PCA is an unsupervised learning method that converts the original data into a new data space. It is widely used in various kinds of fields, but isn't limited to machine learning problems. Although PCA doesn't force us to map the data to a different dimensional space, it is frequently used in dimensionality reduction. This is because it improves the prediction result efficiently by using the low-dimensional space.
PCA is a useful approach for finding patterns based on the correlation between features in the observation. In short, the goal of PCA is to discover the coordinates that maximize the variance of each feature:
The preceding diagram is an example of projecting two-dimensional observations into a one-dimensional space. The original data exists in the XY coordinates. Assuming that we mapped the original data points into...
