This repository contains our STA314 project work on classifying pet facial expressions into Angry, Happy, and Sad.

The final and best-performing approach in this repo is the tree-based modeling workflow in ensemble_trees.ipynb. If you want the notebook that represents the final model, start there.

• combined_code.ipynb A combined notebook that places the code from the tree-based workflow, logistic regression baseline, and MobileNetV2 workflow into a single file. The notebook is ordered with trees first, followed by logistic regression, then MobileNetV2, and includes markdown cells introducing each model section.

• ensemble_trees.ipynb The final notebook and strongest model in the repository. This notebook builds the tree-based pipeline, compares feature combinations, and develops the final tree-oriented approach used for the project conclusions.

• MobilenetV2.ipynb A transfer-learning notebook using MobileNetV2 for image classification. This is part of the deep learning exploration, but it is not the final model we selected.

• logisticregression.py A baseline script that converts images to grayscale, resizes them, flattens them into vectors, and fits a logistic regression classifier. This serves as a simpler benchmark against the more advanced models.

• eda.ipynb Exploratory data analysis notebook for inspecting the dataset, image sizes, preprocessing choices, and dataset structure before training the main models.

• data/classification-of-pet-facial-expression/ The dataset directory containing the train and test image folders used by the notebooks and scripts.

• data/classification-of-pet-facial-expression.zip A zipped copy of the dataset.

• LICENSE Project license file.

• .gitignore Git ignore rules for local machine artifacts.