These functions project network data:

  • to_mode() projects a network onto one of its modes. Two nodes of that mode get a tie when they have ties to the same nodes of another mode. Give the mode as an index or as a name. It also projects a network of three or more modes.

  • to_mode1() projects a two-mode network onto its first mode, for example the rows.

  • to_mode2() projects a two-mode network onto its second mode, for example the columns.

  • to_mobility() projects a two-mode network of people and places onto the places, with one tie for each move of a person.

  • to_linegraph() makes the line graph of a network. Each tie becomes a node. Two of these nodes get a tie when their ties share a node.

  • to_hypergraph() makes a hypergraph, where one tie can connect more than two nodes.

  • to_simplicial() makes a simplicial complex, such as the maximal cliques of a one-mode network.

to_mode1(
  .data,
  similarity = c("count", "jaccard", "rand", "pearson", "yule", "match", "overlap",
    "crossmin", "maxcrossmin", "sqdiff", "covariance", "bonacich", "ochiai", "ochiai2",
    "czekanowski", "sokalsneath", "hamann", "rogerstanimoto", "euclidean", "manhattan",
    "hamming", "cosine", "spearman", "kendall", "ruzicka", "agreement", "disagreement",
    "signed", "activity", "association", "fractional")
)

to_mode2(
  .data,
  similarity = c("count", "jaccard", "rand", "pearson", "yule", "match", "overlap",
    "crossmin", "maxcrossmin", "sqdiff", "covariance", "bonacich", "ochiai", "ochiai2",
    "czekanowski", "sokalsneath", "hamann", "rogerstanimoto", "euclidean", "manhattan",
    "hamming", "cosine", "spearman", "kendall", "ruzicka", "agreement", "disagreement",
    "signed", "activity", "association", "fractional")
)

to_mobility(.data, layer = NULL, weighted = FALSE)

to_mode(
  .data,
  mode = 1,
  similarity = c("count", "jaccard", "rand", "pearson", "yule", "match", "overlap",
    "crossmin", "maxcrossmin", "sqdiff", "covariance", "bonacich", "ochiai", "ochiai2",
    "czekanowski", "sokalsneath", "hamann", "rogerstanimoto", "euclidean", "manhattan",
    "hamming", "cosine", "spearman", "kendall", "ruzicka", "agreement", "disagreement",
    "signed", "activity", "association", "fractional"),
  through = NULL
)

to_linegraph(.data)

to_hypergraph(.data)

to_simplicial(.data)

Arguments

.data

An object of a {manynet}-consistent class:

  • adjacency or incidence matrix from {base} R

  • edgelist data.frame from {base} R or tbl/tbl_df from {tibble}

  • stocnet stocnet, from the {manynet} package

  • igraph igraph, from the {igraph} package

  • network network, from the {network} package

  • tidygraph tbl_graph, from the {tidygraph} package

similarity

How to measure how alike two nodes are. The default is "count". Each measure compares the ties of two nodes to the other mode.

The measures are in groups below. The measures in one group put the pairs of nodes in the same order, and differ only in their scale. So the choice of group is more important than the choice in a group.

Measures for valued and binary data:

  • "count" counts the shared ties, as \(XX'\). For valued data, it adds the products of the tie weights, and a repeated tie counts as many times as it is repeated. For signed data, it adds the products of their absolute values, since the signed rules below read the signs. It is the default, and the fastest.

  • "match" counts the nodes of the other mode where the two nodes have the same value. This includes the nodes that neither of them has a tie to. For binary data, it is "rand" times the number of nodes in the other mode. Use it when the level of each tie is important.

  • "crossmin" adds the smaller of the two weights for each shared node. It shows what two nodes can give together. For binary data, it is the same as "count".

  • "overlap" divides "crossmin" by the smaller of the two nodes' totals. This is the Szymkiewicz-Simpson coefficient. For data with no negative values, it is from 0 to 1. Use it when one node is much more active than the other. For valued data, xUCINET divides the count and not "crossmin", so its values can be more than 1.

  • "ruzicka" divides "crossmin" by the sum of the two nodes' totals minus "crossmin". This is Ruzicka's weighted Jaccard coefficient, from 0 to 1. For binary data, it is the same as "jaccard". Use it when you want "jaccard" but must keep the tie weights.

