matthew-fournier-536154-unsplash (1).jpg


Technique: Deep Neural Network

For this project I created a deep neural network with 4 dense layers and 3 dropout layers that had a learning rate of 0.1. After fitting my model to the training data with a batch size of 7 and 100 epochs I used a GridSearch algorithm to help me find the best parameters.


Dogs or Not-Dogs

Technique: Convolutional Neural Network

In this project used a Deep ConvNet to classify images as cats or dogs with 99.1% accuracy. In this project the goal was to predict whether an image was a cat or dog using a convolutional neural network. Using both Tensorflow and Keras I was able to harness the power of a 2 layer CNN.


Wine Quality

Technique: Decision Trees & Random Forests

My tree is split on alcohol, sulfates, and volatile acidity. Higher quality wines tend to have a higher percentage of alcohol and sulfates. The majority of wines in the training set were ranked as 5 or 6, or "Okay" in terms of my grouped ranking.


Trump Regrets

Technique: NLP

It's clear Trump has a major problem with Hillary Clinton. He mentions "hillari" 455 times in the selected tweets and "america" just 227 times.

Here are some frequently used words by people who regret voting for Trump: vote, trump, regret, now, make, support, like, lie, promis, stop. We can extrapolate what we want here, but the topcis are clear, the people who regret voting for Donald Trump are REALLY unhappy about it.