The PDS Data Science Grant provides Microsoft Azure credits and mentorship for motivated students who want to carry out projects using data or data science techniques. This grant was created in Spring 2020, and it has successfully helped fund 6 projects (so far!), which range from creating a dynamic digital 3D model of Streicker Bridge on Princeton’s campus to detecting three different types of retinal diseases from medical scans using neural networks. The grant recipients present their research at a joint Center for Statistics and Machine Learning (CSML) and PDS event at the end of every school year. Princeton Data Science is grateful to the CSML for generously supporting the creation of this grant by providing Microsoft Azure credits for each grant recipient to use for their project. Recipients of the grant receive up to $1000 in Azure credits, which is given out on a per need basis.Grant applications are open and can be found here!
Deadline: September 29th at 11:59 pm
My project aims to create a Digital Twin (a dynamic digital 3D model of a physical asset) of the Streicker Bridge on Princeton’s Campus that integrates Structural Health Monitoring data and updates the model to reflect the real-time state of the bridge throughout its lifecycle. When the Streicker Bridge was constructed in 2009, sensors were installed within it and previous work in the Civil and Environmental Engineering Department has been done on installing, monitoring, and interpreting the data from these sensors. I aim to build on this research and explore new ways to visually represent and analyze this data by constructing a 3D geometrical model of the bridge and link it to Microsoft Azure’s Digital Twins technologies to create a model that is both geometric and computational to serve as a singular efficient source for all information regarding the structure. Further work in data visualization, mixed reality, and machine learning will explore the possibilities of such a model. This project is rooted in civil engineering and infrastructure management, but explores the intersection between engineering, computer science, and data science and I greatly appreciate the support of Princeton Data Science!
My project aims to use transfer learning to examine optical coherence tomography (OCT) scans in order to detect three different types of retinal diseases: choroidal neovascularization (CNV), Diabetic macular edema (DME), and Multiple drusen (Drusen), and distinguish diseased from normal retinas. The goal of the project is to apply various convolutional neural network architectures that have been pre-trained on other image classification tasks in an attempt to outperform baseline custom CNN models on this task. A stretch goal is to use visualization techniques to highlight the regions of the image that are most informative to the models in classifying the diseases. This work explores an important health issue and could be useful in providing supplementary information to ophthalmologists and retinal specialists in the complex task of grading OCT scans.
My project will largely focus on conducting exploratory data analysis on datasets involving 911 calls in different cities. I hope to explore the different factors that may have an impact on 911 response times and to possibly gain new insights on ways that we could improve these response times.
Satya Nayagam/Brian KangThe aim of this project is to leverage stored data on customer transaction history to make informed predictions about future purchasing behavior, namely next purchase date and spending amount, using ML. Having the ability to gain insights into such features will be significant to businesses for maintaining customer loyalty, reducing churn and acquisition costs, and boosting customer lifetime value. The short term focus of this project will be to improve the accuracy of our current model at predicting the next purchase date of a customer.
Please do note that Princeton Data Science is a student organization and not a university department. Consequently, we do not sponsor PhD or Masters students.