Implemented Updates Describe > Two Variable Summarise #9074

Vitalis95 · 2024-07-16T14:02:04Z

Fixes partly #4952
@rdstern , have a look at the progress made

rdstern · 2024-07-19T15:18:19Z

@Vitalis95 that's looking very promising. Here is the option for 2 variables when Numeric by Numeric

a) In the 3 variable case, with the first and second numeric could you add the Means and Sig Level as well and also the Swap and the misssing values option.
b) @lilyclements may want to comment on this point. I suggest, when we have Means (both 2 variable and 3 Variable) we should change the word Means to Model. The Model is the means when a variable is a factor, and it is the Slope (or Slopes) for a Variate. So that's what we should be giving there?
c) In 3 variable with ANOVA could there be another option, with now some radio buttons. There will be more, but this one to start with is Interaction. If chosen then it gives the ANOVA with the interaction term.
d) And a third radio button could we have Compare as the label. This would not have the option of means. It would include sig level and the swap. I forget the command, but it would give the ANOVA line with just the second variable, then the third and then the intereaction. So one could find the decrease in the residual when we add the third variable (so the second in the x of the model, and then the extra when we also add the interaction term.

Vitalis95 · 2024-07-30T12:00:33Z

@Vitalis95 that's looking very promising. Here is the option for 2 variables when Numeric by Numeric

a) In the 3 variable case, with the first and second numeric could you add the Means and Sig Level as well and also the Swap and the misssing values option. b) @lilyclements may want to comment on this point. I suggest, when we have Means (both 2 variable and 3 Variable) we should change the word Means to Model. The Model is the means when a variable is a factor, and it is the Slope (or Slopes) for a Variate. So that's what we should be giving there? c) In 3 variable with ANOVA could there be another option, with now some radio buttons. There will be more, but this one to start with is Interaction. If chosen then it gives the ANOVA with the interaction term. d) And a third radio button could we have Compare as the label. This would not have the option of means. It would include sig level and the swap. I forget the command, but it would give the ANOVA line with just the second variable, then the third and then the intereaction. So one could find the decrease in the residual when we add the third variable (so the second in the x of the model, and then the extra when we also add the interaction term.

@lilyclements , what is your comment on this? The interaction term is not in the $anova_tables2 function

rdstern · 2024-09-23T13:06:47Z

@lilyclements and @Vitalis95 I was delaying on this dialog. But it has now become urgen, because the merged version of this dialog, doesn't do the ANOVA at all. It gives a bug. The 2-way ANOVA is working on this dialog, so let's merge it as soon as we can.
If @lilyclements has time this week to improve the function, then let's do this too? If so, then here are the 2 items it would be good to add for the 2-variable ANOVA options. Lily may have more?
a) Add a Total checkbox, default unchecked.
b) Change the Means option to Means/Model. Lily, currently it gives the means whatever the x. Here is an example:

When x is numeric it should give the intercept and slope.

In this example the means are even more ridiculous if you swap the variables.

c) The categorical by numeric should give summaries. This is fine in the current version, but has gone (wrongly) to yet more ANOVA tables in this pull request.

Here is the running version:

If we can get these 2-variable options working well, then either attack the 3 variable stuff quickly as well - ideal, or disable the 3 variables and merge this. Then attack the 3 variable case.

lilyclements · 2024-09-23T16:08:19Z

@rdstern thanks for this - I've fixed (a) and (b) in this PR #9152

rdstern · 2024-09-23T19:16:16Z

@Vitalis95 @lilyclements has now added trhe option for the Total and changed the Means into Means/Model. That's great.
On layout (you may have better ideas. I wonder about 2 rows of these options now
Checkbox Total instead of Means and then Sig Level as now.
Then the Means Checkbox on the line below, by itself, and now written as Means/Model. I assume that it will give either the means or the model, automatically, depending on whether the x is numeric or a factor.

update

Vitalis95 · 2024-09-24T08:31:14Z

@rdstern about c), you had suggested earlier the the Swarp First/Second variables checkbox be visible, when checked it gives ANOVA, when unchecked it gives summaries that is when we have categorical by numeric

rdstern · 2024-09-24T08:39:09Z

@Vitalis95 that's right. Did I get that wrong in my checking? That would be a relief. So now, does categorical by numeric still give summary tables? And when you swap it gives ANOVA? If so, that's great and please disregard c).

