Non-linear robust regression procedure for calibration in flame atomic absorption spectrometry

✍ Scribed by Paweł Kościelniak

• Publisher: Elsevier Science
• Year: 1993
• Tongue: English
• Weight: 929 KB
• Volume: 278
• Category: Article
• ISSN: 0003-2670
• DOI: 10.1016/0003-2670(93)80095-3

✦ Synopsis

Robust regression methods based on the median, namely the single median and the repeated median methods, were developed to adapt them to those cases occurring in flame atomic absorption spectrometry (FAAS) when non-linear calibration relationships are encountered. All essential features of the median methods were preserved owing to the modification suggested and the regression procedures are simple and easy to apply, allowing outliers to be detected and rejected. The resistance of the methods considered to the effect of outliers is discussed. The effectiveness of the tiedian approaches and some advantages over the least-squares non-linear regression method are revealed in examples with both hypothetical data and real calibration data obtained from FAAS.