
Known Issues in Release 2.0
Reference Topic/Area Issue Description / Workaround
EXT-2319 nzSpatial ST_PointOnSurface incorrectly returning a point with empty poly-
gons
EXT-2328 Bayesian Networks Bayesian Networks are not deterministic in the choice of VARX
and VARY
EXT-2332 KMEANS The number of numeric columns supported are limited to 55 in
this release when Mahalanobis distance is used for K-means
clustering.
SWS-62976,
EXT-915
Netezza system,
Netezza Analytics
A known issue has surfaced in recent testing regarding algo-
rithms calling the function rank() or over(). When these functions
are called on large data sets (10 billion rows if called once, or 5
billion rows if called twice), you will receive an error similar to,
"ERROR: SPU swap partition: Disk temporary work space is
full." This is due to the function operating on host where there is
not enough disk space to handle the operation.
To solve this issues properly, a Netezza system fix is needed
(SWS-62976). This is currently slated to be included in a patch
release for Netezza systems (date to be determined). Until the fix
is in place, you are limited to a data set less than 5 or 10 billion
rows as noted above.
The following Algorithms use either rank() or over():
rank()
Correlation and Mutual Information
Discretization
FP-Growth
General Diagnostic Measures
KNN
Moments
Non-Parametric Statistics
PCA added to the Analytics documentation
over()
Correlation and Mutual Information
Decision Trees
Discretization
Divisive Clustering
FP-Growth
General Diagnostic Measures
K-means
KNN
Moments
Non-Parametric Statistics
Regression Trees
Sufficient Statistics
25
Comentários a estes Manuais