Personal Portfolio | Robert J Swetland

Proactive Problem-Solving: Identified problematic areas and implemented strategic solutions to enhance operations and minimize disruptions.
System Audits: Conducted audits of systems and assessed outcomes to ensure optimal performance and compliance.
Asset Preservation: Maintained information security, control structures, and the preservation of organizational assets.
Budget and Vendor Management: Assisted the CIO in budget development, equipment procurement, project completion, vendor negotiations, and ensuring cost-effectiveness.
Inventory and License Management: Oversaw inventory maintenance and software license compliance.
Documentation: Maintained comprehensive systems documentation for IT operations.
Technical Support: Provided escalated on-site support for hardware, software, and network connectivity issues.
Office 365 Administration: Managed all aspects of Office 365, including the office suite, SharePoint, and collaboration tools.
Training Materials: Created training materials for IT systems tailored to end users with varying technical abilities.
Communication: Acted as the primary liaison between IT and operational, business, and functional units.
Quality Initiatives: Supported quality measures by completing tasks related to organizational initiatives.
JIRA Implementation: Led the buildout, implementation, and support of JIRA for five different departments.
Technology Integration: Gathered and interpreted business needs across departments, suggested technology solutions, and oversaw the lifecycle of those implementations.
Reporting and KPIs: Conducted daily team huddles, prepared weekly KPI reports for the CIO, and presented at monthly management meetings.
Vendor Oversight: Managed and oversaw all technology MSP partners and vendors.
Roadmap Development: Translated multi-disciplinary departmental needs into technological roadmaps.
Technology Solutions: Planned, developed, trained on, deployed, and maintained technology solutions, including custom programming, API development and ingestion, HL7 pipelines, and in-house software solutions.
Cross-Department Collaboration: Regularly attended multi-disciplinary staff meetings to understand business needs, develop technological roadmaps, and provide ongoing education and support for systems.

High-Availability Network: Designed and maintained a high-availability network infrastructure supporting 16 locations across 7 cities and one mobile care unit.
Switch Infrastructure: Configured, installed, and optimized 35 Juniper/Mist Layer 2/Layer 3 switches to enhance network performance.
Wireless Connectivity: Deployed and managed 80 Mist access points (APs) for reliable wireless coverage.
Server Management: Administered 23 physical and 35 virtual application and storage servers to support organizational operations.
User Devices: Oversaw 500+ user devices, including desktop PCs, laptops, iPads, Surface tablets, and associated peripherals.
Printing Solutions: Managed 100+ printers, including high-volume multifunction printers (MFPs), prescription printers, desktop printers, label printers, and specialty devices.
Medical Equipment: Supported ECG systems, dental X-ray sensors, and panoramic imaging systems at three locations.
Legacy Systems: Maintained two on-premises legacy EMRs, ensuring operational continuity during transitions.
SharePoint 365: Administered over 200 SharePoint 365 sites, ensuring security, compliance with policies, and uninterrupted collaboration.

Network Upgrade: Successfully upgraded all networking equipment from Cisco Meraki to Juniper Mist, enhancing reliability and performance.
EMR Transition: Transitioned the company from an on-premises EMR to a cloud-based EMR (OCHIN Epic), improving accessibility and scalability.
Server Migration: Migrated servers from Windows Server 2008 R2 and 2012 to Windows Server 2019 and 2022, ensuring system modernization and improved performance.
Hybrid System Implementation: Migrated operations from a fully on-premises setup to a hybrid infrastructure, leveraging two on-premises data centers, a colocation facility, and cloud platforms (AWS and Azure).
Emergency Alert System: Installed, configured, and implemented an emergency mass alert system (Alertus) company-wide, providing training for seamless adoption.
Comprehensive Documentation: Documented all IT assets, processes, SOPs, and training materials for use by both technical teams and staff.
Automation: Automated numerous routine tasks through Bash and PowerShell scripting, significantly improving operational efficiency.
IT Governance: Established IT policies, processes, and procedures, including the adoption of SCRUM and ITIL4 protocols to improve operational standards.
Asset Inventory Management: Implemented an asset inventory system to ensure accurate tracking and annual checks for all technology assets.
Surveillance System Design: Designed and implemented a video surveillance system across all clinics, enhancing security.
Mobile Clinic Communications: Upgraded mobile clinic communications from a Cradlepoint single SIM system to a custom multi-SIM system, ensuring reliable connectivity in remote locations.
Onboarding & Training: Standardized employee onboarding processes and technology training programs for consistency and efficiency.
File Management Migration: Began the migration from DFS to SharePoint 365, streamlining file management and collaboration.
Data Management: Introduced Data Management for healthcare applications using ETL methodologies, provided training, and led discussions on security measures and appropriate use.
AI in Healthcare: Introduced AI for healthcare applications using RAG, provided training, and led discussions on security measures and appropriate use.
*Data Insights and Visualization: Trained the data team to use scripting for analyzing large datasets and introduced Tableau for presenting actionable insights to stakeholders.

Employee Widget : Developed an intranet portal providing a comprehensive employment management solution, including withholding changes, contact/name updates, time-off requests, education request management, training logs, suggestion submissions, paystub access, and policy retrieval.

Patient Intake Paperwork : Transitioned the patient intake process from paper to electronic, incorporating pre-filled forms for physical signatures and enabling long-term storage and retrieval via a custom-coded SharePoint library application.

New Employee Orientation Binder : Digitized and customized the 500+ page orientation binder to provide a tailored user experience by discipline, leveraging SharePoint widgets, SQL stored procedures, and custom code. This reduced new employee turnover by simplifying the onboarding process.

SOAP Templates : Designed and developed Windows applications (initially in VB 6.0, later re-written in VB 12) to support nursing, social work, and spiritual care staff by prompting for relevant information during intake, routine, and crisis care visits. These tools pre-formatted data for seamless integration into EMR software.

Application Request Tool : Built an application to track, promote, and manage application requests in alignment with leadership goals. Automated communication of progress, archived discussions, and provided updates to stakeholders to enhance transparency and collaboration.

Catalog : Created an e-commerce platform to monitor the consumption and costs of promotional materials and other consumables, allocating expenses to appropriate cost centers.

Incident Report System : Developed a system for long-term tracking and disposition of patient-related incidents, enabling trend analysis, compliance tracking, and improved decision-making.

HRIS Portal : Converted employment records to an electronic format and streamlined employee management from application through all phases of employment.

Additional Projects : Delivered numerous other projects aimed at improving employee satisfaction, increasing efficiency, identifying trends, measuring ROI on training, enhancing compliance automation, and boosting company profitability.

Agile Development Adoption : Introduced and championed the adoption of the Agile development methodology, significantly improving stakeholder satisfaction and reducing scope creep.

Virtualization Project Lead : Led the IT department in optimizing server utilization through virtualization, reducing hardware replacement costs. Implemented long-term plans for maintenance and disaster mitigation using VMware and Microsoft Hyper-V platforms, presenting detailed strategies to IT teams and leadership.

Project Description

Project: Video Presentation Slide Extraction
Date: 2025
Description: This Video Presentation Slide Extraction softare automates the extraction of individual presentation slides from a video lecture. Using OpenCV for video processing, it sequentially reads frames and applies image analysis techniques to detect slide transitions.
Key methods include grayscale conversion , frame differencing , Structural Similarity Index (SSIM) comparisons, and Optical Character Recognition (OCR) via Tesseract to recognize textual changes indicative of slide transitions.
Additionally, the script employs perceptual hashing (phash) to identify and eliminate duplicate or near-identical slides, ensuring a set of unique images. The final output consists of individually saved slides, neatly compressed into a ZIP archive, simplifying review and sharing.

Key functionalities:
- Frame extraction and processing from video lectures.
- Slide transition detection using SSIM and OCR.
- Removal of duplicate slides through perceptual hashing.
- Automated creation of a ZIP file containing the extracted unique slides.
Language(s): Python 3
Skills: NumPy CV2 OS ZipFile SYS Subprocess PIL ImageHash SkImage PyTesseract
Project URL: Click Here

Project Description

Project: Image De-Duplication
Date: 2025
Description: The Image De-Duplication software is a powerful and efficient tool designed for detecting and eliminating duplicate images within a specified directory. Leveraging advanced hashing techniques, feature extraction, and machine learning-based similarity metrics, this software ensures precise and scalable image comparison. It supports both raster and vector image formats, including JPG, PNG, BMP, GIF, and SVG .

