive bayes classifier models for predicting customer churn.
- The results of the comparison between different classification models reveal that the Random Forest model achieved an F1 score of 90%.
- Future work will aim to enhance the model further through the utilization of advanced techniques such as grid search.
2. Medical Insurance Premium Prediction with Machine Learning
- Collected data from a diverse sample of insured individuals, covering age, gender, BMI, smoking habits, region, and medical charges.
- Conducted feature selection to pinpoint key factors for predicting insurance premiums.
- Utilized linear regression and artificial neural network (ANN) models, resulting in R-squared values of 0.69 and 0.74.
- Optimized ANN hyperparameters, increasing the R-squared value to 0.97.
3. Deep Learning Respiratory Disease Detection and Classification
- Compiled a medical imaging dataset with X-rays and CT scans featuring respiratory disease conditions, including Viral and Bacterial pneumonia, and Covid-19.
- Explored deep learning models, focusing on employing ResNet50 in conjunction with additional Convolutional Neural Networks (CNNs) for image classification tasks.
- Conducted model evaluation and hyperparameter tuning, leading to the choice of a robust CNN architecture.
- Achieved effective classification of healthy and viral pneumonia (F1 > 87%), with room for improvement in bacterial pneumonia (F1 < 63%) and Covid-19 (F1 < 75%).
- Future work will focus on enhancing bacterial pneumonia and Covid-19 classification.
Additional Projects
- Machine Learning with PySpark: Customer Churn Analysis
- Deep Learning for Real-Estate Price Prediction
- Predict Sales Revenue with scikit-learn
- NBA Rookies Career Longevity using Scikit-learn
- Diabetes Disease Detection with XG-Boost and Neural Networks
- Medical Diagnosis using Support Vector Machines Will add when as soon as possibe. TBD