Summary
In this chapter, we learned about the Fourier transform in its original form as well as its discrete counterpart, the discrete Fourier transform. We touched on the algorithm used to compute that discrete version, the Fast Fourier Transform, and detailed the application of data analysis from the perspective of the frequency domain.
With the audio signal data converted into cross-sectional data in the frequency domain, we built a classification model to predict the source of the signal. To do this, we employed a binning technique to aggregate adjacent frequencies from the output of the Fourier transform into average columns. This significantly reduced the dimensionality of the newly generated cross-sectional data without significant information. We further tuned these bins to be narrower in the center band of the frequency spectrum where we believe most information are held. This assumption was made because the audio signals were for human hearing and this middle section corresponded...
