Contributors: Di Ye, Hao Wang, and Hannah Pu
Department of Statistics, University of Illinois at Urbana-Champaign
LASSO Regression, logistic regression, neural network, random forest, and xgboost have been used in this project to identify whether each image in the Kaggle datasets contains volcanoes or not, and if there is any, how many volcanoes are there. The analysis of the images revealed that there are several volcanoes on Venus. Neural network xgboost both had good performance to identify whether the images have volcanoes in this project for image classification with an area under the curve of 0.9629 (neural network) and 0.9680 (xgboost), and F-1 score of 0.9747 (neural network) and 0.9628 (xgboost). The comparison between different models is also included.
Keywords:
- Image Classification
- LASSO Regression
- Logistic Regression
- Marginal Screening
- Neural Network
- Random Forest
- Xgboost
- K-means
- Linear Discriminant Analysis
- Final Project Rmd File (proj-final-report.rmd)
- Final Project pdf File (proj-final-report.pdf)
- Project proposal (Proposal folder)
- Cleaned R Data link: https://goo.gl/J6AAqp
- Images and more details can be found in https://github.com/diye2/Volcanoes-Project