A feature selection method combined with ridge regression and recursive feature elimination in quantitative analysis of laser induced breakdown spectroscopy

Abstract

In the spectral analysis of laser-induced breakdown spectroscopy, abundant characteristic spectral lines and severe interference information exist simultaneously in the original spectral data. Here, a feature selection method called recursive feature elimination based on ridge regression (Ridge-RFE) for the original spectral data is recommended to make full use of the valid information of spectra. In the Ridge-RFE method, the absolute value of the ridge regression coefficient was used as a criterion to screen spectral characteristic, the feature with the absolute value of minimum weight in the input subset features was removed by recursive feature elimination (RFE), and the selected features were used as inputs of the partial least squares regression (PLS) model. The Ridge-RFE method based PLS model was used to measure the Fe, Si, Mg, Cu, Zn and Mn for 51 aluminum alloy samples, and the results showed that the root mean square error of prediction decreased greatly compared to the PLS model with full spectrum as input. The overall results demonstrate that the Ridge-RFE method is more efficient to extract the redundant features, make PLS model for better quantitative analysis results and improve model generalization ability.

A Versatile Method for Quantitative Analysis of Total Iron Content in Iron Ore Using Laser-Induced Breakdown Spectroscopy

“…In addition to the emission line of the target element Fe for quantitative analysis, the emission lines of other elements are also retained, such as, K, Na, Al, Ca, Mn, Mg, Si, Ti, Zr, and Li, indicating that the calibration results of TFe content are considerably depending on a lot of other elements. 24 As expected, these selected features are the peaks and the wavelengths around the peaks, that is, the emission peak region contains more useful LIBS spectral information for predicting. 21 It is worth noting that partial spectral baseline of 540-610 nm and 820-970 nm were selected in which there were almost no characteristic emission lines.…”

“…2,21 Feature selection is another effective measure to increase the LIBS-based quantitative accuracy, through which the dimension of data set is reduced by selecting a most effective feature subset from the full spectrum. 2,9,[21][22][23][24] In particular, the commonly used feature selection methods in LIBS analysis are principal component analysis (PCA), 25 SelectKBest, 23 particle swarm optimization (PSO), 2 and genetic algorithms (GA). 26 In recent years, variable importance (VI) produced by RF has been used to improve predicting ability in the fields of metallurgical industry and environmental protection.…”

Psychophysiological Arousal in Young Children Who Stutter

“…Hence, this section aims to identify the features’ relative importance rather than feature selection. Following prior studies [ 22 , 29 , 82 , 93 , 94 ], we employ ridge-regression. Ridge regression adds ‘squared magnitude’ of coefficient as penalty term to the loss function, hence highly penalizes coefficient of less important features.…”

Way-out for laser-induced breakdown spectroscopy

“…In addition, there are ten articles regarding data processing and different applications. A series of different new data processing methods were proposed, such as variable selection [12][13][14], principal component analysis Mahalanobis distance [15], combination of support vector machine and partial least square [16], as well as the classification method based on the elemental intensity ratio [17]. It is worth mentioning that variable selection is an important way to improve the quantification performance and has attracted more interest [12][13][14].…”

“…A series of different new data processing methods were proposed, such as variable selection [12][13][14], principal component analysis Mahalanobis distance [15], combination of support vector machine and partial least square [16], as well as the classification method based on the elemental intensity ratio [17]. It is worth mentioning that variable selection is an important way to improve the quantification performance and has attracted more interest [12][13][14]. The application not only included steel classification [17], aluminum alloy analysis [14], coal analysis [12,16], geographical authenticity evaluation [18], and plants classification [15], but also included some new or uncommon application fields such as uranium measurements [19], mercury measurements [20], soil pH measurements [13], and rock salt analysis [21].…”