The MDMVCC platform accepts data in the following three allowable data formats:
Raw Data
Use continuous process data to build a control chart that is based on the principal components of the data. The default dimension of the control chart is based on the number of principal components that account for 85% of the process variation. This number is based on the cumulative percent of the principal component eigenvalues.
Principal Components
Use principal component columns that were previously saved from a principal component analysis (PCA). The default dimension of the control chart is the number of components specified as process variables.
Partial Least Squares Score Data
Use score columns that were previously saved from a partial least squares (PLS) analysis to build a control chart that is based on the score columns. The default dimension of the control chart is the number of scores specified as process variables.
Notes:
• PCA or PLS models built with a frequency or weight column are not supported.
• PCA or PLS models built with historical data must use the same number of historical data rows as specified in the MDMVCC launch window.
• PCA models built from within the Multivariate platform are not supported.