Fourier transform infrared spectra of a representative pool of the ceramic bodies of 75 shards excavated in the archaeological district of Canosa (Puglia) were analyzed by principal component analysis (PCA) and soft independent modeling of class analogy (SIMCA) with the aim to establish reference groups for the purpose of assigning future samples. PCA analysis using the spectral data comprised between 1260 and 440 cm−1 (410 wave numbers or data points) showed that the first three principal components (PC) describe most of the total spectral variance. By means of PCA, most of the information related to firing temperature and temper type was explained by the first PC: the score plot on the first and second PCs confirmed the same grouping of the samples as previously performed according to classification criteria determined by means of a detailed attribution of all the mid infrared absorbance peaks. SIMCA modeling carried out at 95% confidence level on second derivative pre processed data were successful too, but one object, which was assigned to a wrong class. Interestingly, SIMCA proved to be a promising tool to rapidly classify ceramic samples.
FTIR-chemometric tools as aids for data reduction and classification of pre-Roman ceramics
DE BENEDETTO, Giuseppe, Egidio;
2005-01-01
Abstract
Fourier transform infrared spectra of a representative pool of the ceramic bodies of 75 shards excavated in the archaeological district of Canosa (Puglia) were analyzed by principal component analysis (PCA) and soft independent modeling of class analogy (SIMCA) with the aim to establish reference groups for the purpose of assigning future samples. PCA analysis using the spectral data comprised between 1260 and 440 cm−1 (410 wave numbers or data points) showed that the first three principal components (PC) describe most of the total spectral variance. By means of PCA, most of the information related to firing temperature and temper type was explained by the first PC: the score plot on the first and second PCs confirmed the same grouping of the samples as previously performed according to classification criteria determined by means of a detailed attribution of all the mid infrared absorbance peaks. SIMCA modeling carried out at 95% confidence level on second derivative pre processed data were successful too, but one object, which was assigned to a wrong class. Interestingly, SIMCA proved to be a promising tool to rapidly classify ceramic samples.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
