Extraction of Maximum Analytical Information from the Data

4 Extraction of Maximum Analytical Information from the Data 

Conventionally at present, the analyst may perform many different determinations on one sample, but then use each result independently for the decision-making process, for which the analysis is demanded. More recently, it has been recognised that the results used in totality may yield much more information about the sample. Three such methods will be taken as representative. 

Factor analysis. It is a truism that any observation may be a compound of several factors. In analysis one often seeks to hold all but one factor constant and measure the one against some standard.- This approach has much to commend it a result is obtained and the complexities involved in complete understanding of the system are avoided. In other disciplines it has not proven so easy to avoid the complications e.g. in psychology and social science the relative importance of hereditary and environmental characteristics or the relationship between race and intelligence are matters of controversey, and so the field of factor analysis has been developed. The objective is to extract from the data available, the number of independent factors present and their relative importance. 

A chemical example is the mass spectrum of a mixture of oils. It is possible to construct a matrix which relates relative intensity of a component A at an m/e value n, an etc.  

The observed signal at m/e of n is given by XAan + XBbn + X Cn. .. where XA is the fraction of component A in the mixture. By appropriate matrix manipulation, which constitutes the factor analysis, it is possible to calculate XA, XB etc. Interestingly, it is also possible to deduce the number of independent factors, so that if an unexpected oil were present in the mixture this would be shown up. 

---