Experimental Design of Applying Intelligent Computation to NIR Spectral Data Mining

Near infrared spectroscopy (NIRS) has been widely used in various areas of scientific research and technological engineering. This technique needs to deal with large volume of data collection and analysis. The procedure involved becomes a serious problem to undergraduate students learning this course as they usually lack of adequate mathematic knowledge. Fortunately, many programs of data analysis with intelligent computation modules are now available for learners. In this paper, an experimental design procedure is elaborated on how to use intelligent computation modules involved in data processing programs for the NIR spectral data modeling. First, the possible relationship between the physical and chemical features of experimental samples and their spectral data could be hypothetically created based on observation. Next, experimental samples as well as spectrometer should be well prepared for spectral data collection. Then, the collected spectral data are transferred into some of intelligent computation modules for data processing. Lastly, the computational result should be carefully summarized and written in the form of academic paper. To illustrate the whole process of how to apply intelligent computation to NIR spectral data mining, a case of variety identification of fragrant mushrooms based on NIR spectroscopy is introduced.