At the early stage of my undergraduate career, I assisted George Ferguson's teaching in machine learning course and started doing research in our math and data science department. There I am fortunate to be advised by Alex Iosevich. We have discovered high dimension reduction tools involving fractal phenomena in theory and real life. Now we turn our attention toward a universal approximation for the situation under functions that the "typical approximator," Neural networks, cannot perform well on, including but not limited to the high dimensional functions with the fractal pattern. I also work with Yu Shen, studying the prediction models applied to topics related to environmental science.