The Sun is capable of generating massive explosions called solar flares. The energy powering these events is derived from the Sun's complex magnetic field and is released spontaneously in a process known as "magnetic reconnection". During such an event, the entire electromagnetic spectrum experiences an enhancement, and a variety of instruments analyze different aspects of the spectrum to uncover the underlying physics of flares. Because solar flares can negatively impact the earth's geomagnetic envelope and compromise many of the electronics that our society depends on, it has become paramount to better understand and possibly predict these events. In service of these ambitions, we use modern machine learning techniques to analyze large volumes of spectral data collected by NASA's IRIS satellite. We find techniques that allow us to automatically analyze millions of spectra efficiently, calculate information-theoretic quantities, as well as provide the first flare prediction results using a novel spectral dataset.