  • "maxcrossmin" takes the largest of these smaller weights, not their sum. So one strong shared tie is enough. For binary data, it only shows if two nodes share any tie.

  • "sqdiff" adds the squared differences of the weights, as \(d\). It gives \(1/(1+d)\), so that a larger value means more alike. Two nodes with the same ties get 1. To get \(d\) back, calculate \(1/x - 1\).

  • "pearson" is the correlation of the two nodes' ties. "covariance" is the same, but not standardised. Use "covariance" when the variance of the ties is important.

  • "spearman" and "kendall" are rank correlations. Use them when the order of the weights is reliable, but the spacing is not. An example is ordinal ratings.

  • "cosine" is the cosine of the angle between the two nodes' ties. It is like "pearson", but it does not centre the ties first. So two nodes are alike when their ties are proportional. For binary data, it is the same as "ochiai". For a node with no ties, "cosine" gives 0, but "ochiai" divides by zero.

  • "euclidean" and "manhattan" use the distance \(d\) between the two nodes' ties, and give \(1/(1+d)\), as "sqdiff" does. "manhattan" adds the absolute differences. "euclidean" gives more weight to one large difference than to many small differences. To get \(d\) back, calculate \(1/x - 1\).

Measures for binary data only. These measures change valued data to binary data first, and give a warning. For two nodes, \(a\) is the number of nodes in the other mode that both have a tie to. \(b\) and \(c\) are the numbers that only one of them has a tie to. \(d\) is the number that neither has a tie to.

  • Measures that do not use \(d\). Use them when two missing ties tell you nothing, for example in a sparse network with many events. "jaccard" is \(a/(a+b+c)\). "czekanowski" (also Dice or Sorensen) is \(2a/(2a+b+c)\), which gives more weight to \(a\). "sokalsneath" is \(a/(a+2(b+c))\), which gives more weight to \(b\) and \(c\). For valued data, use "ruzicka" in place of "jaccard".

  • Measures that count \(d\) as a match. Use them when a missing tie tells you as much as a tie. "rand" (simple matching) is \((a+d)/(a+b+c+d)\). "hamann" is \(((a+d)-(b+c))/(a+b+c+d)\), from -1 to 1. It is positive when there are more matches than differences. "rogerstanimoto" is \((a+d)/(a+2(b+c)+d)\).

  • Measures from the odds ratio \(ad/bc\). "yule" is Yule's Q, \((ad-bc)/(ad+bc)\). "bonacich" is \(\sqrt{ad}/(\sqrt{ad}+\sqrt{bc})\), which is Yule's Y from 0 to 1. Both give 1 when \(b\) or \(c\) is 0.

  • Geometric measures. "ochiai" is \(a/\sqrt{(a+b)(a+c)}\), the cosine for binary data. "ochiai2" (Sokal and Sneath's fifth measure) is \(ad/\sqrt{(a+b)(a+c)(d+b)(d+c)}\), which also uses \(d\). These two do not put pairs in the same order as any other group, so try them too.

  • "hamming" counts the differences, \(b+c\), as the Hamming distance. It gives \(1/(1+b+c)\). It puts pairs in the same order as "rand".

Measures that change "count" for how active each node is:

  • "activity" divides "count" by the average number of ties of the two nodes. For binary data, it is the same as "czekanowski".

  • "association" divides "count" by the count that chance would give. A value above 1 means that two nodes share more than chance would give. This is the association strength of van Eck and Waltman (2009).

  • "fractional" shares each node of the other mode among its members. For example, a work with \(n\) authors adds \(1/(n-1)\) to each pair of its authors. Each author then gets a total of 1 from each work. A work with one author adds nothing. This is the fractional counting of Perianes-Rodriguez et al. (2016).

Measures that read the sign of each tie: "agreement", "disagreement", and "signed". See the section on weighted and signed networks below.

These six measures count every tie, also when a node has more than one tie to the same node.

layer

For a network of people and places with more than one layer of ties between people and places, the name of the layer that says where each person is.

weighted

Logical. If TRUE, the ties of the moves between two places are summed into one tie, with the number of moves as its weight. The default is FALSE, one tie for each move.

mode

The mode to project the network onto. Give an index, such as 1 or 2, or the name of the mode. For a two-mode network, FALSE is the first mode and TRUE the second. The default is 1, the first mode, for example the rows.

