Essays, Posts & Presentations

New 12.1 Dataset Interactive Controls and Formatting Options

Christopher Carlson

Author

Christopher Carlson

Title

New 12.1 Dataset Interactive Controls and Formatting Options

Description

Big boost in functionality for Dataset in Wolfram Language 12.1 helps you communicate your data effectively and gain new insights. See interactive controls and examples of new functions.

New 12.1 Dataset Interactive Controls and Formatting Options

June 23, 2020

Christopher Carlson, Senior User Interface Developer, User Interfaces

In his blog post announcing the launch of Mathematica Version 12.1, Stephen Wolfram mentioned the extensive updates to

Dataset

that we undertook to make it easier to explore, understand and present your data. Here is how the updated Dataset works and how you can use it to gain deeper insight into your data.

New Interactive Features We have added items to Dataset column header context menus for sorting and reverse sorting your data:

If a Dataset has multiple levels of data, you can sort multiple columns simultaneously:

Sort row headers by hovering near the corner of the blank cell atop a row header column. When the menu indicator (

) appears, right-click it to bring up the context menu and choose a sort item:

Hide and Show items are also in the context menus of all Dataset cells, used to collapse parts of datasets for focused views of particular data:

In[]:=

Take[ExampleData[{"Dataset","Planets"}],5,{2,3}]

Out[]=

	Radius	Moons
	Radius		Mass	Radius
Mercury	2439.7km
Venus	6051.9km
Earth	6367.4447km	Moon	7.3459× 22 10 kg	1737.5km
Mars	3386.km	Phobos	1.072× 16 10 kg	11.1km
Mars	3386.km	Deimos	1.5× 15 10 kg	6.2km
Jupiter	69173.km	Metis	1.× 17 10 kg	21.5km
		Adrastea	7.× 15 10 kg	8.2km
		Amalthea	2.07× 18 10 kg	83.45km
		Thebe	1.5× 18 10 kg	49.3km
		63 total ›

All the Options Sorting and hiding give you interactive tools for exploring your data. With Dataset’s new formatting options, you can present your data in ways that make it easier to understand and spot patterns.

The following is a complete set of new Dataset options:

Alignment	Grid-like formatting for Dataset items
Background
ItemSize
ItemStyle
HeaderAlignment	Grid-like formatting for Dataset headers
HeaderBackground
HeaderSize
HeaderStyle
ItemDisplayFunction	complete control of item and header formatting
HeaderDisplayFunction	complete control of item and header formatting
HiddenItems	which items are initially hidden
MaxItems	maximum number of items to display without a scrollbar or elision
DatasetDisplayPanel	initial drill-down position
ScrollPosition	initial scroll positions

In the subsequent sections, I’ll explain the basic functions of these options and then do a deep dive into option value syntax. It lets you apply option values to Dataset data in tons of useful ways.

Alignment, Background, ItemSize, ItemStyle and Their Header Equivalents These options, familiar from

Grid

, now work in Dataset as well. Here is a dataset with default styling:

Here is the same Dataset with right-aligned ages, orange backgrounds and italic “children” entries (to change a Dataset’s options, wrap it with Dataset[...] and specify the new options):

In[]:=

Dataset

people data

,Alignment{"age"Right},BackgroundLightOrange,ItemStyle{"children"Italic}

Out[]=

	age	sex	children
				age	sex
Deb	62	female	Hal	29	male
			Kat	31	female
Eva	43	female
Bob	41	male	Bob	1	male
			Bri	3	female
			3 total ›
Ann	35	female	Amy	6	female
Cal	60	female

Each of the styling options has an analogous header option that operates on the Dataset’s headers rather than the items:

In[]:=

Dataset

people data

,Alignment{"age"Right},BackgroundLightOrange,ItemStyle{"children"Italic},HeaderAlignment{"age"Right},HeaderBackgroundLightRed,HeaderStyleBold

Out[]=

	age	sex	children
				age	sex
Deb	62	female	Hal	29	male
			Kat	31	female
Eva	43	female
Bob	41	male	Bob	1	male
			Bri	3	female
			3 total ›
Ann	35	female	Amy	6	female
Cal	60	female

ItemDisplayFunction, HeaderDisplayFunction If the basic styling options don’t meet your needs, you can take complete control of item and header formatting with the

ItemDisplayFunction

and

HeaderDisplayFunction

options.