Key Features :
- Multi-Format Support : Detects duplicates in both raster images (JPEG, PNG, BMP, etc.) and vector images (SVG).
- Flexible Similarity Threshold : User-defined threshold (0.0 - 1.0) allows fine-tuning of duplicate detection sensitivity.
- Fast and Efficient Processing : Optimized for large datasets, utilizing OpenCV, NumPy, and Scikit-learn for fast computations.
- Robust Image Hashing & Feature Extraction :
  - Perceptual Hashing (pHash, aHash, dHash) : Identifies near-duplicates by comparing image fingerprints.
  - Histogram Analysis : Uses color distribution comparison for enhanced accuracy.
  - SVG Normalization & Hashing : Ensures consistent vector image comparison.
- Comprehensive Logging & Reporting : Generates detailed logs and outputs duplicate sets to a specified file.
- Cross-Platform Compatibility : Runs on Windows, macOS, and Linux with a simple Python 3 setup.

Ideal Use Cases

Organizing Large Image Libraries : Automatically detect and remove redundant files.
Preventing Redundant Storage : Optimize disk usage by identifying and eliminating duplicate images.
Automating Image Deduplication in Workflows : Useful for media companies, photographers, and digital asset management.

Language(s): Python 3
Skills: OpenCV NumPy Scikit-learn ImageHash Pillow ArgParse Collections Logging Time XML Parsing hashlib
Project URL: Click Here

Project Description

Project: MIT: MLS Case study: Unsupervised Learning
Date: 2024
Description: This project explores Unsupervised Learning techniques such as Clustering and Dimensionality Reduction to analyze customer segmentation. Methods like -Means and K-Medoids are applied to categorize customers based on spending behaviors and interaction patterns with the bank.
Skills: Pandas NumPy Matplotlib Seaborn Exploratory Data Analysis (EDA) Data Preprocessing Data Visualization K-Means Clustering K-Medoids Clustering Gaussian Mixture Model Unsupervised Learning
Project URL: Click Here

Project Description

Project: MIT: MLS Case Study PCA
Date: 2024
Description: This project explores Principal Component Analysis (PCA), a Dimensionality Reduction technique. It demonstrates how PCA can be used to transform high-dimensional data into a lower-dimensional representation while preserving the most significant variance.
Skills: NumPy Pandas Matplotlib Seaborn Scikit-learn Exploratory Data Analysis (EDA) Data Preprocessing Data Visualization Data Scaling Principal Component Analysis (PCA)
Project URL: Click Here

Project Description

Project: MIT: Socio-economic Factors for Geographic Clustering
Date: 2024
Description: This project explores Clustering techniques such as K-Means, K-Medoids, Gaussian Mixture Model, Hierarchical Clustering, Agglomerative ClusteringDBSCAN to segment data and analyze economic patterns.
Skills: NumPy Pandas Matplotlib Seaborn Scikit-learn Exploratory Data Analysis (EDA) Data Scaling K-Means Clustering K-Medoids Gaussian Mixture Model Hierarchical Clustering Dendograms Agglomerative Clustering DBSCAN
Project URL: Click Here

Project Description

Project: MIT: Graph-based Clustering
Date: 2024
Description: This project explores the concepts of eigenvectors and spectral clustering, focusing on their applications in data science. The project involves analyzing Graph-based Clustering techniques using Spectral methods to identify clusters in complex datasets.
Skills: NumPy Pandas Matplotlib Scikit-learn Spectral Clustering Eigenvalues & Eigenvectors
Project URL: Click Here

Project Description

Project: MIT: Spectral Clustering
Date: 2024
Description: This project explores the use of eigenvectors for dimensionality reduction and visualization techniques. It applies spectral methods to analyze datasets and extract meaningful patterns, providing insights into high-dimensional data.
Skills: NumPy Pandas Matplotlib Seaborn Eigenvectors Spectral Clustering
Project URL: Click Here

Project Description

Project: MIT: Text Processing
Date: 2024
Description: This project involves web scraping and topic modeling using Latent Dirichlet Allocation (LDA) to analyze human-generated text. It aims to extract meaningful patterns and summarize large text corpora through various Machine Learning (ML) techniques.
Skills: Scikit-learn Web Scraping BeautifulSoup Latent Dirichlet Allocation (LDA) Topic Modeling SVILDA algorithm
Project URL: Click Here

Project Description

Project: MIT: Fantasy Sports Clustering
Date: 2024
Description: This project explores clustering techniques for segmenting fantasy sports data, analyzing player trends, and identifying hidden patterns to enhance fantasy league strategies using techniques like Data Scaling, Principal Component Analysis (PCA), K-Means Clustering, K-Mediods Clustering, Hierarchical Clustering, Dendograms, Agglomerative Clustering, Gaussian Mixture Model (GMM) Clustering, and Density-Based Spatial Clustering of Applications with Noise (DBSCAN).
Skills: Pandas NumPy Matplotlib Seaborn Scikit-Learn Exploratory Data Analysis (EDA) Data Visualization Outlier Analysis Data Scaling Principal Component Analysis (PCA) K-Means Clustering K-Mediods Clustering Hierarchical Clustering Dendograms Agglomerative Clustering Gaussian Mixture Model (GMM) Clustering Density-Based Spatial Clustering of Applications with Noise (DBSCAN) Algorithm Comparison
Project URL: Click Here

Project Description

Project: MIT: Celestial Object Detection
Date: 2024
Description: This project involved creating a classification model for classifying astronomical objects into three categories: stars, galaxies, and quasars. The project required Exploratory Data Analysis (EDA), k-Nearest Neighbors (k-NN), Data Scaling, Principal Component Analysis (PCA), Decision Trees, model evaluation using k-Fold Cross Validation,to analyze the dataset, build and test the model.
Skills: NumPy Pandas Matplotlib Seaborn CSV JSON OS Scikit-learn Exploratory Data Analysis (EDA) Data Visualization Data Preparation k-Nearest Neighbors (k-NN) Data Scaling Principal Component Analysis (PCA) Decision Tree Classifier K-Fold Cross Validation
Project URL: Click Here

Project Description

Project: MIT: Facial Recognition
Date: 2024
Description: In this project, we begin building a simple application for classifying faces in photo images using Principle Component Analysis (PCA).
Skills: OS NumPy MatPlotLib Principle Component Analysis (PCA) Eigenfaces
Project URL: Click Here

Project Description

Project: MIT: Handwritten Character Recognition
Date: 2024
Description: This project employs t-Distributed Stochastic Neighbor Embedding (t-SNE) to visualize high-dimensional handwritten digit data from the MNIST dataset. The goal is to reduce the dimensionality of the dataset using Principal Component Analysis (PCA) while preserving its structure to enhance visualization and understanding of digit clustering in image recognition tasks.
Skills: MNIST Dataset Time Numpy Matplotlib Seaborn Scikit-learn t-distributed Stochastic Neighbor Embedding (t-SNE) Principal Component Analysis (PCA) Data Visualization
Project URL: Click Here

Project Description

Project: MIT: Advertising Analysis
Date: 2024
Description: This project explores the impact of advertising expenditures on sales performance. It applies multiple Regression Modeling techniques to analyze data from multiple advertising channels, identifying key influencers on sales outcomes.
Skills: Pandas NumPy Matplotlib Scikit-learn Data Visualization Simple Linear Regression Multiple Linear Regression LVC 2 - Model Evaluation: Cross validation and Bootstrapping Train/Test/Validation Split Data Regularization K-Fold and LooCV Statsmodel Bootstrapping
Project URL: Click Here