A name is compared with the names of the modes from mode_names(). Set these names with add_info(). The comparison ignores case, plurals, and other words in the name. For example, if the modes are "women" and "social events", then "social events", "events", and "event" all give the second mode. If a name matches more than one mode, the function stops. Then give a name that tells the modes apart, or give an index.

through

The modes to project mode through, as indices or names, as for mode. The default is all the other modes. This makes a difference only for a network of three or more modes. The nodes of all the modes in through then count as one mode. Ties within one mode, and ties to other modes, are not used.

For example, take a network of authors, works, and keywords. to_mode(.data, "authors", through = "works") ties two authors who wrote a work together. to_mode(.data, "works") ties two works that share an author or a keyword. To keep two modes, for example to tie authors to keywords, use to_twomode(.data, through = "works").

Value

An object of the same class as the function was given, modified as explained in the function description, details, or section. Functions that split a network return a list of such objects.

Details

Not all functions have methods available for all object classes. Below are the currently implemented S3 methods:

              data.frame default igraph matrix network stocnet tbl_graph
to_hypergraph                  *      *              *       *         *
to_linegraph                   *      *
to_mode1               *       *      *      *       *       *         *
to_mode2               *       *      *      *       *       *         *
to_simplicial                  *      *              *       *         *

Comparison of two-mode projection methods

CategoryFeaturemanynet::to_mode1()/to_mode2()igraph::bipartite_projection()network/sna manual
InputDedicated functionyesyesno
Accepted input classesigraph, network, tidygraph, matrix, edgelistigraph onlyany (manual extraction)
Detects mode membership frommode node attributetype vertex attributebipartite network attr (positional)
ProjectionReturns both projections at onceno — one per callyes — list of twotwo manual calls
Projects mode 1 (actors)to_mode1()which = "true"A %*% t(A)
Projects mode 2 (events)to_mode2()which = "false"t(A) %*% A
WeightsRaw co-membership countsyesyes (multiplicity = TRUE)yes
Binary (unweighted) outputyesyes (multiplicity = FALSE)threshold manually
Jaccard normalisationyesnocode manually
Cosine normalisationyesnocode manually
Other similarity measures31 in all, see similaritynocode manually
AttributesRetains node attributesyesyesno — lost in matrix round-trip
Retains edge attributesweight onlyweight onlyno
Removes self-loops automaticallyyesyesdiag(P) <- 0 manually
OutputOutput class matches inputyesnono
Directed projection supportlimitedno — undirected onlyyes — asymmetric matrix
UsabilityLines of code (basic case)114–6
Lines of code (Jaccard weights)1~8 manual~8 manual
Pipe-friendlyyeswith wrappersno

Weighted and signed networks

Most measures read the weight of a tie as how strong the tie is. The measures for binary data change each weight to 1 first, and give a warning.

Three measures read the sign of a tie instead. The sign shows if a node is for (1) or against (-1) a node of the other mode. For example, in irps_nuclear, each speaker supports or opposes each concept.

  • "agreement" counts the pairs of ties with the same sign. Two speakers who both support a concept agree. Two speakers who both oppose a concept also agree.

  • "disagreement" counts the pairs of ties with opposite signs.

  • "signed" is "agreement" minus "disagreement". The result is a signed network. A negative tie means that two nodes disagree more often than they agree.

If a speaker makes two statements about the same concept, both statements count. To keep agreement and disagreement in one network, join the two networks with from_layers().

The weights can also be on a scale, for example from -2 to 2. Then each pair of ties adds a part of 1 to "agreement", and the rest to "disagreement". Two ties with the same value add 1 to "agreement". Two ties at the two ends of the scale add 1 to "disagreement". In general, "disagreement" adds \(|s_1 - s_2|/r\), where \(r\) is the range of the scale.

To weight a two-mode network of documents and terms before the projection, use to_normalised(rule = "tfidf").

Mobility

to_mobility() projects a two-mode network of people and places onto the places. The people are the first mode, and the places the second mode. The ties between people and places say where each person is. They have begin and end times, or the time of each wave of a panel. A person holds one place at a time. A move is the end of one stay of a person and the begin of the next. The result is a directed network of the places, with their attributes, and one tie from the origin to the destination for each move. The tie has the time of the move and the person who moves.

to_mobility() also takes a network of places with an occupant record, as play_mobility() gives. The person of each tie is then the name of the occupant, and the nodes keep only the attributes of the places. See also as_monan(), which gives the places of the people at two times to {MoNAn}.

