Bird Bone Classification: Distinguishing Terrestrial and Flying Birds Based on Skeletal Structure

Overview:

This repository contains a classification model that identifies birds as either terrestrial or flying based on their skeletal structure.

Key Findings:

• Model accuracy: ~85.5% on the testing dataset.
• TPR for flying birds: 86.4%
• TNR for terrestrial birds: 82.4%

Dataset:

Measurements from 420 bird skeletons sourced from the National History Museum of Los Angeles County.

Methodology:

1. Data Preprocessing:

   • Standardization of predictors.
   • 25/75 split for testing and training.

2. Modeling:

   • K-Nearest-Neighbour classification.
   • K value: 3.

3. Features:

   • Wing length (ulna and humerus).
   • Leg length (tarsus, tibiotarsus, femur).

Impact:

A benchmark for researchers to classify birds by bone structure.

Future Directions:

• Explore wing-to-leg length ratio and bird body mass.
• Consider bone diameter or density.
• Examine individual bone lengths.

Contributors:

• Jiakang Huang