Posted by Christopher Mertin on December 21, 2018 in Project • 7 min read
Using a word2vec model trained on tweets allows for finding what words are most similar to a given emoji. This is provided through an interactive Flask API.
Posted by Christopher Mertin on August 21, 2017 in Project • 13 min read
Using statistics, deep learning, SVMs, and Gaussian Mixture Models on customer banking information to build a predictive model to optimize who the bank markets term deposits to so that marketing resources are not wasted on those unlikely to enroll.
Posted by Christopher Mertin on July 24, 2017 in Project • 16 min read
Uses public data from Rossmann stores with feature creation to analyze store sales and create a business forecast. Built models to predict sales at certain stores for a given day, as well as found the most important factors driving sales.
Posted by Christopher Mertin on June 09, 2017 in Project • 8 min read
Using data pertaining to Super Symmetry (susy), I explore the use of various algorithms to try and identify particles from data. I look at the use of a Random Forest, Gaussian Naïve Bayes, Multi-Layer Perceptron, and Deep Learning with the use of Dropout.
Posted by Christopher Mertin on May 05, 2017 in Project • 2 min read
Masters project at the Univeristy of Utah which explores the use of Hierarchical Matrices to increase the learning rate and computation speed of Deep Neural Networks.
Posted by Christopher Mertin on April 29, 2017 in Project • 1 min read
Explores the use of eigenfaces and clustering algorithms to differentiate people with religious heritage with up to 80% accuracy.
Posted by Christopher Mertin on February 19, 2017 in Project • 13 min read
An extension of the previous movie recommendation system, however with the use of Latent Factors with Keras to create a Neural Network to calculate the similarities instead of using MapReduce. This is a much more accurate implementation as it allows the addition of a bias term.
Posted by Christopher Mertin on January 15, 2017 in Project • 8 min read
With the use of the MapReduce ecosystem, and user movie ratings from the movie lens data set, I build a system using the cosine distance between user movie ratings to find movies that are similar to each other.
Posted by Christopher Mertin on December 22, 2016 in Project • 5 min read
Uses data from the Titanic to predict if someone would survive or not on the titanic based on various characteristics.
Posted by Christopher Mertin on December 02, 2016 in Visualization • 1 min read