Here is an item display function that replaces “male” and “female” with the symbols for male and female, and a header display function that changes the “sex” headers accordingly:

In[]:=

Dataset

people data

,ItemDisplayFunction{"sex"(If[#==="male",♂,♀]&)},HeaderDisplayFunction{"sex"("♂/♀"&)}

Out[]=

	age	♂/♀	children
				age	♂/♀
Deb	62	♀	Hal	29	♂
			Kat	31	♀
Eva	43	♀
Bob	41	♂	Bob	1	♂
			Bri	3	♀
			3 total ›
Ann	35	♀	Amy	6	♀
Cal	60	♀

The display function is given three arguments: the item or header value, the path to the item or header and the entire dataset itself. Here is a header display function that uses the second (path) argument to highlight children with the same name as their parent:

In[]:=

Dataset

people data

,HeaderDisplayFunction(If[MatchQ[#2,{x_,"children",x_}],Style[#,Bold,Red],#]&)

Out[]=

	age	sex	children
				age	sex
Deb	62	female	Hal	29	male
			Kat	31	female
Eva	43	female
Bob	41	male	Bob	1	male
			Bri	3	female
			3 total ›
Ann	35	female	Amy	6	female
Cal	60	female

HiddenItems Specify which Dataset items are initially hidden with the

HiddenItems

option:

In[]:=

Dataset

people data

,HiddenItems{"Eva","sex"}

Out[]=

	age	+	children
				age	+
Deb	62	•	Hal	29	•
			Kat	31	•
+	•	•	•
Bob	41	•	Bob	1	•
			Bri	3	•
			3 total ›
Ann	35	•	Amy	6	•
Cal	60	•

To hide all items by default and unhide individual items, use All to hide everything and then make exceptions using pathFalse:

In[]:=

Dataset

people data

,HiddenItems{All,{"Bob"}False}

Out[]=

	age	sex	children
				age	sex
+	•	•	+	•	•
			+	•	•
+	•	•	•
Bob	41	male	Bob	1	male
			Bri	3	female
			3 total ›
+	•	•	+	•	•
+	•	•	•

Make exceptions to the exceptions to hide unhidden items using pathTrue:

In[]:=

Dataset

people data

,HiddenItems{All,{"Bob"}False,"sex"True}

Out[]=

	age	+	children
				age	+
+	•	•	+	•	•
			+	•	•
+	•	•	•
Bob	41	•	Bob	1	•
			Bri	3	•
			3 total ›
+	•	•	+	•	•
+	•	•	•

MaxItems Pre-12.1, the only control you had over how many Dataset items were displayed was via Dataset`$DatasetTargetRowCount. In 12.1, the

MaxItems

option gives you control over the number of rows displayed as well as columns and deeper levels. To limit the number of rows displayed to 3, specify MaxItems3:

In[]:=

Dataset

planets data

,MaxItems3

Out[]=

	Radius	Moons
			Mass	Radius
Mercury	2439.7km
Venus	6051.9km
Earth	6367.4447 km	Moon	7.3459× 22 10 kg	1737.5km
rows 1–3 of 8

Give a list to specify limits at multiple levels (rows, columns):

In[]:=

Dataset

planets data

,MaxItems{3,1}

Out[]=

	Radius
Mercury	2439.7km
Venus	6051.9km
Earth	6367.4447km
rows 1–3 of 8 columns 1–1 of 2

You can specify limits at any depth. Here, the number of each planet’s moons displayed is limited to 1:

