Other Version 20 Enhancements

Numerous other enhancements have been made in Version 20. Many are listed on this page.

Design of Experiments - Doehlert Designs

Doehlert designs have been added to the catalog of experimental designs that may be created and analyzed in Statgraphics 20. These designs are a type of uniform space-filling design that may be used for response surface optimization. Compared to central composite and Box-Behnken designs, Doehlert designs are usually smaller. The predictive variance is also more uniform throughout a spherical region. However, the designs are not orthogonal.

Statgraphics 20 provides Doehlert designs ranging from a 7-run design for 2 factors through a 43-run design for 6 factors. The 7-run design is shown below:

Distribution Fitting and Weibull Analysis

The Distribution Fitting (Censored and Uncensored data) and the Weibull Analysis procedure have been modified to include confidence intervals for distribution parameters, percentiles, and tail areas. The intervals are available when fitting the normal, lognormal, logistic, loglogistic, exponential, smallest extreme value and Weibull distributions. Sample output is shown below:

Excel and Minitab Import

Statgraphics 20 adds the ability to import workbooks and project files consisting of multiple worksheets.  Previously, each worksheet needed to be imported separately. In addition, Statgraphics V20 can import any of the Minitab formats shown below: 

Date-Time Formats

New date-time formats have been added to specify data in terms of milliseconds: 

Operators

2 new operators have been added for use in Statgraphics expressions:

SORTASC(X,k) - sorts the data in column X and returns the first k (smallest) values.

SORTDESC(X,k) - sorts the data in column X and returns the first k (largest) values.

X may be either numeric or non-numeric.

Tabulation and Frequency Tables

A new Packed Bubble Chart has been added to the Tabulation and Frequency Tables procedures. It displays a bubble for each level of the tabulated variable. The areas of the bubbles are proportional to the frequencies of each class, which is also indicated by a color palette ranging from blue to red. 

Other new charts include a funnel chart, a pyramid chart, and a rectangle area chart.

Subset Analysis

Several enhancements have been made to the Subset Analysis procedure. The first is the addition of a dot diagram that shows the distribution of the data variable X with dots color coded according to the subset. Dots may be plotted at each unique value of X or collected into non-overlapping intervals as in a frequency histogram.

The second enhancement is a packed bubble chart containing a bubble for each subset of the data. The area of the bubble is proportional to the number of observations in that subset. The color of the bubble represents the value of a selected statistic for the subset.

The third enhancement is the ability to fit a selected distribution to the data in each subset. Any of 27 different probability distributions may be fit. Various goodness-of-fit statistics may be used to test whether the selected distribution adequately fits the data in each subset.


The fitted distributions for each subset may be displayed in a single plot.

Trellis Plots

Four new options have been added to the trellis plots for quantitative variables:

1.      Violin plots

2.      Dot diagrams

3.      Diamond plots

4.      Summary statistics