Spectral analysis technology has been widely used in the qualitative and quantitative analysis of complex systems due to its simple，fast，speed and non-destructive．However，spectra often contain hundreds or thousands of wavelengths（variables）．Some of them may not be related to the research object，which will lead to a large amount of calculation and reduce the prediction accuracy of model．Therefore，it is necessary to select the variables before establishing the model．The least absolute shrinkage and selection operator（LASSO） could be used to shrink the regression coefficients to zero，so as to achieve the purpose of variable selection．In this study，LASSO was used to select the variables for near infrared（NIR） spectra of ternary blend oil and Raman spectra of bio-fluid samples．The contents of sesame oil and sarcosine were quantitatively analyzed by building partial least squares（PLS） and multiple linear regression（MLR） models．The methods are compared with the three variable selection methods，i.e. uninformative variable elimination－PLS（UVE－PLS），Monte Carlo uninformative variable elimination－PLS（MCUVE－PLS） and random test－PLS（RT－PLS）．The result shows that the variable selection methods based on LASSO retain the least variable numbers and the fastest calculation speed．LASSO－PLS shows the highest prediction accuracy for ternary blend oil samples，while LASSO－MLR shows the highest prediction accuracy for bio-fluid samples．Therefore，the variable selection methods based on LASSO are expected to be well applied in the field of spectral analysis.