The main.py file demonstrates how rTCA address the problem of covariate shift. This example is a 2-class classificaction problem. First, we use data_generator.py file to generate the two 3-dimensional datasets under the scenario of covariate shift. We can generate different types of covariate shift by adjusting the 'translation' and 'angle' parameters.

The rTCA.py file contains the algorithm of reduced-kernel TCA, we can adjust the parameters to achieve the better result.