Data mining and Process Analytics systems are powerful tools which can aid you in understanding the causes of process problems such as reduced yield and poor quality. They use advanced algorithms such as Principal Component Analysis to “mine” plant data. An example is the MS2 system which is now in use in over 15 sites. To help you get the best out of these complex algorithms and the wealth of tools available in MS2, training is essential and in response to requests from users, we are now able to provide this. A greater understanding of the general principles of data mining will also be included; the course will also be useful for people who are thinking about starting to use data mining techniques.. The course will cover the essential aspects of the MS2 system.

AGENDA

• Introduction to Process Analytics and Data Mining
• Importing Data - Single and multiple file import
• Data Pre-screening - Missing Data, Interpolation, Bad sample isolation, non-numeric data
• Univariate Analysis - Trend plots, CUSUM charts, Shewart charts, Distribution plots
• Parallel Coordinate Visualisation
  
• Principal Component Analysis - Eigenvalues and Eigenvectors, Scree plots, contributions, loadings, score plots, data cluster analysis, Hotelling's T2, Q statistic (Squared Prediction Error), Mulitway PCA
• Practical Issues - Recognising bad data, choosing the right algorithm
• Some Examples