Options - Analyses/Graphs: Limits
In the Options
dialog box, select the Limits
tab (below Analyses/Graphs) to access options to customize
limits in Statistica. The specifications made on this tab will be used
as the default operating conditions whenever you open Statistica.
Number
of processors (cores) to use. Use these options to specify how
Statistica will take advantage of multiple processors/cores available
on your machine. In most cases, you should use the default maximum allowed,
but this value can be adjusted in special circumstances if necessary.
Note that for both options, you are always limited by the licensed number
of cores.
Default.
Select this option button to use the number of cores for which your hardware
is currently configured. This is the recommended default selection.
Limit to
_ cores. Select this option button and enter in the text box any
value 1 or greater. If you specify a value greater than you are licensed
for, a message will be displayed and the value changed to the license
limit. Note that this value ignores the number of cores actually on your
machine, so if you are licensed for more cores than you have on your machine,
you can "over-subscribe" and configure the threading to use
more threads than cores. Note that changing this value will dynamically
adjust the number of cores Statistica will use for multi-threaded operations
(i.e., no need to restart the program).
Computation
of percentiles. This drop-down list contains options from which
you can choose the method for computing percentile values, including medians (50th percentile)
and quartiles
(25th and 75th percentiles).
The method you select here will apply to every place in Statistica where
medians, quartiles, or percentiles are computed, for example, to the computation
of:
For more details on a specific
method, see Percentile Calculation Options in Statistica.
Automatically
close Graph menu dialogs after a graph is produced (do not treat these
graphs as Analyses to be continued after the initial output is produced).
Select this check box if you want the graph creation dialog boxes to automatically
close after you create a graph. If this check box is cleared, the graph
creation dialog box is minimized upon the creation of the graph.
Limit maximum memory (MB) for each Analysis.
In this box, specify the maximum amount of memory you want allocated to
a single analysis. If an analysis exceeds the allocated memory, an error
message is displayed. Note that this limit applies to a single analysis
only, not to graphs or Statistica as a whole. By default,
the maximum amount of memory allocated is 75% of the installed RAM on
your computer.
Reset. Click
this button to set the maximum memory limit for each analysis to the default,
75% of the installed RAM on your computer.
When
using non-incremental learning algorithms (e.g. data mining, GRM, etc.),
automatically apply sampling when the dataset is larger than _ MB.
Select this check box to enable Statistica to use systematic sampling
when running analyses (e.g., data mining, GRM, etc.) on data sets larger
than the amount entered in the MB
box in order to use system resources more efficiently.
Statistica will choose the sample using systematic
sampling methods, i.e., the value for k
is set automatically based on the size of the data set and the MB
limit. The size of the data file is divided by limit, and the resulting
value is rounded to the nearest integer and set as k. For example, if
the limit is set to 25 MB and
the data set being analyzed is 70 MB, Statistica will set k
equal to 3 since 70/25 = 2.8.
If systematic sampling is not the desired sampling
method, there are other sampling tools available to use. For more information,
see Create
a Random Sample.
When running analyses from SVB, display warnings dialog.
In many instances throughout interactive analyses, you may receive non-fatal
error (warning) messages, for example, to inform you that some missing
data were replaced by means, that some variables were dropped from the
analyses, that some parameters were reset, etc. Usually, these warnings
will not require that the respective analyses be terminated, but the information
provided in these warnings may be important for you to be aware of. When
such warning messages occur during the execution of a Statistica Visual Basic (SVB) program,
the Analysis
Warnings dialog box is displayed, which lists the warnings that
have occurred, when they occurred (the time and in which specific analysis),
the approximate line numbers in the SVB program code where the warnings
occurred, etc. For example, this log of errors may look like this:

When output to the report
window is specified (see Options - Output Manager tab), the warnings can be printed
as well:

The maximum number of such warnings that will
be retained for a single SVB program run can be configured via the respective
edit field (default is 10; if more than 10 errors occur, then the first
one will be "dropped" from the list of non-fatal errors that
can be displayed). If you do not want to display the Analysis
Warnings dialog box, clear this check box.