Update data_object_R6.R

Vitalis95 · 2024-09-24T08:58:15Z

@Vitalis95 @lilyclements has now added trhe option for the Total and changed the Means into Means/Model. That's great. On layout (you may have better ideas. I wonder about 2 rows of these options now Checkbox Total instead of Means and then Sig Level as now. Then the Means Checkbox on the line below, by itself, and now written as Means/Model. I assume that it will give either the means or the model, automatically, depending on whether the x is numeric or a factor.

@lilyclements , following this comment, could the total be passed like the sign_level and mean arguments, such that when the Total checkbox is checked it produces the totals

rdstern · 2024-09-24T09:02:39Z

@Vitalis95 that's exactly what I was hoping for with the Total line.

lilyclements · 2024-09-24T09:38:29Z

instat/static/InstatObject/R/data_object_R6.R

@@ -4524,8 +4524,7 @@ DataSheet$set("public", "has_labels", function(col_names) {
 }
 )

-DataSheet$set("public", "anova_tables2", function(x_col_names, y_col_name, signif.stars = FALSE, sign_level = FALSE, means = FALSE) {
-
+DataSheet$set("public", "anova_tables2", function(x_col_names, y_col_name, total = TRUE, signif.stars = FALSE, sign_level = FALSE, means = FALSE) {


@Vitalis95 apologies. I have now added total as a parameter> I've set it to be TRUE by default but if you want it as FALSE by default then set this line here to be total = FALSE instead of total = TRUE

Make sure you pull changes into your branch to avoid potential conflicts too.

Vitalis95 · 2024-09-24T12:12:02Z

@lilyclements , I pulled the changes and now there is an error,

…mmaries Three var summaries

lilyclements · 2024-09-24T17:19:44Z

@Vitalis95 thanks for this - yes, this is a great point. I forgot to update anova_tables2 in the instat_object after updating in the data_object file. This is done here and ready for you to merge into this branch if you are happy with it.

Anova changes

rdstern · 2024-09-25T08:27:33Z

@Vitalis95 and @lilyclements this is looking great!
a) Vitalis I remember Lily mentioned that the default was to give the total, but you could change that, in her function. Pleasec could you do that. Currently it gives the total whatever I do. The default is not to give the total, because that's what R does.
b) Please could you move Sig Level up, so it is after the total. (They are both concerned with the anova table, so they come above the Means/Model.
c) @lilyclements everything is there now, but can the output be improved? I hope that's not too time consuming. Could Francis help if needed? Here is an example:

It isn't a Means table of yield. I presume we could say just Model?
Could there be blanks instead of NA?
Could the F probability be formatted better?
And do we need to know it is a tibble and have all the [[1]] and list the type of each variable?

Vitalis95 · 2024-09-25T08:48:02Z

@rdstern , let me include the above comments then you will check

rdstern · 2024-09-25T14:51:36Z

@Vitalis95 looks really good now. I'll just wait to see if @lilyclements makes any changes?

lilyclements

@Vitalis95 looks good to me too!

lilyclements · 2024-10-09T09:45:10Z

@Vitalis95 @rdstern apologies didn't mean to approve. I will make some changes to the R code

Removing lists to remove the square brackets, and adding cat to print

lilyclements · 2024-10-09T12:36:30Z

@rdstern I made the changes you requested. You can now test it

rdstern

@lilyclements you now give the means twice, and not the ANOVA table!
If possible could you, or @Vitalis95 also improve the spacing of the controls, when there are both numeric variables

lilyclements · 2024-10-09T17:27:01Z

@rdstern so I did, apologies. I've fixed that now (hopefully!). Very sorry for that

rdstern

@lilyclements that's looking much better now.

a) Great that the F-prob is like that now. Thanks
b) Could you swap the order back, so the ANOVA comes before the means or the model? I see the ANOVA as the basic result, optionally followed by the model or means when needed.
c) Can you get rid of the extra tibble details, and the [[1]], etc? If that's too big to do now, then we could approve with these, and then improve the presentation, when we add the 3-variable stuff in a new pull request?

lilyclements

@Vitalis95 I've updated map to be walk for two classes/functions?

Are these classes functions only used for anova_table2 bits?

walk means that it will not print output, so will not give the list output!

