turn on suggestions

Auto-suggest helps you quickly narrow down your search results by suggesting possible matches as you type.

Showing results for

- JMP User Community
- :
- Discussions
- :
- Why is RSquare nothing? - Summary of Fit of Model

Topic Options

- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Aug 30, 2017 6:46 PM
(836 views)

Why is RSquare not being calculated for this model?

Here is the script the model gave me. Why does it show "Mixture" instead of the actual components?

```
Fit Model(
Y( :FW5% ),
Effects(
:FH & Mixture,
:TotFlow & Mixture,
:Pch & Mixture,
:ExFlow & Mixture,
:FH * :TotFlow,
:FH * :Pch,
:TotFlow * :Pch,
:FH * :ExFlow,
:TotFlow * :ExFlow,
:FH * :TotFlow * :ExFlow,
:TotFlow * :Pch * :ExFlow,
:ExFlow * :ExFlow,
:ExFlow * :ExFlow * :ExFlow,
:FH * :TotFlow * :Pch * :ExFlow,
:Pch * :ExFlow,
:FH * :TotFlow * :Pch,
Scheffe Cubic( :FH, :ExFlow ),
Scheffe Cubic( :TotFlow, :ExFlow ),
:FH * :Pch * :ExFlow
),
No Intercept,
Center Polynomials( 0 ),
Personality( "Standard Least Squares" ),
History(
Effects(
:FH & Mixture,
:TotFlow & Mixture,
:Pch & Mixture,
:ExFlow & Mixture,
:FH * :TotFlow,
:FH * :Pch,
:TotFlow * :Pch,
:FH * :ExFlow,
:TotFlow * :ExFlow,
:FH * :TotFlow * :ExFlow,
:TotFlow * :Pch * :ExFlow,
:ExFlow * :ExFlow,
:FH * :FH,
:ExFlow * :ExFlow * :ExFlow,
:FH * :TotFlow * :Pch * :ExFlow,
:Pch * :ExFlow,
Scheffe Cubic( :FH, :TotFlow ),
Scheffe Cubic( :FH, :Pch ),
:FH * :TotFlow * :Pch,
Scheffe Cubic( :TotFlow, :Pch ),
Scheffe Cubic( :FH, :ExFlow ),
Scheffe Cubic( :TotFlow, :ExFlow ),
:FH * :Pch * :ExFlow,
Scheffe Cubic( :Pch, :ExFlow )
),
Effects(
:FH,
:TotFlow,
:Pch,
:ExFlow,
:FH * :TotFlow,
:FH * :Pch,
:TotFlow * :Pch,
:FH * :ExFlow,
:TotFlow * :ExFlow,
:FH * :TotFlow * :ExFlow,
:TotFlow * :Pch * :ExFlow,
:ExFlow * :ExFlow,
:FH * :FH,
:ExFlow * :ExFlow * :ExFlow,
:FH * :TotFlow * :Pch * :ExFlow
),
Effects(
:FH,
:TotFlow,
:Pch,
:ExFlow,
:FH * :TotFlow,
:FH * :Pch,
:TotFlow * :Pch,
:FH * :ExFlow,
:TotFlow * :ExFlow,
:FH * :TotFlow * :ExFlow,
:TotFlow * :Pch * :ExFlow,
:ExFlow * :ExFlow,
:FH * :FH,
:ExFlow * :ExFlow * :ExFlow
),
Effects(
:FH,
:TotFlow,
:Pch,
:ExFlow,
:FH * :TotFlow,
:FH * :Pch,
:TotFlow * :Pch,
:FH * :ExFlow,
:TotFlow * :ExFlow,
:FH * :TotFlow * :ExFlow,
:FH * :Pch * :ExFlow,
:TotFlow * :Pch * :ExFlow,
:ExFlow * :ExFlow,
:FH * :FH,
:ExFlow * :ExFlow * :ExFlow
),
Effects(
:FH,
:TotFlow,
:Pch,
:ExFlow,
:FH * :TotFlow,
:FH * :Pch,
:TotFlow * :Pch,
:FH * :ExFlow,
:TotFlow * :ExFlow,
:FH * :TotFlow * :ExFlow,
:Pch * :ExFlow,
:FH * :Pch * :ExFlow,
:TotFlow * :Pch * :ExFlow,
:ExFlow * :ExFlow,
:FH * :FH,
:ExFlow * :ExFlow * :ExFlow
),
Effects(
:FH,
:TotFlow,
:Pch,
:ExFlow,
:FH * :TotFlow,
:FH * :Pch,
:TotFlow * :Pch,
:FH * :ExFlow,
:TotFlow * :ExFlow,
:FH * :TotFlow * :ExFlow,
:Pch * :ExFlow,
:FH * :Pch * :ExFlow,
:TotFlow * :Pch * :ExFlow,
:ExFlow * :ExFlow,
:FH * :FH
),
Effects(
:FH,
:TotFlow,
:Pch,
:ExFlow,
:FH * :TotFlow,
:FH * :Pch,
:TotFlow * :Pch,
:FH * :ExFlow,
:TotFlow * :ExFlow,
:FH * :TotFlow * :ExFlow,
:Pch * :ExFlow,
:FH * :Pch * :ExFlow,
:TotFlow * :Pch * :ExFlow,
:ExFlow * :ExFlow,
:FH * :FH,
:FH * :TotFlow * :Pch * :ExFlow
),
Effects(
:FH,
:TotFlow,
:Pch,
:ExFlow,
:FH * :TotFlow,
:FH * :Pch,
:TotFlow * :Pch,
:FH * :ExFlow,
:TotFlow * :ExFlow,
:FH * :TotFlow * :ExFlow,
:Pch * :ExFlow,
:FH * :Pch * :ExFlow,
:TotFlow * :Pch * :ExFlow,
:ExFlow * :ExFlow,
:FH * :FH
),
Effects(
:FH,
:TotFlow,
:Pch,
:ExFlow,
:FH * :TotFlow,
:FH * :Pch,
:TotFlow * :Pch,
:FH * :TotFlow * :Pch,
:FH * :ExFlow,
:TotFlow * :ExFlow,
:FH * :TotFlow * :ExFlow,
:Pch * :ExFlow,
:FH * :Pch * :ExFlow,
:TotFlow * :Pch * :ExFlow,
:FH * :TotFlow * :Pch * :ExFlow,
:ExFlow * :ExFlow,
:Pch * :Pch,
:FH * :FH
)
),
Emphasis( "Effect Leverage" ),
Run(
:FW5% << {Plot Actual by Predicted( 1 ), Plot Regression( 0 ),
Plot Residual by Predicted( 1 ), Plot Effect Leverage( 1 )}
),
SendToReport(
Dispatch(
{"Response FW5%", "FH", "Leverage Plot"},
"1",
ScaleBox,
{Format( "Fixed Dec", 12, 1 ), Min( -1.22802306518175 ),
Max( 1.2593984962406 ), Inc( 0.5 ), Minor Ticks( 1 )}
),
Dispatch(
{"Response FW5%", "FH", "Leverage Plot"},
"2",
ScaleBox,
{Min( -20 ), Max( 19 ), Inc( 5 ), Minor Ticks( 1 )}
),
Dispatch(
{"Response FW5%", "TotFlow", "Leverage Plot"},
"1",
ScaleBox,
{Min( -2.5 ), Max( 2.43421052631579 ), Inc( 1 ), Minor Ticks( 1 )}
),
Dispatch(
{"Response FW5%", "TotFlow", "Leverage Plot"},
"2",
ScaleBox,
{Min( -20 ), Max( 38.5 ), Inc( 5 ), Minor Ticks( 1 )}
),
Dispatch(
{"Response FW5%", "Pch", "Leverage Plot"},
"1",
ScaleBox,
{Min( -20 ), Max( 16.0902255639098 ), Inc( 5 ), Minor Ticks( 1 )}
),
Dispatch(
{"Response FW5%", "Pch", "Leverage Plot"},
"2",
ScaleBox,
{Min( -20 ), Max( 25.5 ), Inc( 5 ), Minor Ticks( 1 )}
),
Dispatch(
{"Response FW5%", "ExFlow", "Leverage Plot"},
"1",
ScaleBox,
{Min( -12.6872632709593 ), Max( 12.218045112782 ), Inc( 2 ),
Minor Ticks( 1 )}
),
Dispatch(
{"Response FW5%", "ExFlow", "Leverage Plot"},
"2",
ScaleBox,
{Min( -20 ), Max( 28.1 ), Inc( 5 ), Minor Ticks( 1 )}
),
Dispatch(
{"Response FW5%", "FH*TotFlow", "Leverage Plot"},
"2",
ScaleBox,
{Min( -20 ), Max( 34.6 ), Inc( 5 ), Minor Ticks( 1 )}
)
)
)
```

