Analyze the Performance Data

Analyze the Data

The next step after the data have been entered and checked for accuracy is analysis of the data with the ultimate goal of determine optimal mixture proportions. Before proceeding with the analysis, you may make any desired changes to the response variable limits (performance criteria), weights, and optimization function types which were entered in Step 1.

Change response variable limits, weights, and function types

COST provides both graphical and numerical analysis of your data. The graphical analysis includes various types of plots which are described below. The numerical analysis includes fitting an empirical model to the data, verifying adequacy of the model, and numerical optimization.

Performing the Analysis - Task by Task

Task 1: Characterize the Response Variables
Tool: Summary Statistics

Task 2: Assess the Balance of the Design
Tool: Counts Plot Matrix of Factors

Task 3: Assess Optimality of Design Points for All Responses Jointly
Tool: Matrix of Counts-in-admissible-region plots

Task 4: Assess Optimality of Design Points for All Responses Jointly
Tool: Matrix of Percentage-in-admissible-region plots

Task 5: Determine Interrelationships between Response Variables
A. Tool: Scatter Plot Matrix of Response Variables

Scatter Plot Matrix of Response Variables versus Factors

Task 6: Assess Relationship Between Response Variables & Factors
Tool: Mean Plots of Response Variables versus Factors

Task 7: Determine Optimal Settings for Each Factor
A. Tool: Best Settings Based on Mean Values
B. Tool: Best Settings Based on Individual Runs

Task 8: Model Fitting and Verification

Model Fitting Tool

Task 9: Numerical Optimization

Best Settings - Maximization of Total Score Fitted Function

Task 10: Response Prediction
Tool: Response Prediction Tool

Return to top of page