My name is Seth Newton and I am a cheminformatics data scientist. As most of my work has been with the federal government, I cannot post the code here.
Nevertheless, here are some repositories of personal projects I'd like to direct you to that showcase my skills:
- Recurrent Neural Network to Generate Novel Boy Names - popular baby boy names were fed into a recurrent neural network that works at the character level to predict novel boy names. It generated some interesting names!
- Unsupervised Learning to Categorize Stock Headlines - here I use unsupervised learning techniques to cluster stock news headlines, finding that one cluster contains more editorial/opinion type articles compared to reports on actual company events
- Supervised Learning to Predict Stock Movements Based on Headlines - after classification of news headlines using unsupervised learning, I use sentiment analysis of event-driven headlines fed into a support vector machines model and a random forest model to classify news headlines into those likely to cause a positive price movement, negative movement, or unlikely to cause any movement