Hi class, in this project, we carried out model evaluation and selection for predictive analytics on image data. As data scientists, we often need to evaluate different modeling/analysis strategies and decide what is the best. Such decisions need to be supported by sound evidence in the form of model assessment, validation and comparison. Because of this we used 3 different models to compare each analysis. Each one had a different efficiency. We had to walk the fine line of complexity vs efficiency in this project. In addition, we also need to communicate our decision and supporting evidence clearly and convincingly in an accessible fashion. Here is our repository organized for reproducibility and ease of understanding.
Term: Spring 2021
-
Team 6
-
Team members
- Weiwei Song
- Changhao He
- Aurore Gosmant
- Zi Fang
- Jimmy Smiley
-
Project summary: In this project, we created a classification engine for facial emotion recognition.
- 78 fiducials points of each emotion are marked in order for analysis.
- Distributions of vertical distance between right pupil(1) and right brow peak(21) are shown in histograms. For example, the distance of an angry face tends to be shorter than that of a surprised face.
- The distributions of vertical distances between right mouth corner(50) and the midpoint of the upper lip(52). For example, the distance of an happy face tends to be shorter than that of a sad face.
Contribution statement: (default)
Weiwei Song implemented several algorithms counting XGBoost, XGBoost+SMOTE, Random Forest algorithms. She is also responsible for writing the related .Rmd and R script files, and preparing the slides for the class presentation.
Aurore Gosmant is responsible for exploring through the data and the features generated by the fiducial points. She also implemented the PCA + XGBoost algorithm, tuned hyperparameters, ran cross validation and generated the .Rmd and .R script files corresponding.
Zi Fang implemented and tuned the PCA + Linear Discriminant Analysis and KNN algorithms, and generated the associated .Rmd and .R scripts corresponding. She also worked on XGBoost cooperatively with Weiwei Song.
Changhao He is responsible for implementing and running the baseline model, tuning hyper parameters. He wrote the R files about the baseline model and finalized the main.Rmd. He also ran the best model on the test data given in class.
Following suggestions by RICH FITZJOHN (@richfitz). This folder is orgarnized as follows.
proj/
├── lib/
├── data/
├── doc/
├── figs/
└── output/
Please see each subfolder for a README file.