Project Description

Project: MIT: Healthcare Analytics
Date: 2024
Description: This project focuses on predicting hospital length of stay (LOS) using various machine learning models. The study involves Data Preprocessing, Feature Engineering, Exploratory Data Analysis, and regression model development including Linear Regression, Ridge Regression, Lasso Regression, Decision Trees Regressors, Random Forests Regressors, and Bagging Regressors to estimate patient LOS based on various demographic and clinical factors.
Skills: Pandas NumPy Matplotlib Seaborn Scikit-learn Exploratory Data Analysis (EDA) Data Processing Feature Engineering Train/Test Split cross_val_score KFold Linear Regression Ridge Regression Lasso Regression Decision Tree Regressor Random Forest Regressor Bagging Regressor LabelEncoder GridSearchCV Statistical Analysis Prediction
Project URL:

Project Description

Project: MIT: Retail Analytics - BigMart
Date: 2024
Description: This project involves predicting sales for different BigMart stores using a variety of Machine Learning models. The analysis includes Data Preprocessing, Feature Engineering, in-depth Statistical Analysisand the development of Predictive Models to estimate sales based on various factors such as store size, location, and product attributes.
Skills: NumPy Pandas Seaborn Matplotlib StatsModels Scikit-learn Exploratory Data Analysis (EDA) Data Processing Missing Values Treatment Data Visualization Feature Engineering Data Scaling (w/MinMaxScaler) Feature Selection (removing Multicollinearity w/Variance Inflation factor) Train/Test Split Statistical Analysis of Data Testing for Heteroscedasticity/Homoscedasticity
Project URL: Click Here

Project Description

Project:MIT: Retail Analytics - Cars4U
Date: 2024
Description: This project analyzes car sales data, evaluating trends and customer preferences using data-driven regression techniques (Linear Regression, Ridge Regression, Decision Tree Regression, and Random Forest Regression) and Hyperparameter Tuning to optimize sales strategies.
Skills: Pandas NumPy Matplotlib Seaborn Scikit-learn Exploratory Data Analysis (EDA) Data Preprocessing Data Visualization Feature Engineering Missing Value Treatment Train/Test Split Linear Regression Ridge Regression Decision Tree Regressor Random Forest Regressor Hyperparameter Tuning - Decision Tree Decision Tree Regressor Hyperparameter Tuning - Random Forest Random Forest Regressor Statistical Analysis
Project URL: Click Here

Project Description

Project: Wage Prediction
Date: 2024
Description: This project analyzes wage patterns and attempts to predict wages based on demographic and work-related factors using Linear Regression and Polynomial Regressionon techniques.
Skills: NumPy Pandas Matplotlib Seaborn Exploratory Data Analysis (EDA) Data Visualization Train/Test Split Linear Regression Polynomial Features
Project URL: Click Here

Project Description

Project: MIT: Gender Wage Gap
Date: 2024
Description: This project analyzes the gender wage gap in the U.S., exploring factors such as education level, industry, and work experience. Using statistical methods and machine learning techniques (Linear Regression, Polynomial Regression, Linear Specification , Quadratic Specification), it aims to quantify wage disparities and identify key influencing variables using Statistical Hypothesis Testing.
Skills: NumPy Pandas Matplotlib Seaborn Exploratory Data Analysis (EDA) Linear Regression Polynomial Features Partialling Out: (Linear Specification - Basic Model) Partialling Out: (Quadratic Specification - Flexible model) Statistical Hypothesis Testing
Project URL: Click Here

Project Description

Project: MIT: Economic Growth Analysis
Date: 2024
Description: This project examines economic growth disparities between rich and poor countries using GDP growth data. It involves Visualization, Linear Regression, Lasso Regressor and Statistical Modeling to understand macroeconomic trends and contributing factors.
Skills: NumPy Pandas Scikit-learn StatsModels Data Visualization Linear Regression Lasso Regressor Statistical Modeling
Project URL: Click Here

Project Description

Project: MIT: Wage Prediction (Using Modern Techniques)
Date: 2024
Description: This project analyzes wage prediction using various Statistical and Machine Learning techniques, examining key factors such as education, experience, and gender.
Skills: NumPy Pandas Matplotlib Seaborn Scikit-learn Exploratory Data Analysis (EDA) Data Visualization Polynomial Features Linear Regression ElasticNet CV Gradient Boosting Regressor Decision Tree Regressor Statistical Analysis
Project URL: Click Here

Project Description

Project: MIT: Retail Analytics - SuperKart
Date: 2024
Description: This project analyzes sales data from SuperKart to predict future sales trends. It employs in-depth Statistical Analysis, Data Preprocessing, Feature Engineering, and OLS Regression to extract insights that can improve sales strategies.
Skills: NumPy Pandas Scikit-learn Matplotlib Seaborn StatsModel SciPy PyLab Statistical Analysis Exploratory Data Analysis (EDA) Data Visualization Data Preprocessing Feature Engineering Outlier Checks Train/Test Split Model Training Multicollinearity Analysis OLS Regression Prediction
Project URL: Click Here

Project Description ^★

Project: MIT: CAPSTONE PROJECT 1: Customer Prediction Models
Date: 2024
Description: This project aims to build a machine learning model to predict the likelihood of leads converting to paid customers for ExtraaLearn, an EdTech startup. The goal is to analyze lead interactions, identify key conversion factors, and develop a predictive model to optimize marketing efforts and resource allocation.

The project required extensive:
- Data Cleansing
- Exploratory Data Analysis (EDA)
- Data Visualization, Preprocessing, and Feature Engineering
- Outlier Detection and Treatment
- Data Scaling
The following machine learning techniques were then used:
- Logistic Regression
- Principal Component Analysis (PCA)
- Decision Tree Classifier
- Decision Tree Pruning using:
  - Maximum Depth
  - Limited Depth
  - Minimum Samples per Leaf
  - Minimum Samples to Split)
- Decision Tree Classifier Hyperparameter Tuning
- Random Forest Classifier with Pruning and Hyperparameter Tuning
Due to a large data imbalance the following techniques were used:
- Class Weighting
- SMOTE
- Undersampling
- Hybrid - SMOTE Oversampling and Undersampling
Advanced regression techniques were then tested including:
- Random Forest Classifier with Polynomial Features
- XGBoost with Class Weights and Polynomial Features
- XGBoost with Threshold Tuning and Polynomial Features
- Majority Class Undersampling with Polynomial Features
- Optimized Hyperparameters for Recall
All models were then compared and the best used to analyze the dataset, and build the prediction model.
Skills: NumPy Pandas Matplotlib Seaborn SciPy Scikit-learn StatsModels XGBoost IPython Data Cleansing Exploratory Data Analysis (EDA) Data Visualization Data Preprocessing Feature Engineering One Hot Encoding Outlier Detection and Treatment Target Feature Separation Data Split Data Scaling Logistic Regression Principal Component Analysis (PCA) Decision Tree Classifier Decision Tree Pruning (Maximum Depth) Decision Tree Pruning (Limited Depth) Decision Tree Pruning (Minimum Samples per Leaf) Decision Tree Pruning (Minimum Samples to Split) Decision Tree Classifier Hyperparameter Tuning Random Forest Classifier Random Forest Pruning Random Forest Hyperparameter Tuning Data Imbalance (Class Weighting) Data Imbalance (SMOTE) Data Imbalance (Undersampling) Data Imbalance (Hybrid - SMOTE Oversampling and Undersampling) Advanced Feature Engineering Random Forest Classifier with Polynomial Features Gradient Boosting with Polynomial Features XGBoost with Class Weights and Polynomial Features XGBoost with Threshold Tuning and Polynomial Features Majority Class Undersampling with Polynomial Features SMOTE and Undersampling with Polynomial Features Optimized Hyperparameters for Recall Customer Profile Statistical Analysis
Project URL: Click Here

Project Description

Project: MIT: Celestial Object Detection
Date: 2024
Description: This project involved creating a classification model for classifying astronomical objects into three categories: stars, galaxies, and quasars. The project required Exploratory Data Analysis (EDA), k-Nearest Neighbors (k-NN), Data Scaling, Principal Component Analysis (PCA), Decision Trees, model evaluation using k-Fold Cross Validation, to analyze the dataset, build and test the model.
Skills: NumPy Pandas Matplotlib Seaborn CSV JSON OS Scikit-learn Exploratory Data Analysis (EDA) Data Visualization Data Preparation k-Nearest Neighbors (k-NN) Data Scaling Principal Component Analysis (PCA) Decision Tree Classifier K-Fold Cross Validation
Project URL: Click Here