In[]:=

Dataset

planets data

,MaxItems->{Automatic,Automatic,1}

Out[]=

	Radius	Moons
			Mass	Radius
Mercury	2439.7km
Venus	6051.9km
Earth	6367.4447km	Moon	7.3459× 22 10 kg	1737.5km
Mars	3386.km	Phobos	1.072× 16 10 kg	11.1km
		2 total ›
Jupiter	69173.km	Metis	1.× 17 10 kg	21.5km
		63 total ›
Saturn	57316.km	Tarqeq	—	3.5km
		61 total ›
Uranus	25266.km	Cordelia	4.5× 16 10 kg	20.1km
		27 total ›
Neptune	24553.km	Naiad	1.9× 17 10 kg	33.km
		13 total ›

DatasetDisplayPanel When you click a Dataset header, you drill down to that level in the dataset:

Specify the initial drill-down position directly with DatasetDisplayPanel, giving the path to drill down to:

In[]:=

Dataset

planets data

,DatasetDisplayPanel{"Earth"}

Out[]=

Radius	6367.4447km
Moons	 … 1 

ScrollPosition When a Dataset has scrollbars, you can specify the initial scroll positions with

ScrollPosition

, giving the initial vertical and horizontal positions:

In[]:=

Dataset

planets data

,MaxItems{3,1},ScrollPosition{2,2}

Out[]=

	Moons
		Mass	Radius
Venus
Earth	Moon	7.3459× 22 10 kg	1737.5km
Mars	Phobos	1.072× 16 10 kg	11.1km
Mars	Deimos	1.5× 15 10 kg	6.2km
rows 2–4 of 8 columns 2–2 of 2

Deep Dive into Options Syntax Dataset’s styling options have a rich syntax that supports patterns, cyclic specifications and value functions. To show you how those work, I’ll take a deep dive into

Background

syntax. Other styling options work similarly.

To apply the same Background color to all items in a Dataset, specify a single color:

In[]:=

Dataset

people data

,BackgroundYellow

Out[]=

	age	sex	children
				age	sex
Deb	62	female	Hal	29	male
			Kat	31	female
Eva	43	female
Bob	41	male	Bob	1	male
			Bri	3	female
			3 total ›
Ann	35	female	Amy	6	female
Cal	60	female

To specify different colors for successive levels of a Dataset, give a list:

In[]:=

Dataset

people data

,Background{Yellow,Cyan}

Out[]=

	age	sex	children
				age	sex
Deb	62	female	Hal	29	male
			Kat	31	female
Eva	43	female
Bob	41	male	Bob	1	male
			Bri	3	female
			3 total ›
Ann	35	female	Amy	6	female
Cal	60	female

But wait, that colored everything green! That’s because the yellow rows and cyan columns blend to give green items. You can see what’s going on more clearly in the next example.

Giving a list at a given level applies the colors to successive elements. In this case, the first row is yellow, the second is cyan and the rest are the default color:

In[]:=

Dataset

people data

,Background{{Yellow,Cyan}}

Out[]=

	age	sex	children
				age	sex
Deb	62	female	Hal	29	male
			Kat	31	female
Eva	43	female
Bob	41	male	Bob	1	male
			Bri	3	female
			3 total ›
Ann	35	female	Amy	6	female
Cal	60	female

If you color the columns similarly, the colors blend at their intersections. Thus the {"Eva","age"} and {"Deb","sex"} items are green, the blend of yellow and cyan:

In[]:=

Dataset

people data

,Background{{Yellow,Cyan},{Yellow,Cyan}}

Out[]=

	age	sex	children
				age	sex
Deb	62	female	Hal	29	male
			Kat	31	female
Eva	43	female
Bob	41	male	Bob	1	male
			Bri	3	female
			3 total ›
Ann	35	female	Amy	6	female
Cal	60	female

As in Grid, you can specify background colors to be used at the beginning, middle and end at a given level. This example makes the first row red, the second orange, then rows cyclically yellow and white until the last row, which is again red:

