In this chapter, we introduced the traditional way of analyzing sequential data. Fourier transformation is an established method of unfolding data into multiple factors, just like Taylor expansion does. The most important thing about Fourier transformation is that it is able to find the fundamental building blocks of the periodic sequential data, which is the frequency in a complex numerical space.
TensorFlow.js allows us to access the fast implementation of Fourier transformation. Due to this, we looked into how to use FFT and inversed FFT for compound cosine curves and observed the results. In the next chapter, we will look at dimensionality reduction and t-SNE.
