Supervised Machine Learning models are trained on a combined data-set from CalTech and STScI to identify probable planet baring stars in NASA’s TESS satellite light-curve data.
NASA’s TESS satellite is a started taking light readings of 200,000 of the brightest, closest stars in 2018. They expect to find about 300 earth-sized exoplanets within the next year. Data from this on-going project is being made available to the public. For my very first co-op project, another Data Science student and I paired up to see if we could find anything interesting.
To be clear, neither of us are astrophysicists. Nonetheless, after brainstorming together, we came up with a way for a way that machine learning might contribute to NASA’s project without any reference to physics. To train a classification model, all we needed was a set of data to train on with and without exoplanets —think hotdog / not a hotdog from the show Silicone Valley. We had access to the the raw data without any classification. If we could get a list of stars in the data that had already been established as likely to have planets, then we were in business. …
An App to Rank Cities based upon user selected factors
Project Roles & Objectives:
This project was put together over the course of eight weeks by a team of 14 students with four disciplines — UX, iOS, Full Stack Web, and Data Science. We were given two weeks to plan, five weeks to build, and one week to clean and present our documentation. Much of the user research and data collection had been done by a previous team. There were two DS students on this team. …
A CNN for image recognition is trained on results from Google Images API for city names. The array of probabilities provided by a softmax output layer are examined to determine subtle similarities between cities. The plans for a “City Vibes” swiping app are here discussed.
This project originally started as a feature for iOS implementation of another app — Best Places to Live (link). In the third release cycle of our project, I teamed up with an iOS student to build out the feature described here. Fortunately for my partner, he was offered paid position midway through development. As a consequence, the front end of this feature has not yet been completed. Despite this setback, I have included my work on this feature as a stand-alone story in my project. …
