Detection and analysis of unresolved multiplet chromatographic peaks

Computerized pattern classification of strongly overlapped peaks in stationary electrode polarography

“…A'pJA'p, (Areas) Efunction at 80 % ip AE at 75%/%, A'p, -A'p, (net area) / at AEip = 8 mV £ at 70% /%, Efunction at 75% /" Prediction ability for the 32 features of cycle No. 10 and the 22 features of cycle No. 12 was examined.…”

“…Intuitively, individual component curve shapes are more important than peak-height ratios or peak separations in the location of a pattern point. The fact that prediction ability varied considerably more between cycle No 10. and cycle…”

“…A trained observer can detect distortions in the data or its derivatives for more severe overlap conditions. Beyond the limit of visual interpretation, it may be possible to use simultaneous equations (1)(2)(3)(4), curve-fitting (5)(6)(7), calibration plots of second-derivative peak parameters (S), width parameters (9,10), moment analysis (11), self-modeling curve resolution (12), and characteristic vector analysis (13). Learning machine pattern classification methods have been reported for qualitative (14) and semiquantitative (15) analysis of mixtures.…”

Inferences from observations. Graphical intuition to Bayesian probability

“…Although straight lines are best, nonlinear models can be handled by use of an appropriate nonlinear transformation of the data. An example is the probit analysis of overlapping peaks (3).…”

“…The meter readings are proportional to absorbance. The particular form of the law applicable here, w'hich now may be called a model, is At = a(Vt + Fe) (3) where At is the meter reading, Vt is the volume of standard added for the ith solution, a is a constant, and F0 is the volume of standard equivalent to the amount of copper from the sample. Although it is the amount of copper in the sample that is of analytical interest, wre will limit our concern to the determination of the equivalent volume, Vc.…”