Batch level predictionset, including workset batch with excluded observation, may be incorrect (Q923)

The information in this article applies to: 
SIMCA 16
SIMCA 15
Summary: 

When the batch level predictionset includes a workset batch that has excluded observations in the batch evolution model, the predictionset, and consequently predictions, may be incorrect.

Symptoms: 

The predictionset is different for the predictionset batches not in the batch evolution model, depending on wether workset batches are included or not.

Workaround: 

Build your predictionset only from batches not in the evolution model. Alternatively, if you want to include workset batches, include only workset batches without excluded observations in the evolution model in your predictionset. 

More Information: 

When the predictionset only includes batches not included in the batch evolution model, it is always correct.

Date modified: May 21, 2019
Article type: 
Problem