1 ACCEPTED SOLUTION

Accepted Solutions

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Aug 31, 2017 7:38 AM
(1549 views)

Solution

To add to my colleague's post, and dive a little deeper, are the factors truly mixture factors? In other words, do they add up to a constant value (usually 1)? If not, that could also be causing your lack of an RSquare value. By having the "&Mixture" attribute turned on, JMP is expecting those factors to add up to a constant. That is why the "no-intercept" box is checked. However, the model would be a Scheffe model, not truly a no-intercept model as Mark correctly points out (the intercept is included with the main effects in a Scheffe model). But if those factors do NOT add up to a constant, then JMP reverts to the no-intercept model and is likely the cause of your RSquare value missing.

Dan Obermiller

2 REPLIES

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Aug 31, 2017 6:37 AM
(807 views)

R square is generally the ratio of the *model sum of squares* to the *corrected total sum of squares*. The corrected total sum of squares is computed with the average response. This *intercept-only* or *reduced model* is independent of any factor effects. The R square and *F*-ratio in the analysis of variance are used to compare the performance and significance, respectively, of the full model compared to the reduced model. The full model **must** contain the reduced model (all of its terms) for this comparison to be valid.

The effects of the mixture components in your full model *include the intercept*. It means that the effects of the mixture components and the intercept are inseparable. An explicit intercept term is not included. (Enforced by the No Intercept argument in your script.) The lack of an intercept means that the R square does not exist and any computation that supposedly results in R square is invalid.

Learn it once, use it forever!

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Aug 31, 2017 7:38 AM
(1550 views)

To add to my colleague's post, and dive a little deeper, are the factors truly mixture factors? In other words, do they add up to a constant value (usually 1)? If not, that could also be causing your lack of an RSquare value. By having the "&Mixture" attribute turned on, JMP is expecting those factors to add up to a constant. That is why the "no-intercept" box is checked. However, the model would be a Scheffe model, not truly a no-intercept model as Mark correctly points out (the intercept is included with the main effects in a Scheffe model). But if those factors do NOT add up to a constant, then JMP reverts to the no-intercept model and is likely the cause of your RSquare value missing.

Dan Obermiller