ocr: - 155 05 S U - 0.5 15 15 :0.5 0 0.5 1 15 2 PCI Figure 5: Clustering of model resolts using first wo principal components. Regions have different likelihoods of model correchness, suggosting principal components can be used to assign a relative confidence level to each day's predictions.
(2000).iso/pcaik/8b1/orouf105.gif?hex){kind=small}
(2000).iso/pcaik/8b1/orouf105.gif "Download"){kind=small}
(2000).iso/pcaik/8b1/orouf104.gif){kind=link}
(2000).iso/pcaik/8b1/sherf101.gif){kind=link}