Correlation Analysis
Quantify the linear correlation strength between two variables. |r| = 1 means perfect correlation, 0 means no correlation. Run correlation analysis first to confirm regression is worthwhile.
A process engineer suspects injection pressure (X1) and mold temperature (X2) are highly correlated; if they move together, it is hard to tell which parameter truly drives dimensional variation. Run a correlation matrix first to detect multicollinearity before regression or DOE, avoiding a misleading model. Quality databases often log dozens of process parameters. Correlation analysis quickly screens candidate factors significantly related to the CTQ, focusing improvement resources.
Input 20 records each of injection pressure X1 and mold temperature X2 Pearson r(X1, X2) = 0.85, p < 0.001 → Strong positive correlation; temperature rises with pressure → Likely controlled by the same heating system — verify separately Pearson r(X1, Y_dimension) = 0.72, p = 0.0003 Pearson r(X2, Y_dimension) = 0.69, p = 0.0008 → Both process parameters correlate significantly with the dimension