Project Description

Project: MIT: Gender Wage Gap
Date: 2024
Description: This project analyzes the gender wage gap in the U.S., exploring factors such as education level, industry, and work experience. Using statistical methods and machine learning techniques (OLS Regression, Polynomial Regression, Linear Specification , Quadratic Specification), it aims to quantify wage disparities and identify key influencing variables using Statistical Hypothesis Testing.
Skills: NumPy Pandas Matplotlib Seaborn Exploratory Data Analysis (EDA) OLS Regression Polynomial Features Partialling Out: (Linear Specification - Basic Model) Partialling Out: (Quadratic Specification - Flexible model) Statistical Hypothesis Testing
Project URL: Click Here

Project Details

Project: MIT: Hospital Length of Stay
Date: 2024
Description: This project explores advanced classification techniques and hypothesis testing in the context of data-driven decision-making. It involves extensive exploratory data analysis (EDA), feature engineering, and statistical hypothesis testing to uncover significant insights.
The project also includes the implementation and evaluation of multiple regression and classification models, with a strong focus on ensemble learning techniques such as Random Forest, Gradient Boosting, and XGBoost. By leveraging statistical methods and machine learning algorithms, this project refines predictive models and ensures data-driven optimization for real-world applications.
Skills: NumPy Pandas Matplotlib Seaborn Exploratory Data Analysis (EDA) Data Visualization Statistical Analysis Data Cleanup Feature Engineering Data Preparation Train/Test Split Decision Tree Regressor Ensemble Learning Bagging Regressor (Bootstrap Aggregating) Random Forest Regressor Ada Boost Regressor Gradient Boosting Regressor XGB Regressor Tuned Random Forest Regressor Hypothesis Testing Classification Models
Project URL: Click Here

Project Details

Project: MIT: Credit Card Churn Analysis
Date: 2024
Description: This project analyzes credit card customer churn, using data-driven approaches to identify patterns and predict customer attrition. The analysis includes Exploratory Data Analysis (EDA), Feature Engineering, and Machine Learning techniques (including Missing Value Imputation, Categorical Encoding, Logistic Regression, Support Vector Machine (SVM), Decision Tree Classifier, Random Forest Classifier) to build a predictive model.
Skills: Pandas NumPy Matplotlib Seaborn Scikit-learn Data Preparation Exploratory Data Analysis (EDA) Data Visualization Missing Value Imputation Train/Test Split Categorical Encoding Logistic Regression Support Vector Machine (SVM) Precision-Recall Curve for Logistic Regression RBF Kernel Decision Tree Classifier Random Forest Classifier
Project URL: Click Here

Project Details

Project: MIT: Method Performance Prediction
Date: 2024
Description: This project predict the quality of a roasting method's outputs, using data-driven approaches to identify patterns and predict customer attrition. The analysis includes Exploratory Data Analysis (EDA), and Machine Learning techniques (including Missing Value Imputation, Categorical Encoding, Logistic Regression, Support Vector Machine (SVM), Decision Tree Regressor, Random Forest Regressor) to build and evaluate predictive models, performs an evaluation of those models and creates a pipeline for its use.
Skills: Pandas NumPy Matplotlib Seaborn Scikit-learn Exploratory Data Analysis (EDA) Data Visualization Data Preparation Train/Test Split Missing Value Imputation Decision Tree Regressor Random Forest Regressor Bagging Regressor Adaboost Regressor Gradient Boosting Regressor XGB Regressor Model Performance Evaluation Statistical Analysis Hyperparameter Tuning Tuned Random Forest Regressor Tuned Bagging Regressor Tuned Decision Tree Regressor Model pipeline Construction
Project URL: Click Here

Project Details

Project: MIT: Employee Attrition Prediction
Date: 2024
Description: This project is to determine the factors the affect employee attrition and predict which employees are likely to leave the company. The analysis includes Exploratory Data Analysis (EDA), Data Preparation and Scaling Machine Learning techniques including Logistic Regression, Support Vector Machine (SVM) using both Linear Kernel and RBF Kernel, Decision Tree Classifier, Random Forest Classifier, Hyperparameter Tuning using both Grid Search and Random Search, evaluate models of those models and Grid Search Tuned Decision Tree Classifier, and Grid Search Tuned Decision Random Forest Classifier.
Skills: Pandas NumPy Matplotlib Seaborn Scikit-learn Exploratory Data Analysis (EDA) Data Visualization Data Preparation Train/Test Split Data Scaling Logistic Regression Model Statistical Analysis Precision-Recall Curve for Logistic Regression Support Vector Machines - Linear Kernel Support Vector Machines - RBF Kernel Decision Tree Classifier Random Forest Classifier Hyperparameter Tuning - Grid Search Hyperparameter Tuning - Random Search Grid Search Tuned Decision Tree Classifier Grid Search Tuned Decision Random Forest Classifier
Project URL: Click Here

Project Details

Project: MIT: Travel Package Purchase Prediction
Date: 2024
Description: This project is to analyze the customers' data and information to provide recommendations to the marketing department and build a model to predict the potential customer who is going to purchase the newly introduced travel packagey. The analysis includes Exploratory Data Analysis (EDA), Data Preparation and Scaling Machine Learning techniques (including Logistic Regression, Support Vector Machine (SVM) using both Linear Kernel and RBF Kernel, Decision Tree Classifier, Random Forest Classifier), Hyperparameter Tuning using both Grid Search and Random Search, performs an evaluation of those models and Grid Search Tuned Decision Tree Classifier, and Grid Search Tuned Decision Random Forest Classifier..
Skills: Pandas NumPy Matplotlib Seaborn Scikit-learn Exploratory Data Analysis (EDA) Data Visualization Data Preparation Impute Missing Values Train/Test Split Data Scaling Logistic Regression Model Support Vector Machines - Linear Kernel Support Vector Machines - RBF Kernel Decision Tree Classifier Hyperparameter Tuning - Grid Search Random Forest Classifier
Project URL: Click Here
Published Model Project URL: Click Here

Project Details

Client: MIT: Restaurant Waiting Time
Date: 2024
Description: This project applies a Decision Tree Classifier to restaurant data to analyze trends, and identify the key factors in whether a customer will wait for service or leave, to improve data-driven decision-making.
Skills: NumPy Matplotlib Scikit-learn Decision Tree Classifier
Project URL: Click Here

Project Details

Client: MIT: Space Shuttle Challenger Disaster Analysis
Date: 2024
Description: The Space Shuttle Challenger exploded 73 seconds after liftoff on January 28, 1986, killing all seven astronauts on board, including school teacher Christa McAuliffe. The disaster resulted from the failure of an O-ring seal in one of the solid rocket boosters, which allowed hot gases to escape and led to the structural failure of the shuttle. Engineers from Morton Thiokol, the company responsible for the boosters, were aware of the risks associated with launching in cold temperatures. They urged NASA to delay the launch, warning that the O-rings might fail. However, their concerns were ultimately overruled under pressure to maintain the launch schedule, leading to the catastrophe.

In this analysis, we will use Logistic Regression to predict O-ring failure as temperatures decrease.
Skills: NumPy Pandas StatsModels
Project URL: Click Here

Project Description ^★

Project: MIT: CAPSTONE PROJECT 1: Cutomer Prediction Models
Date: 2024
Description: This project aims to build a machine learning model to predict the likelihood of leads converting to paid customers for ExtraaLearn, an EdTech startup. The goal is to analyze lead interactions, identify key conversion factors, and develop a predictive model to optimize marketing efforts and resource allocation.

