Version 16.1 of STATGRAPHICS Centurion adds
several new features and additional statistical
procedures not contained in Version 16.0. These
include:
1. A new 64-bit edition that can access as
much RAM as is contained in your computer. While
both the 32-bit and 64-bit editions will run on
64-bit operating systems, the 64-bit edition can
handle larger data sets.
2. Four new statistical procedures for
multivariate visualization. These procedures let
you examine data involving many quantitative factors
on a case-by-case basis.
3. A new procedure for generating
statistical
tolerance limits based on normal, lognormal, and
Weibull distributions. Nonparametric limits may also
be calculated.
4. Three new procedures for
Monte Carlo simulation.
The main procedure lets you generate random samples
for relationships that are too complex to derive
analytically. The other two procedures create random
samples from 48 probability distributions and from
ARIMA time series models.
5. Two new sample size determination
procedures. One determines the necessary sample size
for estimating capability indices such as Cpk, while
the other determines the required sample size for
creating statistical tolerance limits.
6. Dynamic videos showing interaction with
graphs may now be recorded and saved in AVI files.
7. Complex SQL queries may be entered
manually rather than working through the Query
Wizard dialog boxes.
8. Slanted ticmarks may be selected for the
X-axis on any graph.
9. The order of factor levels may be changed
interactively in many procedures such as a barchart
or analysis of variance.
10. Enhancements to the DOE Wizard, including
the ability to optimize only selected responses.
HIGHLIGHTS
Monte
Carlo Simulation - When analyzing
complex systems, the relationships between an output
variable Y and various input variables X1, X2, X3,
..., may be too complex to derive theoretically. In
such cases, Monte Carlo simulation may be used to
determine the distributional properties of Y. A new
procedure in Version 16.1 lets you perform such a
simulation by:
-
Defining the relationship between the inputs and
outputs.
-
Specifying a probability distribution for each
input.
-
Generating repeated samples for each input and
calculating the value of the outputs.
This
methodology is an important component of Lean Six
Sigma. The dialog box below shows a typical example,
where the output variable is the total time to
complete 12 tasks:
Users interested in time series forecasting may also
appreciate the new procedure for generating samples
from ARIMA models:
Multivariate Visualization
- Four new procedures have been added for
visualizing multiple quantitative variables. These
include:
A
parallel coordinate plot connects the
standardized values of each variable with a separate
line for each row of data:
In an
Andrews plot, the values of each variable are
encoded using a trigonometric function:
Star glyphs draw polygons for each row in which
the distance to each vertex represents the value of
a specific variable:
Chernoff faces use various features of a cartoon
face to represent the value of each variable:
Sample
Size Determination - Two new procedures have been added for
determining required sample sizes. One
procedure determines how many samples are required
for precise estimation of capability indices:
The second procedure calculates the sample size
needed when calculating statistical tolerance
limits:
Statistical Tolerance Limits - This
procedure calculates statistical tolerance limits
and bounds for data sampled from selected
probability distributions:
Statistical tolerance limits are an important
technique for demonstrating that specification
limits for a manufacturing process are being met.
Video
Recording - Videos may now be created
demonstrating user interaction with STATGRAPHICS
graphs.
