Design of Experiments Wizard
The Experimental Design section of
STATGRAPHICS contains a new wizard that assists
users in constructing and analyzing designed
experiments. It guides the user through twelve
important steps in the creation of the design:
Step 1: Define responses
Step 2: Define exp. factors
Step 3: Select design
Step 4: Select model
Step 5: Select runs
Step 6: Evaluate design
Step 7: Save experiment
Step 8: Analyze data
Step 9: Optimize responses
Step 10: Save results
Step 11: Augment design
Step 12: Extrapolate
It also includes:
1. New diagnostic plots
such as Prediction Variance Plots, Prediction
Profiles, Variance Dispersion Plots, and
Fraction of Design Space plots.
2. The ability to include
both process and mixture variables in a single
design.
3. Integrated facilities
for multiple response optimization based on
desirability functions.
4. An option for
interactively exploring multiple dimensions
using either response surface or contour plots.
5. Easier extrapolation of
fitted models, including path of steepest ascent
identification.
6. New facilities for
creating robust parameter design using either
crossed factors or Montgomery's combined
approach.

3D Contour Plot
3D Mesh Plot
Fraction of Design Space Plot

Monte Carlo
Simulation Three new procedures are
available for performing Monte Carlo
simulations:
1. General Simulation Models
2. Random Number Generation
3. Simulation of ARIMA Time Series
The General Simulation Models procedure lets
you create a response variable Y that depends on
several input variables X. Probability
distributions are then specified for each X.
During the simulation, random values are
generated for each X and the resulting
distribution of Y is obtained. 

Parallel
Coordinate Plots and Andrews Plots These
plots are used to visualize multivariate data on
a casebycase basis. Each line connects the
values of multiple variables in a single row.
Positive and negative correlations between the
variables are evident by the parallel or
intersecting nature of the lines. By color
coding the lines, relationships with a
categorical variable can often be observed.


Star Glyphs and
Chernoff Faces Glyphs are graphic images
in which different features are scaled according
to the value of quantitative variables. They are
useful for identifying clusters of similar
multivariate observations and outliers. 

Sequential Sampling
Sequential probability ratio tests can be
constructed for testing the mean or standard
deviation of a normal distribution, the
probability parameter of a binomial
distribution, or the rate parameter of a Poisson
distribution. Unlike tests with a predetermined
sample size, sequential tests obtain samples one
at a time. After each sample is obtained, one of
three decisions is made: accept the null
hypothesis, reject the null hypothesis, or
continue sampling. In many cases, a decision
will be reached more quickly than from a fixed
size hypothesis test. In addition to the test,
operating characteristic (OC) and average sample
number (ASN) curves are provided.


Correspondence Analysis
This procedure creates a map of the rows and
columns in a twoway contingency table for the
purpose of providing insights into the
relationships amongst the categories of the row
and/or column variables. Often, no more than two
or three dimensions are needed to display most
of the variability or “inertia” in the table. An
important part of the output is a correspondence
map on which the distance between two categories
is a measure of their similarity. 

Multiple Correspondence
Analysis
This procedure
creates a map of the associations among
categories of two or more variables. It
generates a map similar to that of the
Correspondence Analysis procedure. However,
unlike that procedure which compares categories
of each variable separately, this procedure is
concerned with interrelationships amongst the
variables.


Reliability of Repairable
Systems Two procedures
have been added to analyze data
consisting of failure times from systems that
can be repaired, one for continuous times and
one for interval data. It is assumed that when
the system fails, it is immediately repaired and
placed in service again. The goal of the
analysis is to develop a model that can be used
to estimate failure rates or quantities such as
the MTBF (mean time between failures).


OneDimensional Point Process
Models Constructs models for events
occurring in one dimension (usually time or
space). Includes homogeneous and nonhomogeneous
Poisson process and renewal models with various
interevent time distributions.


Frequency Tables
This procedure analyzes a single column
containing previously tabulated counts. It
displays the counts using either a barchart or
piechart. Statistical tests may also be
performed to determine whether the data conform
to a set of multinomial probabilities. 

Sampling Distributions
The Sampling Distributions procedure
calculates tail areas and critical values for
four common sampling distributions. It also
plots the calculated results. 

Dashboard Gage Create
gages to display the status of critical quality
statistics. 

Statistical Tolerance Limits
A new procedure allows you to calculate
statistical tolerance limits from a sample of
data based on:
1. A normal distribution
2. A lognormal distribution
3. A Weibull distribution
4. A normal distribution after transforming the
data
5. A nonparametric approach
Twosided limits and onesided bounds are
available. 

Sample Size Determination for
Capability Indices The program will now
determine required sample sizes for estimating
common capability indices.


Sample Size Determination for
Statistical Tolerance Limits The program
will also determine required sample sizes for
creating statistical tolerance limits.

