This is a Jupyter Notebook that implements PCA for face recognition using the Olivetti Faces dataset provided by scikit-learn's fetch_olivetti_faces function. The notebook demonstrates how to use the PCA algorithm to reduce the dimensionality of the dataset and visualize the results.
PCA is a dimensionality reduction technique that transforms a set of correlated features into a set of uncorrelated components that capture most of the variance in the data. PCA can be used to reduce noise, visualize data, and improve performance of machine learning models.
The notebook contains the following steps:
- Loading and exploring the dataset
- Applying PCA to reduce the number of features from (400, 4096) to 2d plane
- Visualizing the 2D projection of the data with different colors for each activity
- Evaluating the explained variance ratio of PCA