Artificial Intelligence By Example

Creating the feature vector

The remaining step is to sort the eigenvalues from the highest to the lowest value. The highest eigenvalue will provide the principal component (most important). The eigenvector that goes with it will be its feature vector. You can choose to ignore the lowest values or features. In the dataset, there will be hundreds and often thousands of features to represent. Now we have the feature vector:

Feature Vector=FV={eigenvector1,eigenvector2...n}

n means that there could be many more features to transform into a PCA feature vector.