Exploring a Data Jungle

As a new product is taken from the laboratory, through the pilot, and into the production plant, a lot of information will accumulate that helps to better understand the product and the associated production process. In general, the major characteristics are soon evident and will be exploited to optimize the production. Fine-tuning the process can only be done, however, if hitherto hidden aspects are revealed. 

Situation A case in point is a purification/drying step before the sample 

Data Analysis Because of the danger of false conclusions if only one or two parameters were evaluated, it was deemed better to correlate every parameter with all the others, and to assemble the results in a triangular matrix, so that trends would become more apparent. The program CORREL described in Section 5.2 retains the sign of the correlation coefficient (positive or negative slope) and combines this with a confidence level (probability p of obtaining such a correlation by chance alone). 

So much for theory The interesting thing is to ponder what went contrary to expectations, and to try to find explanations. It is here that graphical presentations are helpful to judge the plausibility of any effect A correlation, is only a correlation, and can come about because of a single point that bucks the trend. 

Interpretation It seems as if the assay figures (columns 7, 8) were strongly negatively correlated with the presence of residual solvents (rows 3-5, not as expected) and to a lesser degree with water. The up to 8% water have a direct effect on the assay, unless compensated for, while even 50 ppm of E 

---