A basic Android application to find cocktail and mocktail recipes through use of an API.
Libraries/Resources used: Android Studio, TheCocktailDB's free API
This was a solo project in which I used a combination of Kotlin and XML in Android Studio. It can take you to a random drink, and you can save drinks for easy access later. Just a fun, simple app overall!

For this project, my group took inspiration from Google's Quick, Draw! to create a program that could identify a user's drawings.
Libraries/Resources used: Numpy, PyTorch, tkinter, matplotlib, PIL, QuickDraw Database
Our neural network model was trained on 10,000 images for each thing it can recognize: Axe, Baseball, Bed, Star, Cat, Flower, Suitcase, Snowman. Each image in the dataset is a grayscale 28x28 image, which we converted into an array of grayscale values from 0-255, totaling an array length of 784.
The first layer of our network is made up of 784 neurons. It goes through a ReLu activation function, and outputs 512 neurons. Another ReLu function scales it down to just 8 neurons. Here you can see it identify a snowman!

In this project, our group compared classification methods Logistic Regression, Decision Tree, and K-NN to find the most effective classifier for predicting heart disease from our data.
Libraries/Resources used: Kagglehub, Pandas, SKLearn, Matplotlib, Numpy, Heart Disease Indicator Dataset
We used data published by the CDC with NaNs removed. We encoded ordinal data to nominal data, then min-max scaled and trandformed the dataset. We concluded that the most effective classifier was Decision Tree, as it had high accuracy and the best recall and precision our of the models we tested. Additionally, we had to consider that false negatives are more dangerous in the medical field than false positives, so we aimed to minimize false negatives - which the Decision Tree did best.