The project required extensive:
- Data Cleansing
- Exploratory Data Analysis (EDA)
- Data Visualization, Preprocessing, and Feature Engineering
- Outlier Detection and Treatment
- Data Scaling
The following machine learning techniques were then used:
- Logistic Regression
- Principal Component Analysis (PCA)
- Decision Tree Classifier
- Decision Tree Pruning using:
  - Maximum Depth
  - Limited Depth
  - Minimum Samples per Leaf
  - Minimum Samples to Split)
- Decision Tree Classifier Hyperparameter Tuning
- Random Forest Classifier with Pruning and Hyperparameter Tuning
Due to a large data imbalance the following techniques were used:
- Class Weighting
- SMOTE
- Undersampling
- Hybrid - SMOTE Oversampling and Undersampling
Advanced regression techniques were then tested including:
- Random Forest Classifier with Polynomial Features
- XGBoost with Class Weights and Polynomial Features
- XGBoost with Threshold Tuning and Polynomial Features
- Majority Class Undersampling with Polynomial Features
- Optimized Hyperparameters for Recall
All models were then compared and the best used to analyze the dataset, and build the prediction model.
Skills: NumPy Pandas Matplotlib Seaborn SciPy Scikit-learn StatsModels XGBoost IPython Data Cleansing Exploratory Data Analysis (EDA) Data Visualization Data Preprocessing Feature Engineering One Hot Encoding Outlier Detection and Treatment Target Feature Separation Data Split Data Scaling Logistic Regression Principal Component Analysis (PCA) Decision Tree Classifier Decision Tree Pruning (Maximum Depth) Decision Tree Pruning (Limited Depth) Decision Tree Pruning (Minimum Samples per Leaf) Decision Tree Pruning (Minimum Samples to Split) Decision Tree Classifier Hyperparameter Tuning Random Forest Classifier Random Forest Pruning Random Forest Hyperparameter Tuning Data Imbalance (Class Weighting) Data Imbalance (SMOTE) Data Imbalance (Undersampling) Data Imbalance (Hybrid - SMOTE Oversampling and Undersampling) Advanced Feature Engineering Random Forest Classifier with Polynomial Features Gradient Boosting with Polynomial Features XGBoost with Class Weights and Polynomial Features XGBoost with Threshold Tuning and Polynomial Features Majority Class Undersampling with Polynomial Features SMOTE and Undersampling with Polynomial Features Optimized Hyperparameters for Recall Customer Profile Statistical Analysis
Project URL: Click Here

Project Description

Project: MIT: AudioMNIST Digit Recognition
Date: 2024
Description: This project utilizes machine learning and deep learning to recognize spoken digits from the AudioMNIST dataset . The primary goal is to extract meaningful features from audio signals and train a robust classification model.

Key techniques include:
- MFCC Spectrogram Feature Extraction
- Spectrogram Analysis
- Data Preprocessing (Noise Reduction, Augmentation)
- Deep Learning using CNNs & RNNs
- Hyperparameter Optimization
- Evaluation Metrics (Accuracy, Confusion Matrix, F1-Score)
Skills & Frameworks: Python NumPy Pandas Matplotlib Librosa TensorFlow Keras Scikit-learn MFCC Spectrograms Spectrogram Analysis CNNs RNNs Model Evaluation Hyperparameter Tuning
Project URL: Click Here

Project Description

Project: MIT: Brain Tumor Image Classifier
Date: 2024
Description: This project employs deep learning techniques to classify brain tumors from medical imaging datasets. The primary goal is to use transfer learning to enhance classification accuracy.

Key techniques include:
- Image Preprocessing (Resizing, Normalization, Augmentation)
- Transfer Learning with Pretrained Models
- Feature Extraction using CNN Architectures
- Model Fine-Tuning
- Evaluation Metrics (Accuracy, Precision, Recall, F1-Score)
Skills & Frameworks: Python NumPy Pandas Matplotlib OpenCV TensorFlow Keras Scikit-learn CNNs Transfer Learning VGG16 ResNet Inception Model Evaluation Hyperparameter Tuning
Project URL: Click Here

Project Description

Project: MIT: Data Scientist Employee Attrition
Date: 2024
Description: This project employs machine learning techniques to predict employee attrition in a data science workforce. The goal is to identify key drivers of attrition and build predictive models for proactive employee retention.

Key techniques include:
- Exploratory Data Analysis (EDA)
- Data Preprocessing (Missing Value Handling, Encoding, Scaling)
- Feature Engineering & Selection
- Machine Learning Models:
  - Logistic Regression
  - Decision Trees
  - Random Forest Classifier
- Performance Metrics (Accuracy, Precision, Recall, F1-Score)
Skills & Frameworks: Python NumPy Pandas Matplotlib Seaborn Scikit-learn StatsModels Data Preprocessing Feature Engineering Logistic Regression Decision Tree Classifier Random Forest Classifier Model Evaluation
Project URL: Click Here

Project Description

Project: MIT: Food Image Classification
Date: 2024
Description: This project applies deep learning techniques for food image classification. The primary objective is to leverage transfer learning and CNN models for accurate recognition.

Key techniques include:
- Image Preprocessing (Resizing, Normalization, Augmentation)
- Feature Extraction using CNNs
- Transfer Learning with Pretrained Models
- Model Fine-Tuning
- Evaluation Metrics (Accuracy, Precision, Recall, F1-Score)
Skills & Frameworks: Python NumPy Pandas Matplotlib OpenCV TensorFlow Keras Scikit-learn CNNs Transfer Learning VGG16 ResNet Inception Model Evaluation Hyperparameter Tuning
Project URL: Click Here

Project Description

Project: MIT: Hands-On Artificial Neural Networks (ANNs)
Date: 2024
Description: This project focuses on building and training Artificial Neural Networks (ANNs) using deep learning frameworks. The objective is to develop multi-layer perceptrons (MLPs) for various predictive modeling tasks.

Key techniques include:
- Data Preprocessing (Normalization, Feature Scaling)
- Designing Multi-Layer Perceptrons (MLPs)
- Activation Functions: ReLU, Sigmoid, Softmax
- Optimization Algorithms: Gradient Descent, Adam, RMSprop
- Loss Functions: MSE, Cross-Entropy Loss
- Model Evaluation (Accuracy, Loss Tracking, Confusion Matrix)
Skills & Frameworks: Python NumPy Pandas Matplotlib TensorFlow Keras Scikit-learn Artificial Neural Networks (ANNs) Multi-Layer Perceptron (MLP) ReLU Sigmoid Softmax Gradient Descent Adam Optimizer RMSprop Model Evaluation
Project URL: Click Here

Project Description

Project: MIT: Fashion MNIST Using CNN
Date: 2024
Description: This project implements CNNs to classify images from the Fashion MNIST dataset . The goal is to recognize different clothing items based on grayscale image data .

Key techniques include:
- Data Preprocessing (Normalization, Reshaping Images)
- Building CNN Architectures for Feature Extraction
- Using Batch Normalization & Dropout for Regularization
- Optimization with Adam & Cross-Entropy Loss
- Model Evaluation (Accuracy, Loss Analysis, Confusion Matrices)
Skills & Frameworks: Python NumPy Pandas Matplotlib TensorFlow Keras Scikit-learn Convolutional Neural Networks (CNNs) Batch Normalization Dropout Regularization Adam Optimizer Model Evaluation
Project URL: Click Here

Project Description

Project: MIT: Predicting Chances of Admission
Date: 2024
Description: This project predicts the likelihood of a student being admitted to a graduate program using machine learning regression models . The goal is to analyze key academic and personal factors influencing admissions.

Key techniques include:
- Exploratory Data Analysis (EDA) and Correlation Analysis
- Feature Engineering & Preprocessing (Handling Missing Values, Scaling, Encoding)
- Machine Learning Models:
  - Linear Regression
  - Decision Tree Regression
  - Random Forest Regression
- Model Evaluation (R² Score, RMSE, MAE)
Skills & Frameworks: Python NumPy Pandas Matplotlib Seaborn Scikit-learn Linear Regression Decision Tree Regression Random Forest Regression Model Evaluation
Project URL: Click Here

Project Description

Project: MIT: CNN Sequential Model
Date: 2024
Description: This project implements Convolutional Neural Networks (CNNs) using the Sequential API in TensorFlow/Keras. The goal is to build a robust deep learning model for image classification .

