Thinking back to when you first learned regression analysis, do you remember discussing the differences between input/predictor variables and response/ measured variables, or simply X’s and Y’s?
As analysts, we are taught to divide our variables into these two categories. But, as data sets grow to include multiple unit operations, categories of ingredients, or categories of product features, there are better ways to think about your variables… in logical groupings, or blocks.
Grouping variables into blocks for multivariate analysis allows you to answer questions like:
- Which unit operation has the most impact on final product quality?
- Do raw material properties matter more than process conditions?
- Which category of ingredients has the biggest influence on product performance?
Depending on what questions you hope to answer, these blocks can move between the X-space and Y-space in multivariate analysis. Want to learn more about multi-block modeling in multivariate analysis? Check out our multivariate analysis courses.