As a part of non-invasive and unaware measurement of physiological signal in the house of live-alone person, Raman spectroscopy was applied for urine component analysis in the toilet set. 785nm, 250-300mW output solid state diode laser and 2048 element linear silicon TE cooled CCD array were incorporated for this system. Several tests were performed for setting up Raman spectroscopy in non-constrained situation: toilet set in the house. The effect of dark current, integration time, warming up time of laser, property of probe and interference of water in the toilet were tested and controlled for appropriate measurement in this environment. The spectra were obtained immediately when the subject uses the toilet set, and they can be transmitted to the server though Bluetooth. Those spectra were pre-processed for removing or correcting the effect of undesired light scattering, sample path-length difference and baseline-effect. The preprocessed data were enhanced for more exact result of multivariate analysis. The training data was prepared for predicting unknown component and its concentration by using multivariate methods. Several kinds of multivariate methods: PCA, PCR, PLS were performed to validate what is the fittest method in this environment. Through quantitative and qualitative analysis of Raman spectroscopy’s spectra obtained in the house's toilet set, we could know the component and its concentration of urine which can be index of disease.