Key techniques include:
- Building CNN Architectures with Sequential API
- Feature Extraction using Convolutional Layers (Conv2D, MaxPooling2D)
- Regularization with Dropout Layers
- Optimization using Adam & Cross-Entropy Loss
- Model Evaluation (Accuracy, Loss Tracking, Confusion Matrix)
This project is a foundational implementation for image recognition, applicable in fields like medical imaging, security, and automation.
Skills & Frameworks: Python NumPy Pandas Matplotlib TensorFlow Keras Scikit-learn Convolutional Neural Networks (CNNs) Sequential API Conv2D MaxPooling2D Dropout Regularization Adam Optimizer Model Evaluation
Project URL: Click Here

Project Description

Project: MIT: Plant Seedlings Classification CNN
Date: 2024
Description: This project implements a CNN-based image classification model to identify different plant seedling species . The goal is to develop a robust deep learning pipeline for automated plant classification .

Key techniques include:
- Image Preprocessing (Normalization, Resizing, Augmentation)
- Building a CNN Architecture for Feature Extraction
- Transfer Learning for Improved Accuracy
- Optimization using Adam & Cross-Entropy Loss
- Model Evaluation (Accuracy, Loss Analysis, Confusion Matrices)
This project is useful for agriculture, environmental monitoring, and automated plant species identification.
Skills & Frameworks: Python NumPy Pandas Matplotlib OpenCV TensorFlow Keras Scikit-learn Convolutional Neural Networks (CNNs) Transfer Learning Data Augmentation Adam Optimizer Model Evaluation
Project URL: Click Here

Project Description

Project: MIT: CNN Practice Project (SVHN)
Date: 2024
Description: This project applies CNNs to classify images of house numbers from the SVHN dataset . The goal is to develop an automated digit recognition system for real-world applications.

Key techniques include:
- Data Preprocessing (Normalization, Resizing, Augmentation)
- *Building a CNN Architecture for Feature Extraction
- Optimization using Adam & Cross-Entropy Loss
- Model Evaluation (Accuracy, Loss Analysis, Confusion Matrices)
This project showcases deep learning techniques for real-world digit classification , making it relevant for automated document processing , license plate recognition , and postal address identification .
Skills & Frameworks: Python NumPy Pandas Matplotlib TensorFlow Keras OpenCV Scikit-learn Convolutional Neural Networks (CNNs) Data Augmentation Adam Optimizer Model Evaluation
Project URL: Click Here

Project Description

Project: MIT: NN Practice Project (SVHN)
Date: 2024
Description: This project applies Neural Networks (NNs) to classify house numbers from the SVHN dataset . Unlike CNN-based approaches, it focuses on fully connected architectures .

Key techniques include:
- Image Preprocessing (Normalization, Resizing, Augmentation)
- Building Fully Connected Neural Networks (FNNs)
- Optimization using Adam & Cross-Entropy Loss
- Model Evaluation (Accuracy, Loss Analysis, Confusion Matrices)
This project extends previous CNN-based implementations by experimenting with different NN architectures , making it relevant for automated digit recognition , OCR , and smart address processing .
Skills & Frameworks: Python NumPy Pandas Matplotlib TensorFlow Keras OpenCV Scikit-learn Fully Connected Neural Networks (FNNs) Data Augmentation Adam Optimizer Model Evaluation
Project URL: Click Here

Project Description

Project: MIT: Book Recommendation Systems
Date: 2024
Description: This project develops a book recommendation system using machine learning and collaborative filtering . The goal is to generate personalized book suggestions by analyzing user preferences and item metadata .

Key techniques include:
- Data Preprocessing (Normalization, Encoding, Feature Engineering)
- User-Based & Item-Based Collaborative Filtering
- Matrix Factorization (SVD, ALS, SVD++)
- Hybrid Recommendation Models *
- Evaluation Metrics (RMSE, Precision-Recall, Ranking Metrics)
This project is applicable to e-learning platforms , bookstores , and digital libraries , enabling a highly personalized book discovery experience.
Skills & Frameworks: Python NumPy Pandas Matplotlib Seaborn Scikit-Learn Surprise (Recommendation Algorithms) NLTK (Text Processing) TensorFlow/Keras (Deep Learning Models)
Project URL: Click Here

Project Description

Project: MIT: MLS Yelp Reviews
Date: 2024
Description: This project performs sentiment analysis on Yelp reviews using Natural Language Processing (NLP) . The goal is to classify reviews into positive, negative, or neutral sentiment using machine learning models.

Key techniques include:
- Text Preprocessing (Tokenization, Stopword Removal, Lemmatization)
- Feature Engineering (TF-IDF, Word Embeddings, Word2Vec)
- Supervised Learning Models (Naïve Bayes, Logistic Regression, Neural Networks)
- Model Evaluation (Confusion Matrices, Precision-Recall, F1 Score)
- Data Visualization (Word Clouds, Sentiment Distribution Graphs)
This project is valuable for businesses, marketers, and customer experience teams looking to automate sentiment analysis and gain insights from customer reviews .
Skills & Frameworks: Python NumPy Pandas Matplotlib Seaborn Scikit-learn Natural Language Toolkit (NLTK) TF-IDF Vectorization Word2Vec Logistic Regression Naïve Bayes Deep Learning (Neural Networks) Model Evaluation (Confusion Matrix, F1 Score) Data Visualization
Project URL: Click Here

Project Description

Project: MIT: Million Songs Recommendation System
Date: 2024
Description: This project builds a music recommendation system based on the Million Song Dataset . The goal is to suggest songs to users based on their listening habits using collaborative filtering and content-based techniques .

Key techniques include:
- Data Cleaning & Preprocessing
- TF-IDF Feature Extraction for Content-Based Filtering
- User-Based & Item-Based Collaborative Filtering
- Co-Clustering for Grouping Similar Users & Songs
- Precision, Recall, and RMSE for Model Evaluation
This project showcases recommendation system techniques that can be extended to streaming services like Spotify and Apple Music .
Skills & Frameworks: Python NumPy Pandas Matplotlib Seaborn Scikit-learn Surprise Library TF-IDF Collaborative Filtering User-Based Recommendations Item-Based Recommendations Co-Clustering Cosine Similarity RMSE Evaluation Precision & Recall Metrics
Project URL: Click Here

Project Description ^★

Project: MIT: CAPSTONE PROJECT 2: Amazon Product Recommendation System
Date: 2024
Description: In today\’s digital marketplace, personalized recommendations play a crucial role in enhancing user experience and increasing sales. The project utilizes Amazon's product review dataset, which includes user ratings for electronic products.

The primary objective of this project is to build a scalable recommendation system using different machine learning techniques to enhance personalized suggestions. The dataset lacks product descriptions and review texts to avoid bias, ensuring that the recommendations are purely based on user interaction patterns.

This project applies collaborative filtering , content-based filtering , and machine learning techniques to build an intelligent **Amazon product recommendation system**.

Key techniques include:
- Data Preprocessing (Handling Missing Data, Normalization, Filtering Sparse Data)
- Exploratory Data Analysis (EDA) to Understand User Behavior, product interactions, and rating distributions
- Building Recommendation Models :
  - Rank-Based Recommendations (Top-Rated Product Ranking)
  - User-User Collaborative Filtering
  - Item-Item Collaborative Filtering
  - Matrix Factorization (SVD) for Latent Factor Analysis
  - Clustering-Based Recommendation Systems (K-Means, Hierarchical Clustering)
  - Content-Based Filtering using TF-IDF
  - Hybrid Recommendation Systems (Combining Multiple Models)
  - Grid Search Cross-Validation for Hyperparameter Tuning
- Model Evaluation using RMSE, Precision@K, Recall@K, and F1-score
- Hyperparameter Tuning: Using grid search cross-validation to optimize recommendation models
This project highlights how machine learning-based recommendation systems improve customer engagement, optimize inventory management, and drive sales in e-commerce platforms.
Skills & Frameworks: Python NumPy Pandas Matplotlib Seaborn SciPy Scikit-learn Surprise Library Matrix Factorization (SVD) Clustering (K-Means, Hierarchical Clustering) TF-IDF Feature Engineering Grid Search Cross-Validation Precision@K, Recall@K, RMSE Evaluation
Project URL: Click Here

Project Description

Project: MIT: Enron Network Analysis
Date: 2024
Description: The Enron scandal was one of the biggest corporate frauds in history, leading to the company's bankruptcy. This case study investigates the email network of 50 senior officials to identify key players, their relationships, and their influence in the organization.
Key tasks in this project include:
This project demonstrates how network analysis can uncover patterns in organizational communication, which can be used for fraud detection, leadership analysis, and security insights.
Skills & Frameworks: Pandas NetworkX Matplotlib Graph Theory Social Network Analysis Fraud Detection Eigenvector Centrality Betweenness Centrality Closeness Centrality
Project URL: Click Here

Project Description

Project: MIT: Financial Market Price Estimation
Date: 2024
Description: This project applies machine learning techniques to predict financial market prices based on historical data and market indicators.