In[]:=

Dataset[IdentityMatrix[8],Background{{Red,Orange,{Yellow,White},Red}}]

Out[]=

1	0	0	0	0	0	0	0
0	1	0	0	0	0	0	0
0	0	1	0	0	0	0	0
0	0	0	1	0	0	0	0
0	0	0	0	1	0	0	0
0	0	0	0	0	1	0	0
0	0	0	0	0	0	1	0
0	0	0	0	0	0	0	1

Background colors blend (as they do in Grid) in order to support this kind of styling, which makes it easier to follow long rows and columns:

In[]:=

Dataset[IdentityMatrix[8],Background{{{LightBlue,White}},{{LightGreen,White}}}]

Out[]=

1	0	0	0	0	0	0	0
0	1	0	0	0	0	0	0
0	0	1	0	0	0	0	0
0	0	0	1	0	0	0	0
0	0	0	0	1	0	0	0
0	0	0	0	0	1	0	0
0	0	0	0	0	0	1	0
0	0	0	0	0	0	0	1

In options other than Background, values do not blend. Instead, later values override earlier ones. And within a Background option value, colors only blend when they are part of the same specification. In this example, the column colors override the row colors, except where the column color is None, which lets the row color show through:

In[]:=

Dataset[IdentityMatrix[8],Background{{All}->{{{LightBlue,White}}},{All,All}{None,{{LightGreen,None}}}}]

Out[]=

1	0	0	0	0	0	0	0
0	1	0	0	0	0	0	0
0	0	1	0	0	0	0	0
0	0	0	1	0	0	0	0
0	0	0	0	1	0	0	0
0	0	0	0	0	1	0	0
0	0	0	0	0	0	1	0
0	0	0	0	0	0	0	1

You can specify values at arbitrary levels. To use the default coloring at a given level, specify Automatic. In this example, items in the “children” column, which are at the third level of the Dataset, are colored yellow and orange, while items at higher levels have default coloring:

In[]:=

Dataset

people data

,Background{Automatic,Automatic,{Yellow,Orange}}

Out[]=

	age	sex	children
				age	sex
Deb	62	female	Hal	29	male
			Kat	31	female
Eva	43	female
Bob	41	male	Bob	1	male
			Bri	3	female
			3 total ›
Ann	35	female	Amy	6	female
Cal	60	female

When you hover over a Dataset element, you’ll see its path displayed below the dataset frame. To apply a background color to that element, specify that path on the left-hand side of a rule in the Background value:

In[]:=

Dataset

people data

,Background{{All,"sex"}Cyan}

Out[]=

	age	sex	children
				age	sex
Deb	62	female	Hal	29	male
			Kat	31	female
Eva	43	female
Bob	41	male	Bob	1	male
			Bri	3	female
			3 total ›
Ann	35	female	Amy	6	female
Cal	60	female

If you give a non-list element instead of a path on the left-hand side of a rule, the value is applied to any path that contains that element:

In[]:=

Dataset

people data

,Background{"sex"Cyan}

Out[]=

	age	sex	children
				age	sex
Deb	62	female	Hal	29	male
			Kat	31	female
Eva	43	female
Bob	41	male	Bob	1	male
			Bri	3	female
			3 total ›
Ann	35	female	Amy	6	female
Cal	60	female

Combine level syntax and path syntax to specify a general rule and exceptions, as here where all rows are colored yellow, with the exception of the “Eva” row, which is colored cyan:

In[]:=

Dataset

people data

,Background{Yellow,{"Eva"}Cyan}

Out[]=

	age	sex	children
				age	sex
Deb	62	female	Hal	29	male
			Kat	31	female
Eva	43	female
Bob	41	male	Bob	1	male
			Bri	3	female
			3 total ›
Ann	35	female	Amy	6	female
Cal	60	female

Element paths can contain arbitrary patterns. Here, both the “Eva” and “Ann” rows are colored cyan:

In[]:=

Dataset

people data

,Background{{"Eva"|"Ann"}Cyan}

Out[]=

	age	sex	children
				age	sex
Deb	62	female	Hal	29	male
			Kat	31	female
Eva	43	female
Bob	41	male	Bob	1	male
			Bri	3	female
			3 total ›
Ann	35	female	Amy	6	female
Cal	60	female

Patterns can be arbitrarily complex. This colors any row cyan whose header contains a lowercase or uppercase a:

In[]:=

Dataset

people data

,Background{{_?(!StringFreeQ[#,"a"|"A"]&)}Cyan}

Out[]=

	age	sex	children
				age	sex
Deb	62	female	Hal	29	male
			Kat	31	female
Eva	43	female
Bob	41	male	Bob	1	male
			Bri	3	female
			3 total ›
Ann	35	female	Amy	6	female
Cal	60	female

The restriction imposed by a path is applied after coloring is applied to the Dataset as a whole. Compare these examples. In the first, top-level rows are colored yellow, white and cyan:

In[]:=

Dataset

example data

,Background{{Yellow,White,Cyan}}

Out[]=

	1	2	3
a	1	2	x	a	1
				b	2
				c	3
b	1	2	y	a	1
				b	2
				c	3
c	1	2	z	a	1
				b	2
				c	3

Adding a path specification restricts the coloring to the “3” column:

In[]:=

Dataset

example data

,Background{{All,"3"}{{Yellow,White,Cyan}}}

Out[]=

	1	2	3
a	1	2	x	a	1
				b	2
				c	3
b	1	2	y	a	1
				b	2
				c	3
c	1	2	z	a	1
				b	2
				c	3

To apply the yellow-white-cyan coloring to the individual rows in the {All,"3"} column, specify the coloring at the level of those items, the fourth:

In[]:=

Dataset

example data

,Background{{All,"3"}{None,None,None,{Yellow,White,Cyan}}}

Out[]=

	1	2	3
a	1	2	x	a	1
				b	2
				c	3
b	1	2	y	a	1
				b	2
				c	3
c	1	2	z	a	1
				b	2
				c	3

Since nothing outside of the “3” column is colored in the previous example, the path restriction is redundant. This is another way of specifying the same thing:

In[]:=

Dataset

example data

,Background{None,None,None,{Yellow,White,Cyan}}

Out[]=

	1	2	3
a	1	2	x	a	1
				b	2
				c	3
b	1	2	y	a	1
				b	2
				c	3
c	1	2	z	a	1
				b	2
				c	3

The value of any specification within a styling option can be a function that returns a value. That gives you a useful way of highlighting patterns in data. Here, for example, are the first 100 positive integers, with prime numbers highlighted yellow:

In[]:=

Dataset[Range[100],Background(If[PrimeQ[#],Yellow,White]&)]

Out[]=

1	2	3	4	5	6	7	8	9	10
11	12	13	14	15	16	17	18	19	20
21	22	23	24	25	26	27	28	29	30
31	32	33	34	35	36	37	38	39	40
41	42	43	44	45	46	47	48	49	50
51	52	53	54	55	56	57	58	59	60
61	62	63	64	65	66	67	68	69	70
71	72	73	74	75	76	77	78	79	80
81	82	83	84	85	86	87	88	89	90
91	92	93	94	95	96	97	98	99	100

The arguments of a value function are the value of the item or header, its path within the dataset and the entire dataset itself. Having the dataset available as an argument makes it possible to do local styling based on global properties, as in this example, where rows are colored according to sex. The color of each item is obtained by looking at the value of the “sex” entry in the row that contains the item:

In[]:=

Dataset[ExampleData[{"Dataset","Titanic"}],Background(If[#3[#2[[1]],"sex"]==="male",LightBlue,LightRed]&)]

Out[]=

class	age	sex	survived
1st	29	female	True
1st	1	male	True
1st	2	female	False
1st	30	male	False
1st	25	female	False
1st	48	male	True
1st	63	female	True
1st	39	male	False
1st	53	female	True
1st	71	male	False
1st	47	male	False
1st	18	female	True
1st	24	female	True
1st	26	female	True
1st	80	male	True
1st	—	male	False
1st	24	male	False
1st	50	female	True
1st	32	female	True
1st	36	male	False
rows 1–20 of 1309

Putting It All Together The new Dataset options are intended to help you gain insight into your data and present it effectively. Next are some examples of how you might use them to do so.