Analysis SVB
warnings history queue length. Specify here the maximum length of
the queue for storing non-fatal error and warning messages that occur
during the execution of SVB programs. The historically "oldest"
warnings will be deleted from the queue to make room for new nonfatal
errors and warnings once the number of such messages that occur during
a single SVB program execution exceeds the number specified for the queue
length.
Display warning when creating a graph with data larger than
the data size threshold. In most cases, the graphs produced
in Statistica will contain the data from which they were computed. For
example, when making a histogram for a variable, the actual raw data for
that variable will be stored along with all other information for the
graph; when you save such a graph to a file and later reopen it, you can
still change the method of categorization
or other aspects of the graph that require reprocessing of the raw data.
See also Size
of Statistica Graph files. However, for very large data sets this
may not be desirable, because the files (graph objects) can become very
large (because they contain the raw data); in those cases only storing
the aggregated data (e.g., category counts for histograms) along with
the graph will yield significantly smaller file sizes when you save the
respective graph or when you manage many graphs of large data sets inside
a workbook.
Use this check box to select or clear the option
that the Large Data Warning
dialog box will be displayed, and to specify the minimum number of data
points for a single graph that should trigger this dialog box (cause it
to be displayed). The Large
Data Warning dialog box provides options to either attach the
(large number of) raw data points to the graph, or to detach the graph
from the data; in the latter case, various options for modifying the graph
later will not be available, if they require reprocessing of the raw data.
Note that this option is only relevant to graphs
that do not display the actual raw data points themselves, such as histograms
or pie charts. For other graphs, such as scatterplots, probability plots,
etc., which show as many points as there are individual observations,
the data will always be attached to (stored with) the graph (since otherwise,
the graph could not be plotted).
Data size threshold
( in cells) before displaying the "Large Data Warning" dialog.
Specify here the minimum number of observations for a single graph that
should trigger the Large Data Warning
dialog box. The raw observations will be stored along with the graph for
all graphs computed from a number of observations that is smaller than
this number. See also Size
of Statistica Graph Files.
Display warning about the status of version 5 graph files.
Importing Version 5 Statistica Graphs normally will display a warning
message that the graphs may not look identical after importing into the
current version of Statistica. Clear this check box to suppress the warning
message.
Expand
Histogram binning beyond min/max data values. When creating a histogram
for continuous data, the program has several options for determining how
the categories (binning) will be created. When using the Integer/Auto
option, the program will try and break the histogram binning into neat
intervals based on the range of data. Select this check box to affect
the way that this binning is done. When this check box is selected, a
small value is added to the min/max data values, after the neat intervals
are determined, to ensure that the upper or lower binning boundary will
not fall on the min/max data value. In some cases (commonly when the data
contains zeros), backing off this small amount would produce confusing
graphs. When the check box is cleared, then this small amount is not added
to the min/max data values, and the resulting graphs could have the min/max
data on the upper or lower binning boundaries, if the min/max values end
up being a neat interval.
New
Analyses use Spreadsheet selection conditions by default. Select
this check box to specify that all analyses use the selection conditions
from the spreadsheet (specified via the Spreadsheet Case Selection
Conditions dialog box).
New
Graphs use Spreadsheet selection conditions by default. Select
this check box to specify that all graphs use the selection conditions
from the spreadsheet (specified via the Spreadsheet Case Selection
Conditions dialog box).
Marker
point reduction threshold. Use the options in this box to specify
the thresholds (i.e., the minimum number of data points that must be present
in the graph) for the marker point reduction setting. By default, all
graphs which have fewer points than the Fast
threshold (e.g., by default, less than 10,000 points) are initially created
using the Standard marker point
reduction setting. Note that you can modify the marker point reduction
setting (after graph creation) in the Graphs Options dialog
box. For more information on the Marker
point reduction setting, see, for example, Graph
Layouts (2D Graphs).
Fast.
Enter the lower boundary for the Fast
setting. Whenever the number of points in the data set is between the
value entered here and the value entered for the Aggressive
setting, Statistica will use the Fast
marker point reduction setting to create the initial graph. By default
this value is set at 10,000.
Aggressive.
Enter the lower boundary for the Aggressive
setting. Whenever the number of points in the data set exceeds this value,
Statistica will use the Aggressive
marker point reduction setting to create the initial graph. By default
this value is set at 1,000,000.