Key techniques include:
- Data Preprocessing : Collecting and cleaning financial time-series data.
- Feature Engineering : Extracting meaningful features from historical prices and market indicators.
- Building Predictive Models : Implementing models such as Linear Regression, Random Forest, and Neural Networks.
- Model Training & Optimization : Tuning hyperparameters to improve predictive accuracy.
- Performance Evaluation : Assessing models using RMSE, MAE, and R-squared metrics.

This project demonstrates how machine learning can be leveraged for financial market forecasting, providing valuable insights for traders and investors.

Skills & Frameworks: Python NumPy Pandas Matplotlib Scikit-learn TensorFlow Keras Time-Series Forecasting Regression Analysis Model Evaluation Metrics
Project URL: Click Here

Project Description

Project: MIT: Game of Thrones Network Analysis
Date: 2024
Description: This project uses network analysis to explore character relationships in the Game of Thrones series. The goal is to identify key players and community structures within the character network.

Key techniques include:
- Building a Network Graph : Creating a graph where nodes represent characters and edges represent interactions.
- Centrality Analysis : Identifying key players using degree centrality, betweenness centrality, and eigenvector centrality.
- Community Detection : Using clustering algorithms to identify groups and alliances.
- Graph Visualization : Using NetworkX and Matplotlib to represent the network structure.
- Sentiment Analysis : Analyzing the sentiment of interactions to classify relationships as positive or negative.

This project demonstrates how graph theory and network analysis can be used to uncover hidden patterns in complex narratives.

Skills & Frameworks: Python NetworkX Matplotlib Pandas Gephi (Optional) Graph Theory Community Detection Sentiment Analysis (Optional)
Project URL: Click Here

Project Description

Project: MIT: Kalman Filter 3D Location Tracking
Date: 2024
Description: This project applies Kalman Filtering for 3D location tracking, a technique widely used in robotics, navigation, and motion tracking applications. The Kalman Filter is an optimal estimator that updates predictions based on noisy sensor data, making it highly effective for tracking moving objects.

Key techniques include:
- Data Acquisition & Processing : Simulating or using real-world sensor data for object tracking.
- Kalman Filter Implementation : Designing a state-space model for 3D position and velocity estimation.
- Prediction & Correction Steps : Applying state transition and measurement updates to refine object location estimates.
- Noise Reduction & Smoothing : Filtering out measurement noise to enhance accuracy.
- Visualization : Plotting real vs. predicted locations for evaluation.

This project demonstrates how Kalman Filters can accurately track objects in 3D space, offering applications in robotics, self-driving cars, and augmented reality.

Skills & Frameworks: Python NumPy Pandas Matplotlib SciPy OpenCV Kalman Filter State Estimation Motion Tracking Noise Reduction Sensor Fusion
Project URL: Click Here

Project Description

Project: MIT: Kalman Filter: Tracking Location of an Object Moving with Constant Velocity
Date: 2024
Description: This project implements a Kalman Filter to track an object moving with constant velocity. The algorithm uses prediction and correction steps to estimate an object's position accurately.

Key techniques include:
- State Representation : Modeling the object's position and velocity in a dynamic system.
- Prediction Step : Estimating the next state based on motion dynamics.
- Update Step : Correcting the state estimate using sensor measurements.
- Noise Handling : Filtering out sensor noise to improve accuracy.
- Simulation & Visualization : Running experiments and plotting results to analyze filter performance.

This project demonstrates how Kalman Filters can effectively track objects in motion, making them crucial for applications in autonomous navigation, GPS tracking, and object recognition.

Skills & Frameworks: Python NumPy Pandas Matplotlib Kalman Filter State Estimation Sensor Fusion Noise Filtering Motion Tracking Dynamic Systems
Project URL: Click Here

Project Description

Project: MIT: Network Stock Portfolio Optimization
Date: 2024
Description: This project leverages graph theory and machine learning to construct an optimized stock portfolio. It employs network-based approaches to analyze stock correlations and determine optimal asset allocations that minimize risk while maximizing returns.

Key techniques include:
- Data Collection : Fetching historical stock prices and computing correlation matrices.
- Network Construction : Representing stocks as a graph based on correlation similarities.
- Portfolio Optimization : Applying Markowitz Modern Portfolio Theory (MPT) and Minimum Spanning Tree (MST) techniques to optimize asset allocation.
- Risk Analysis : Measuring diversification benefits and portfolio risk using network properties.
- Performance Evaluation : Comparing the optimized portfolio against traditional allocation strategies.

This project demonstrates how network analysis and financial optimization techniques can enhance investment decision-making and risk management.

Skills & Frameworks: Python Pandas NumPy Matplotlib NetworkX Scipy Markowitz MPT Minimum Spanning Tree (MST) Financial Risk Analysis Data Visualization
Project URL: Click Here

Project Description

Project: MIT: Unemployment Rate Prediction
Date: 2024
Description: This project applies Hidden Markov Model, a predictive modeling techniques to estimate unemployment rates using historical labor market data. It explores various machine learning algorithms to develop an effective forecasting model.

Key techniques include:
- Data Collection & Preprocessing : Gathering labor market data, handling missing values, and feature engineering.
- Exploratory Data Analysis (EDA) : Visualizing unemployment trends and correlations with economic factors.
- Model Development : Implementing Linear Regression, Decision Trees, and Random Forest models to predict unemployment rates.
- Performance Evaluation : Assessing models using RMSE, R-squared scores, and cross-validation.
- Hyperparameter Tuning : Optimizing model performance with GridSearchCV and feature selection techniques.

This project demonstrates how machine learning can help predict unemployment trends, assisting policymakers and businesses in decision-making.

Skills & Frameworks: Python NumPy Pandas Matplotlib Seaborn Scikit-learn Regression Models Hyperparameter Tuning GridSearchCV Feature Engineering Model Evaluation Metrics
Project URL: Click Here

Project Description

Project: MIT: CAVIAR Case Study
Date: 2024
Description: The CAVIAR Case Study project focuses on computer vision and behavior analysis using video surveillance data. The goal is to analyze pedestrian interactions and movement patterns in surveillance videos, leveraging deep learning and computer vision techniques.

Key techniques include:
- Data Preprocessing : Extracting video frames and annotating human activities and interactions.
- Feature Extraction : Using convolutional neural networks (CNNs) and other techniques to extract movement patterns.
- Behavior Classification : Training a model to classify actions such as walking, meeting, split-up, and fighting.
- Trajectory Analysis : Analyzing movement patterns to detect unusual behavior.
- Performance Evaluation : Using metrics like accuracy, confusion matrices, and precision-recall curves.

This project demonstrates how computer vision and deep learning can be used for behavioral analysis in video surveillance, with applications in security, urban planning, and automated monitoring.

Skills & Frameworks: Python OpenCV NumPy Pandas Matplotlib TensorFlow Keras Scikit-learn Deep Learning Convolutional Neural Networks (CNNs) Anomaly Detection
Project URL: Click Here

Project Description

Project: MIT: Automated Retail Spending
Date: 2024
Description: The Automated Retail Spending project is focused on predicting customer spending using automated feature engineering with Featuretools. The goal is to build a classification system do predict if a customer will spend over a threshold amount in the next month based on past transaction data.

Key techniques include:
- Data Exploration & Cleaning : Inspecting, preprocessing, and handling skewed spending distributions.
- Defining the Prediction Problem : Labeling customers based on whether they will spend more than $500 in the next month.
- Feature Engineering with Featuretools :
  - Creating entity sets and deriving new tables for feature extraction.
  - Utilizing deep feature synthesis to generate meaningful features.
  - Applying cutoff times to ensure temporal integrity in predictions.
- Model Training & Evaluation :
  - Using a Random Forest Classifier to predict customer spending.
  - Evaluating performance using ROC AUC, F1 score , and precision-recall curves .
  - Comparing automated feature engineering against a manually engineered baseline.