Hypergraphs

In a hypergraph, one tie can connect any number of nodes. This tie is a hyperedge. Examples are the people at a meeting, or the authors of a paper. to_hypergraph() returns the class it was given. Nothing is lost.

For a two-mode network, each node of the second mode becomes a hyperedge. The hyperedge connects the nodes of the first mode that have ties to that node. The attributes of the second-mode nodes become attributes of the hyperedges. The attributes of the ties, such as weights, stay with each member. to_twomode() gives the two-mode network back.

For a one-mode network, each tie becomes a hyperedge of its two nodes, with its attributes. If the ties are directed, the hyperedges are directed too. The sender is on the 'from' side, and the receiver on the 'to' side. to_mode() gives the one-mode network back.

A 'stocnet' object keeps each hyperedge as one row of its ties. The with column of that row lists the members of the hyperedge (see the Hypergraphs section of make_stocnet()). Other classes have no hyperedges. They keep each hyperedge as a node of a second mode, with a mark. With this mark, is_hypergraph() finds the hypergraph, and coercion back to a 'stocnet' object gives the hyperedges back. A 'network' object is different, because {network} has hyperedges.

Simplicial complexes

A simplicial complex is a hypergraph where each subset of a hyperedge is also a hyperedge. So its largest hyperedges, the faces, describe all of it. to_simplicial() finds these faces, and marks the result as a simplicial complex.

For a one-mode network, the faces are the maximal cliques, because each subset of a clique is also a clique. The direction, weights, and other attributes of the ties are lost. to_mode() gives the one-mode network that the cliques cover.

For a two-mode network or a hypergraph, the faces are the hyperedges that are not a subset of another hyperedge.

References

On two-mode projection

Borgatti, Stephen P., and Daniel S. Halgin. 2011. "Analyzing affiliation networks". In The SAGE Handbook of Social Network Analysis, 417-433. London: SAGE.

Bonacich, Phillip. 1972. "Technique for analyzing overlapping memberships". Sociological Methodology 4: 176-185. doi:10.2307/270732

van Eck, Nees Jan, and Ludo Waltman. 2009. "How to normalize cooccurrence data? An analysis of some well-known similarity measures". Journal of the American Society for Information Science and Technology 60(8): 1635-1651. doi:10.1002/asi.21075

Perianes-Rodriguez, Antonio, Ludo Waltman, and Nees Jan van Eck. 2016. "Constructing bibliometric networks: A comparison between full and fractional counting". Journal of Informetrics 10(4): 1178-1195. doi:10.1016/j.joi.2016.10.006

See also

to_proximity(), which uses the same measures on a one-mode network. It compares nodes by their ties to each other, not by their ties to another mode.

to_cosine(), which gives the cosine over the columns of a matrix without a projection.

Other modifications: modif_backbone, modif_concepts, modif_direction, modif_from_parts, modif_from_ternary, modif_influence, modif_labels, modif_layers, modif_levels, modif_loops, modif_miss, modif_motifs, modif_paths, modif_permutation, modif_proximity, modif_scope, modif_split, modif_weight