@Vitalis95 are you happy with these changes? (Make sure you pull changes too next time you work on this - I've made changes now sorry!)

@rdstern does this now work for you how you would like it?

rdstern

@lilyclements I was not sure what has changed. There is still the tibble-like output.

I am approving, because it is now so much better than the vurrent merged one. We will add another pull request for the 3 variable case, and that also gives another opportunity to improve the formatting of this output.

@N-thony over to you to check, and hopefully merge.

N-thony

@Vitalis95 I made a few comments. I'm working now with Chris to have our first 64 bits only version. If you can make the change quickly, we can get this included in this version.

N-thony · 2024-10-11T08:03:40Z

instat/dlgDescribeTwoVariable.vb

+                'ElseIf IsFactorByFactorByNumeric Then
+                '    ucrReorderSummary.Visible = True
+                '    cmdSummaries.Visible = True
+                '    ucrSaveTable.Visible = True
+                '    ucrChkSummariesRowCol.Visible = True
+                '    ucrSaveTable.Location = New Point(23, 440)
+                '    ucrBase.clsRsyntax.SetBaseROperator(clsJoiningPipeOperator)
+                '    ucrSaveTable.SetPrefix("summary_table")
+                '    ucrSaveTable.SetSaveType(RObjectTypeLabel.Table, strRObjectFormat:=RObjectFormat.Html)
+                '    ucrSaveTable.SetAssignToIfUncheckedValue("last_table")
+                '    ucrSaveTable.SetCheckBoxText("Save Table")
+                '    clsJoiningPipeOperator.SetAssignToOutputObject(strRObjectToAssignTo:="last_table",
+                '                                  strRObjectTypeLabelToAssignTo:=RObjectTypeLabel.Table,
+                '                                      strRObjectFormatToAssignTo:=RObjectFormat.Html,
+                '                                      strRDataFrameNameToAddObjectTo:=ucrSelectorDescribeTwoVar.strCurrentDataFrame,
+                '                                        strObjectName:="last_table")


N-thony · 2024-10-11T08:03:57Z

instat/dlgDescribeTwoVariable.vb

+            'ucrChkDisplayMargins.Visible = IsFactorByFactorByFactor()
+            'ucrInputMarginName.Visible = ucrChkDisplayMargins.Checked AndAlso IsFactorByFactorByFactor()


N-thony · 2024-10-11T08:04:09Z

instat/dlgDescribeTwoVariable.vb

+            'ElseIf rdoThreeVariable.Checked Then
+            '    ucrChkOmitMissing.Visible = IsFactorByNumericByNumeric() OrElse IsNumericByNumericByFactor()
+            'Else
+            '    ucrChkOmitMissing.Visible = False
+            '    cmdMissingOptions.Visible = False


N-thony · 2024-10-11T08:04:58Z

instat/dlgDescribeTwoVariable.vb

+                'If IsNumericByNumericByFactor() Then
+                '    '    If ucrChkSwapXYVar.Checked Then
+                '    '        clsCombineSwapAnova2Table.AddParameter("y", ucrReceiverThreeVariableThirdVariable.GetVariableNames(True), iPosition:=2, bIncludeArgumentName:=False)
+                '    '    Else
+                '    '        clsCombineAnova2Function.AddParameter("y", ucrReceiverThreeVariableThirdVariable.GetVariableNames(True), iPosition:=2, bIncludeArgumentName:=False)
+
+                '    '    End If


Vitalis95 · 2024-10-11T10:55:24Z

@N-thony , have a look at it

N-thony

@rdstern over to you

rdstern

@N-thony I am approving, as this is an important advance. It will be good for it to be merged.
@Vitalis95 please can you make a new pull request for the Three Variable part. That will be even more exciting. And remind @lilyclements about the R bit, for parallel and full, as well as making the result neater for all the ANOVA.

