Finding the Best Pear-forming Model: An Image Classification Problem

Last semester, I took a course all about predictive modeling. Regression, decision trees, clustering, neural networks...you name it, we did it (and if we didn't, we probably will this semester)!

One of the really cool opportunities I had was to work on an image classification project. Our task: create a model that can take a picture of a fruit as an input, and tell us what type of fruit it is. Sounds easy, right? Well sure, between a banana and an apple, that's no problem at all. How about differentiating between 13 different apples? Now, not so much.

So with our goal in mind, we set out to create various models to find the best one. From K-Nearest Neighbors (KNN) and Random Forests, to Convolutional Neural Networks (CNN) and Transfer Learning, we tried a variety of different models to achieve the best classification accuracy. We published our work on Medium; follow along here!