This project demonstrates how automated feature engineering enhances predictive performance in time-series retail analytics, reducing the need for manual feature creation while maintaining interpretability.

Skills & Frameworks: Python NumPy Pandas Featuretools Scikit-learn Random Forest Time-Series Prediction Classification Precision-Recall Evaluation Automated Feature Engineering
Project URL: Click Here

Project Description

Project: MIT: NYC Bus Ride Duration Prediction
Date: 2024
Description: This project applies machine learning to predict bus ride durations using data from various transport services in NYC. The dataset includes pickup and drop-off timestamps , passenger count , trip distance , and geographical coordinates .

Key techniques include:
- Data Preprocessing : Cleaning and transforming data, handling missing values, and selecting relevant features.
- Feature Engineering : Creating new features using automated techniques.
- Exploratory Data Analysis (EDA) : Analyzing relationships between variables such as trip duration and pickup/drop-off locations.
- Model Building : Training and evaluating predictive models, including Linear Regression, Decision Trees, and Random Forest.
- Performance Evaluation : Assessing model accuracy and feature importance to understand key factors affecting ride duration.

This project demonstrates how predictive analytics can optimize urban transportation systems, providing valuable insights for ride-time estimation and traffic management.

Skills & Frameworks: Python NumPy Pandas Matplotlib Seaborn Scikit-learn Linear Regression Decision Trees Random Forest Feature Engineering Model Evaluation
Project URL: Click Here

Project Description ^★

Project: MIT: CAPSTONE PROJECT 1: Cutomer Prediction Models
Date: 2024
Description: This project aims to build a machine learning model to predict the likelihood of leads converting to paid customers for ExtraaLearn, an EdTech startup. The goal is to analyze lead interactions, identify key conversion factors, and develop a predictive model to optimize marketing efforts and resource allocation.

The project required extensive:
- Data Cleansing
- Exploratory Data Analysis (EDA)
- Data Visualization, Preprocessing, and Feature Engineering
- Outlier Detection and Treatment
- Data Scaling
The following machine learning techniques were then used:
- Logistic Regression
- Principal Component Analysis (PCA)
- Decision Tree Classifier
- Decision Tree Pruning using:
  - Maximum Depth
  - Limited Depth
  - Minimum Samples per Leaf
  - Minimum Samples to Split)
- Decision Tree Classifier Hyperparameter Tuning
- Random Forest Classifier with Pruning and Hyperparameter Tuning
Due to a large data imbalance the following techniques were used:
- Class Weighting
- SMOTE
- Undersampling
- Hybrid - SMOTE Oversampling and Undersampling
Advanced regression techniques were then tested including:
- Random Forest Classifier with Polynomial Features
- XGBoost with Class Weights and Polynomial Features
- XGBoost with Threshold Tuning and Polynomial Features
- Majority Class Undersampling with Polynomial Features
- Optimized Hyperparameters for Recall
All models were then compared and the best used to analyze the dataset, and build the prediction model.
Skills: NumPy Pandas Matplotlib Seaborn SciPy Scikit-learn StatsModels XGBoost IPython Data Cleansing Exploratory Data Analysis (EDA) Data Visualization Data Preprocessing Feature Engineering One Hot Encoding Outlier Detection and Treatment Target Feature Separation Data Split Data Scaling Logistic Regression Principal Component Analysis (PCA) Decision Tree Classifier Decision Tree Pruning (Maximum Depth) Decision Tree Pruning (Limited Depth) Decision Tree Pruning (Minimum Samples per Leaf) Decision Tree Pruning (Minimum Samples to Split) Decision Tree Classifier Hyperparameter Tuning Random Forest Classifier Random Forest Pruning Random Forest Hyperparameter Tuning Data Imbalance (Class Weighting) Data Imbalance (SMOTE) Data Imbalance (Undersampling) Data Imbalance (Hybrid - SMOTE Oversampling and Undersampling) Advanced Feature Engineering Random Forest Classifier with Polynomial Features Gradient Boosting with Polynomial Features XGBoost with Class Weights and Polynomial Features XGBoost with Threshold Tuning and Polynomial Features Majority Class Undersampling with Polynomial Features SMOTE and Undersampling with Polynomial Features Optimized Hyperparameters for Recall Customer Profile Statistical Analysis
Project URL: Click Here

Project Description ^★

Project: MIT: CAPSTONE PROJECT 2: Amazon Product Recommendation System
Date: 2024
Description: In today\’s digital marketplace, personalized recommendations play a crucial role in enhancing user experience and increasing sales. The project utilizes Amazon's product review dataset, which includes user ratings for electronic products.

The primary objective of this project is to build a scalable recommendation system using different machine learning techniques to enhance personalized suggestions. The dataset lacks product descriptions and review texts to avoid bias, ensuring that the recommendations are purely based on user interaction patterns.

This project applies collaborative filtering , content-based filtering , and machine learning techniques to build an intelligent **Amazon product recommendation system**.

Key techniques include:
- Data Preprocessing (Handling Missing Data, Normalization, Filtering Sparse Data)
- Exploratory Data Analysis (EDA) to Understand User Behavior, product interactions, and rating distributions
- Building Recommendation Models :
  - Rank-Based Recommendations (Top-Rated Product Ranking)
  - User-User Collaborative Filtering
  - Item-Item Collaborative Filtering
  - Matrix Factorization (SVD) for Latent Factor Analysis
  - Clustering-Based Recommendation Systems (K-Means, Hierarchical Clustering)
  - Content-Based Filtering using TF-IDF
  - Hybrid Recommendation Systems (Combining Multiple Models)
  - Grid Search Cross-Validation for Hyperparameter Tuning
- Model Evaluation using RMSE, Precision@K, Recall@K, and F1-score
- Hyperparameter Tuning: Using grid search cross-validation to optimize recommendation models
This project highlights how machine learning-based recommendation systems improve customer engagement, optimize inventory management, and drive sales in e-commerce platforms.
Skills & Frameworks: Python NumPy Pandas Matplotlib Seaborn SciPy Scikit-learn Surprise Library Matrix Factorization (SVD) Clustering (K-Means, Hierarchical Clustering) TF-IDF Feature Engineering Grid Search Cross-Validation Precision@K, Recall@K, RMSE Evaluation
Project URL: Click Here

Project Details

Project: MIT: Travel Package Purchase Prediction
Date: 2024
Description: This project is to analyze the customers' data and information to provide recommendations to the marketing department and build a model to predict the potential customer who is going to purchase the newly introduced travel packagey. The analysis includes Exploratory Data Analysis (EDA), Data Preparation and Scaling Machine Learning techniques (including Logistic Regression, Support Vector Machine (SVM) using both Linear Kernel and RBF Kernel, Decision Tree Classifier, Random Forest Classifier), Hyperparameter Tuning using both Grid Search and Random Search, performs an evaluation of those models and Grid Search Tuned Decision Tree Classifier, and Grid Search Tuned Decision Random Forest Classifier..
Skills: Pandas NumPy Matplotlib Seaborn Scikit-learn Exploratory Data Analysis (EDA) Data Visualization Data Preparation Impute Missing Values Train/Test Split Data Scaling Logistic Regression Model Support Vector Machines - Linear Kernel Support Vector Machines - RBF Kernel Decision Tree Classifier Hyperparameter Tuning - Grid Search Random Forest Classifier
Project URL: Click Here
Published Model Project URL: Click Here

Robert Swetland

Personal Details

About Me

Experience

Independent Contractor

IT Manager / Senior Software Engineer

Software Engineer

Senior Software Engineer / Systems Administrator

Application Designer / SharePoint Administrator

Technology Specialist

Chief Information Technology Officer

Owner

Education

Degree Programs

University of Colorado

Oregon State University

Oregon State University

Certifications in Progress

University of Colorado

Vanderbilt University

University of Michigan

IBM

Completed Certifications

Massachusetts Institute of Technology

University of Colorado

Vanderbilt University

Duke University

University of Michigan

Adobe

Washington State University

Harvard University

Portfolio