Other projections: modif_concepts

Examples

to_mode1(ison_southern_women)
#> 
#> ── # Projection of Southern Women Data ─────────────────────────────────────────
#> # A labelled, weighted, undirected network of 18 nodes and 139 participation
#> ties
#> # Transformed by projection: mode 1 (count)
#> 
#> ── Nodes 
#> # A tibble: 18 × 4
#>   label     Surname    Title active
#>   <chr>     <chr>      <chr> <lgl> 
#> 1 Evelyn    Jefferson  Mrs   TRUE  
#> 2 Laura     Mandeville Miss  TRUE  
#> 3 Theresa   Anderson   Miss  TRUE  
#> 4 Brenda    Rogers     Miss  TRUE  
#> 5 Charlotte McDowd     Miss  TRUE  
#> 6 Frances   Anderson   Miss  TRUE  
#> # ℹ 12 more rows
#> 
#> ── Ties 
#> # A tibble: 139 × 3
#>    from    to weight
#>   <int> <int>  <dbl>
#> 1     1     2      6
#> 2     1     4      6
#> 3     1     3      7
#> 4     1     5      3
#> 5     1     6      4
#> 6     1     7      3
#> # ℹ 133 more rows
#> 
to_mode2(ison_southern_women)
#> ── # Projection of Southern Women Data ─────────────────────────────────────────
#> # A labelled, weighted, undirected network of 14 nodes and 66 participation
#> ties
#> # Transformed by projection: mode 2 (count)
#> 
#> ── Nodes 
#> # A tibble: 14 × 4
#>   label Surname Title active
#>   <chr> <chr>   <chr> <lgl> 
#> 1 E1    6/27    NA    FALSE 
#> 2 E2    3/2     NA    FALSE 
#> 3 E3    4/12    NA    FALSE 
#> 4 E4    9/26    NA    FALSE 
#> # ℹ 10 more rows
#> 
#> ── Changes 
#> # A tibble: 14 × 4
#>   time        node var        value
#>   <date[1d]> <int> <chr>     <list>
#> 1 1936-02-23    11 active TRUE<lgl>
#> 2 1936-02-25     5 active TRUE<lgl>
#> 3 1936-03-02     2 active TRUE<lgl>
#> 4 1936-03-15     7 active TRUE<lgl>
#> # ℹ 10 more rows
#> 
#> ── Ties 
#> # A tibble: 66 × 3
#>    from    to weight
#>   <int> <int>  <dbl>
#> 1     1     2      2
#> 2     1     3      3
#> 3     1     4      2
#> 4     1     5      3
#> # ℹ 62 more rows
#> 
moves <- make_stocnet(info = list(directed = FALSE, update = "replace"),
  nodes = data.frame(label = c("Ann", "Bob", "Home", "Work"),
                     mode = c("people", "people", "places", "places")),
  ties = data.frame(from = c(1, 1, 2, 2), to = c(3, 4, 3, 3),
                    time = c(1, 2, 1, 2)))
to_mobility(moves)
#> # A longitudinal, labelled, directed network of 2 nodes and 1 arcs over 1 waves
#> 
#> ── Nodes 
#> # A tibble: 2 × 1
#>   label
#>   <chr>
#> 1 Home 
#> 2 Work 
#> 
#> ── Ties 
#> # A tibble: 1 × 4
#>    from    to  time person
#>   <int> <int> <dbl> <chr> 
#> 1     1     2     2 Ann   
#> 
to_mode(ison_southern_women, 2)
#> ── # Projection of Southern Women Data ─────────────────────────────────────────
#> # A labelled, weighted, undirected network of 14 nodes and 66 participation
#> ties
#> # Transformed by projection: mode 2 (count)
#> 
#> ── Nodes 
#> # A tibble: 14 × 4
#>   label Surname Title active
#>   <chr> <chr>   <chr> <lgl> 
#> 1 E1    6/27    NA    FALSE 
#> 2 E2    3/2     NA    FALSE 
#> 3 E3    4/12    NA    FALSE 
#> 4 E4    9/26    NA    FALSE 
#> # ℹ 10 more rows
#> 
#> ── Changes 
#> # A tibble: 14 × 4
#>   time        node var        value
#>   <date[1d]> <int> <chr>     <list>
#> 1 1936-02-23    11 active TRUE<lgl>
#> 2 1936-02-25     5 active TRUE<lgl>
#> 3 1936-03-02     2 active TRUE<lgl>
#> 4 1936-03-15     7 active TRUE<lgl>
#> # ℹ 10 more rows
#> 
#> ── Ties 
#> # A tibble: 66 × 3
#>    from    to weight
#>   <int> <int>  <dbl>
#> 1     1     2      2
#> 2     1     3      3
#> 3     1     4      2
#> 4     1     5      3
#> # ℹ 62 more rows
#> 
to_mode(ison_southern_women, "events")
