dramadas

Portfolio Projects

1) Data Science Projects

These projects are a combination of Machine Learning and Statistics completed in personal time. Below is a summary of projects and skills used.

Statistics

• Statistical Analysis of Medicare Telehealth Trends
  • t-test, ANOVA
• Time Series Forecasting using Exponential Smoothing and ARIMA
  • Autocorrelation Function (ACF), Partial Autocorrelation Function (PACF), Prediction Interval

Machine Learning

• Image Comparison using Perceptual Hashing and Nearest Neighbors
  • Annoy, Hamming distance
• Segmentation modeling using UNet
  • Fine-tuning model parameters, train/test loss evaluation, Accuracy/Precision/F1 metrics, ONNX
• Diffusion model for femur image generation
  • UNet optimization, parametrization, FID metrics
• Segmentation modeling using fine-tuned SAM
  • Transformers, train/test loss evaluation

2) UC Berkeley Projects

These projects were completed during my Master’s in Information and Data Science program at UC Berkeley. If the assignment was completed collaboratively as part of a group, it is noted. Below is a summary of projects and key skills used.

Statistics

• Statistical Analysis of Spend vs Ad Impressions for Snapchat
  • Linear Regression, Data Cleaning
• Linear Regression Modeling to predict bike rental usage
  • Feature Selection, One-Hot Encoding, Ordinary Least Squares Regression
• Timeseries Forecasting using Prophet in Apache Spark (Databricks)
  • Parallelization, Feature Engineering
• Experimental Design for Water Conservation
  • Research Question Formation, Study Design, Sampling, Statistical Methods

Exploratory Data Analysis and Visualization

• Exploratory Data Analysis on Food Insecurity in the US
  • Data Cleaning, Data Visualization