Predictive Modeling using KNN and Logistic Regression

UC San Diego COGS 109 Course: Modeling and Data Analysis

Motivation —

Judging watermelon quality based on its apparent properties such as texture or color is a scientific concern for farmers, distributors and consumers. My team of 3 members wanted to explore this as it was summertime. Our goal was to verify if the sound made when tapping and the amount of sugar are the best predictors for the quality of a watermelon.

My Contribution —

I was in charge of creating all the visualizations, like the one to the right with some of the dataset features. I performed the EDA and developed the KNN model for our prediction task. Then, I reported its accuracy to my team for comparison with logistic regression. I documented most of our report using LaTeX.

Categorical Predictors