#> ── # Projection of Southern Women Data ─────────────────────────────────────────
#> # A labelled, weighted, undirected network of 14 nodes and 66 participation
#> ties
#> # Transformed by projection: mode 2 (count)
#> 
#> ── Nodes 
#> # A tibble: 14 × 4
#>   label Surname Title active
#>   <chr> <chr>   <chr> <lgl> 
#> 1 E1    6/27    NA    FALSE 
#> 2 E2    3/2     NA    FALSE 
#> 3 E3    4/12    NA    FALSE 
#> 4 E4    9/26    NA    FALSE 
#> # ℹ 10 more rows
#> 
#> ── Changes 
#> # A tibble: 14 × 4
#>   time        node var        value
#>   <date[1d]> <int> <chr>     <list>
#> 1 1936-02-23    11 active TRUE<lgl>
#> 2 1936-02-25     5 active TRUE<lgl>
#> 3 1936-03-02     2 active TRUE<lgl>
#> 4 1936-03-15     7 active TRUE<lgl>
#> # ℹ 10 more rows
#> 
#> ── Ties 
#> # A tibble: 66 × 3
#>    from    to weight
#>   <int> <int>  <dbl>
#> 1     1     2      2
#> 2     1     3      3
#> 3     1     4      2
#> 4     1     5      3
#> # ℹ 62 more rows
#> 
to_mode(irps_nuclear, "speakers", "signed")
#> ── # Projection of German nuclear discourse network ────────────────────────────
#> # A labelled, signed, weighted, undirected network of 337 nodes and 11035 ties
#> # Transformed by projection: mode 1 (signed)
#> 
#> ── Nodes 
#> # A tibble: 337 × 9
#>   label           active politician govt  coalition office org       party power
#>   <chr>           <lgl>  <lgl>      <lgl> <lgl>     <lgl>  <chr>     <chr> <int>
#> 1 VfEW            TRUE   FALSE      FALSE NA        FALSE  VfEW      NA        0
#> 2 Angela Merkel   TRUE   TRUE       TRUE  TRUE      TRUE   CDU       31        2
#> 3 Sunday Times    TRUE   FALSE      FALSE NA        FALSE  Sunday T… NA        0
#> 4 SPD             TRUE   TRUE       FALSE FALSE     FALSE  SPD       32        1
#> 5 Norbert Röttgen TRUE   TRUE       TRUE  TRUE      TRUE   CDU       31        1
#> 6 Torsten Krauel  TRUE   FALSE      FALSE NA        FALSE  Die Welt  NA        0
#> # ℹ 331 more rows
#> 
#> ── Ties 
#> # A tibble: 11,035 × 4
#>    from    to weight  sign
#>   <int> <int>  <dbl> <dbl>
#> 1     1     2     -7    -1
#> 2     1     4     -5    -1
#> 3     1     5     -1    -1
#> 4     1    10     -4    -1
#> 5     1    11     -1    -1
#> 6     1    13     -1    -1
#> # ℹ 11,029 more rows
#> 
to_linegraph(ison_adolescents)
#> ── # Line graph ────────────────────────────────────────────────────────────────
#> # A labelled, undirected network of 10 nodes and 20 ties
#> 
#> ── Nodes 
#> # A tibble: 10 × 1
#>   label     
#>   <chr>     
#> 1 Betty-Sue 
#> 2 Sue-Alice 
#> 3 Alice-Jane
#> 4 Sue-Dale  
#> 5 Alice-Dale
#> 6 Jane-Dale 
#> # ℹ 4 more rows
#> 
#> ── Ties 
#> # A tibble: 20 × 2
#>    from    to
#>   <int> <int>
#> 1     1     2
#> 2     2     3
#> 3     1     4
#> 4     2     4
#> 5     4     5
#> 6     2     5
#> # ℹ 14 more rows
#> 
to_hypergraph(ison_southern_women)
to_simplicial(ison_adolescents)
#> ── # The Adolescent Society ────────────────────────────────────────────────────
#> # A labelled, undirected simplicial complex of 8 adolescents and 6 hyperedges
#> of 2 to 3 nodes
#> # Transformed by projection: maximal cliques as hyperedges
#> 
#> ── Nodes 
#> # A tibble: 8 × 1
#>   label
#>   <chr>
#> 1 Betty
#> 2 Sue  
#> 3 Alice
#> 4 Jane 
#> 5 Dale 
#> 6 Pam  
#> # ℹ 2 more rows
#> 
#> ── Ties 
#> # A tibble: 6 × 4
#>    from    to with      label
#>   <int> <int> <list>    <chr>
#> 1    NA    NA <int [2]> h1   
#> 2    NA    NA <int [2]> h2   
#> 3    NA    NA <int [2]> h3   
#> 4    NA    NA <int [3]> h4   
#> 5    NA    NA <int [3]> h5   
#> 6    NA    NA <int [3]> h6   
#>