This is a sample of the built-in Titanic dataset:

In[]:=

Titanic sample

Out[]=

class	age	sex	survived
1st	47	male	False
3rd	32	male	False
1st	54	female	True
2nd	24	male	False
2nd	29	male	False
1st	55	male	False
1st	24	female	True
1st	25	male	True

Styling with

ItemDisplayFunction

and backgrounds makes the data more immediately comprehensible:

In[]:=

Dataset

Titanic sample

,ItemDisplayFunction"class"(StringTake[#,1]&),"age"TooltipStyleSpacer[{2#,20}],Background

,#&,"sex"->(If[#==="male",♂,♀]&),"survived"->(If[#,"✓",""]&),BackgroundSwitch#3[[#2[[1]],"class"]],"1st",

,"2nd",

,"3rd",

&

Since styling options don’t affect the contents of datasets, you can use them to present numeric data in whatever formats make sense without compromising the original data:

In[]:=

Dataset

numeric data

,ItemDisplayFunction{"weight"(Quantity[#,"kg"]&),"factor"(NumberForm[#,{2,2}]&),"yield"(PercentForm[#,2]&)},AlignmentCenter,HeaderAlignmentCenter

Out[]=

weight	factor	yield
19.0168 kg	0.90	32%
23.7387 kg	0.15	44%
5.78343 kg	0.19	76%
21.9207 kg	1.00	34%
22.8375 kg	0.88	41%
4.81656 kg	0.60	67%

Use coloring to make it easier to pick out significant values in data. Here, negative numbers are colored red, and the largest and smallest values in each column are highlighted in blue and pink, respectively:

In[]:=

Dataset[Table[RandomReal[{-1,1}],{7},{3}],ItemStyle(If[#<0,Red,Black]&),Background(Switch[#,Max[#3[[All,#2[[2]]]]],LightBlue,Min[#3[[All,#2[[2]]]]],LightRed,_,White]&)]

Out[]=

-0.0493364	-0.325624	-0.302516
-0.624392	0.569209	0.152446
-0.243085	-0.135257	-0.61027
-0.91199	0.723467	-0.084962
-0.935893	0.348672	0.554716
0.136354	-0.0301653	0.116628
0.817145	0.581321	0.126559

Heat maps are particularly easy using a background color function:

In[]:=

Dataset

random data

,Background(Hue[1,#]&)

Out[]=

0.817389	0.11142	0.789526	0.187803	0.241361	0.0657388	0.542247
0.231155	0.396006	0.700474	0.211826	0.748657	0.422851	0.247495
0.977172	0.825163	0.925275	0.578056	0.29287	0.208051	0.580474
0.128821	0.306427	0.712012	0.390582	0.819967	0.325351	0.59326
0.518774	0.169013	0.472565	0.807161	0.0118355	0.316876	0.789804
0.011978	0.391276	0.458902	0.458845	0.727517	0.790566	0.263269
0.297514	0.443887	0.0553108	0.481571	0.738297	0.203011	0.544772

For a more compact presentation, hide the data behind

Tooltip

. Hovering over an item shows its value:

In[]:=

Dataset

random data

,ItemDisplayFunction{Tooltip[" ",#]&},Background(Hue[1,#]&),ItemSize2

Version 12.1 gives Dataset a big boost in functionality, but we’re not done yet. There’s more to come in future versions. If you have specific requests, leave me a note in the comments section.

Cite this as: Christopher Carlson, "New 12.1 Dataset Interactive Controls and Formatting Options" from the Notebook Archive (2020), https://notebookarchive.org/2020-07-amydw1x