Near-infrared spectroscopy (700–2500 nm) is primarily composed of absorption peaks from the overtone and combination bands of hydrogen-containing groups. The absorption intensity is weak, the sensitivity is relatively low, and the absorption bands are broad with significant overlap. Therefore, for quantitative analysis using near-infrared spectroscopy, if one were to use a single wavelength or multiple wavelengths based on the traditional UV-visible spectroscopy method that relies on the Lambert-Beer law to establish a calibration curve, accurate and reliable analytical results are typically not obtained. As shown in Fig. 3.1, since the late 1970s, the application of chemometric methods in near-infrared spectroscopic analysis has led to a leap in its development. Nowadays, almost all quantitative and of near-infrared spectra are achieved using chemometric methods to establish multivariate calibration models.