This project is part of the Master's Capstone at Duke. In this project, we explore the phenomenon of Prediction Churn and methodology developed to solve it, in the quest of creating a novel method to address our client's needs.
Built an Onsite Earthquake Early Warning system prototype using Gaussian Process Regressions, a class of Machine Learning algorithms, trained on seismological data.
CT Scans take multiple snapshots of subjects that are reconstructed to create the final image. In this project, we reduce the number of snapshots taken to find the optimum at which deeplearning models can perform reliably well on object detection tasks.
Github
In this project, we aimed to determine the causal effect that a host being a super host has on the revenue generated by them, and found a significant effect.
U-Nets used for in image segmetation can classify precise regions where a brain Tumor occus. In this project, we detect and segment a specific type of tumor - Glioblastima Multiforme on clinical 3D MRI data.
Github
Which Star Trek Characters are audience favourites? Here, I have used data from IMDb and episode scripts to help answer this question.
Github
This tool is for everyone who wanted to retain a twitter presence without bothering to actually look at what's trending. Generates synthetic new tweets for you everyday based on tweets from the previous 24hr discussions.
Github