Vitalis95 added 6 commits July 16, 2024 16:56

changes to the code

22b399d

changes to the code

01fe3fb

changes

107bcca

changes

41e4b5e

changes to the code

4f42e53

changes to the code

a18448c

changes

91c887b

changes

8f5f841

Update data_object_R6.R

baaffa5

lilyclements mentioned this pull request Sep 23, 2024

Update ANOVA function #9152

Closed

Vitalis95 and others added 3 commits September 24, 2024 10:01

Merge pull request #279 from IDEMSInternational/master

af4a7d8

update

changes

a85d005

changes

20c187c

Merge pull request #280 from lilyclements/anova_changes

6cfface

Update data_object_R6.R

Adding total as parameter

962c0b2

lilyclements reviewed Sep 24, 2024

View reviewed changes

changes

739793d

lilyclements added 2 commits September 24, 2024 18:17

Merge pull request IDEMSInternational#958 from Vitalis95/three_var_su…

62e5f8f

…mmaries Three var summaries

Update instat_object_R6 to have total = TRUE

05f7ec7

Merge pull request #281 from lilyclements/anova_changes

5e122d0

Anova changes

Changes to the code

681034a

Vitalis95 marked this pull request as ready for review September 25, 2024 08:43

changes

8573f78

rdstern mentioned this pull request Oct 8, 2024

Error in producing an ANOVA table #9180

Open

lilyclements previously approved these changes Oct 9, 2024

View reviewed changes

Tidying up anova_tables2 format

cc91a97

lilyclements dismissed their stale review via cc91a97 October 9, 2024 10:12

Update data_object_R6.R

a447e91

Removing lists to remove the square brackets, and adding cat to print

rdstern reviewed Oct 9, 2024

View reviewed changes

actually returning the model!

9ecd1d6

rdstern reviewed Oct 9, 2024

View reviewed changes

lilyclements added 2 commits October 9, 2024 19:56

Amending anova_tables2 to print

c2bebb2

Update dlgDescribeTwoVariable.vb

d3a4131

lilyclements reviewed Oct 9, 2024

View reviewed changes

rdstern previously approved these changes Oct 9, 2024

View reviewed changes

N-thony requested changes Oct 11, 2024

View reviewed changes

changes

1c600be

Vitalis95 dismissed rdstern’s stale review via 1c600be October 11, 2024 10:54

N-thony approved these changes Oct 11, 2024

View reviewed changes

rdstern changed the title ~~Implemented Describe > Three Variable Summarise~~ Implemented Updates Describe > Two Variable Summarise Oct 13, 2024

rdstern approved these changes Oct 13, 2024

View reviewed changes

N-thony merged commit 243f37e into IDEMSInternational:master Oct 14, 2024
2 checks passed

		'ucrChkDisplayMargins.Visible = IsFactorByFactorByFactor()
		'ucrInputMarginName.Visible = ucrChkDisplayMargins.Checked AndAlso IsFactorByFactorByFactor()

Implemented Updates Describe > Two Variable Summarise #9074

Implemented Updates Describe > Two Variable Summarise #9074

Conversation

Vitalis95 commented Jul 16, 2024 • edited Loading

rdstern commented Jul 19, 2024

Vitalis95 commented Jul 30, 2024

rdstern commented Sep 23, 2024 • edited Loading

lilyclements commented Sep 23, 2024

rdstern commented Sep 23, 2024

Vitalis95 commented Sep 24, 2024

rdstern commented Sep 24, 2024

Vitalis95 commented Sep 24, 2024

rdstern commented Sep 24, 2024

Choose a reason for hiding this comment

Vitalis95 commented Sep 24, 2024

lilyclements commented Sep 24, 2024

rdstern commented Sep 25, 2024

Vitalis95 commented Sep 25, 2024 • edited Loading

rdstern commented Sep 25, 2024 • edited Loading

lilyclements left a comment

Choose a reason for hiding this comment

lilyclements commented Oct 9, 2024

lilyclements commented Oct 9, 2024

rdstern left a comment

Choose a reason for hiding this comment

lilyclements commented Oct 9, 2024

rdstern left a comment • edited Loading

Choose a reason for hiding this comment

lilyclements left a comment

Choose a reason for hiding this comment

rdstern left a comment

Choose a reason for hiding this comment

N-thony left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Vitalis95 commented Oct 11, 2024

N-thony left a comment

Choose a reason for hiding this comment

rdstern left a comment

Choose a reason for hiding this comment

Vitalis95 commented Jul 16, 2024 •

edited

Loading

rdstern commented Sep 23, 2024 •

edited

Loading

Vitalis95 commented Sep 25, 2024 •

edited

Loading

rdstern commented Sep 25, 2024 •

edited

Loading

rdstern left a comment •

edited

Loading