MosaicPlot.m

(*
    Mosaic plot for data visualization implementation in Mathematica
    Copyright (C) 2014-2018  Anton Antonov

    This program is free software: you can redistribute it and/or modify
    it under the terms of the GNU General Public License as published by
    the Free Software Foundation, either version 3 of the License, or
    (at your option) any later version.

    This program is distributed in the hope that it will be useful,
    but WITHOUT ANY WARRANTY; without even the implied warranty of
    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
    GNU General Public License for more details.

    You should have received a copy of the GNU General Public License
    along with this program.  If not, see <http://www.gnu.org/licenses/>.

    Written by Anton Antonov,
    antononcube @ gmail . com,
    Windermere, Florida, USA.
*)

(*
    Mathematica is (C) Copyright 1988-2016 Wolfram Research, Inc.

    Protected by copyright law and international treaties.

    Unauthorized reproduction or distribution subject to severe civil
    and criminal penalties.

    Mathematica is a registered trademark of Wolfram Research, Inc.
*)

(* :Title: MosaicPlot *)
(* :Context: MosaicPlot` *)
(* :Author: Anton Antonov *)
(* :Date: 2018-04-29 *) (* Date of the update of the 2014 implementation.*)

(* :Package Version: 1.0 *)
(* :Mathematica Version: 11.3 *)
(*
  This package defines the function MosaicPlot that summarizes the
  conditional probabilities of co-occurrence of the categorical values
  in a Dataset object or a list of records of the same length.
  (The list of records isassumed to be a full array and the columns to
  represent categorical values.) Note, that if a column is numerical
  but has a small number of different values it can be seen as categorical.

  Descriptions of the mosaic plots can be found in books about
  programming and statistics with R. See for example "R in Action" by
  Robert Kabacoff.
  See also the document in "Mosaic plots for data visualization" at
  https://github.com/antononcube/MathematicaForPrediction/tree/master/Documentation .

  OPTIONS:

  MosaicPlot has options for adjusting the gap between the rectangles,
  the style of the labels, the rotation of the labels, and from which
  side to start the rectangle splitting, and the color of the
  rectangles. MosaicPlot also takes all the options of
  Graphics. (Because MosaicPlot is implemented with Graphics).

  The mosaic plot is made within the rectangle {{0,0},{1,1}}. Using
  the options PlotRange and Frame one make a frame that encompasses
  the rotated labels.

  MosaicPlot takes the following options:

  {"ColumnNames" -> None, "ColumnNamesOffset" -> 0.05,
   "ExpandLastColumn" -> False, "FirstAxis" -> "y", "Gap" -> 0.02,
   "GapFactor" -> 0.5, "LabelRotation" -> {{1, 0}, {0, 1}}, "LabelStyle" -> {},
   "Tooltips" -> True, "ZeroProbability" -> 0.001, ColorRules -> Automatic}

  In addition, MosaicPlot takes all the options of Graphics.

  The options are explained below.

  (o) "ExpandLastColumn" -- visualizing categorical columns + a numerical column

      If the last data column is numerical then MosaicPlot can use it
      as pre-computed contingency statistics. This functionality is
      specified with the option "ExpandLastColumn"->True.

      sData = {{"blond", "blue", 3}, {"blond", "brown", 1},
               {"dark", "blue", 1}, {"dark", "brown", 4}};
      MosaicPlot[sData, "ExpandLastColumn" -> True]

  (o) "Gap" and "GapFactor" -- controlling the size of the gap between the rectangles

      The size of the gaps between the rectangles is controlled with
      the options "Gap" and "GapFactor". The value "Gap" specifies the
      size of the gap between the rectangles derived from the first
      column. MosaicPlot splits the data into rectangles
      recursively. In order to derive the gaps for the subsequent data
      column the values of "Gap" and "GapFactor" are multiplied. In
      other words, if MosaicPlot is given the options
      {"Gap"->g,"GapFactor"->f} then the gap between the rectangles
      corresponding to the i-th column have the size is g f^(i-1).

  (o) "LabelRotation" and "LabelStyle" -- contingency values labels

      The labels derived from the distinct values (levels) of each
      column of the data can be rotated and given style options.

      The option "LabelRotation" takes directional specification for
      Text (the fourth argument of Text). The option "LabelStyle"
      takes options and arguments for the function Style.

      MosaicPlot[censusData[[All, {8, 14}]], "LabelRotation" -> {{1, 0}, {1, 1}},
        "LabelStyle" -> {Bold, Red, FontFamily -> "Times"}]

  (o) "ColumnNames" and "ColumnNamesOffset" -- labels for categorical variables

      The names of the data columns (data's variables) are specified
      with the option "ColumnNames". (The list of names given to
      "ColumnNames" can be formatted with Style.) The distance of the
      column names from the rectangles is specified with the option
      "ColumnNamesOffset".

  (o) "FirstAxis" -- start of the rectangle splitting

      The starting axis of the data splitting is specified by "FirstAxis".

      MosaicPlot[censusData[[All, {9, 14}]], "FirstAxis" -> #] & /@ {"x", "y"}

  (o) "Tooltips" -- tooltips with exact contingency statistics

      MosaicPlot has an interactive feature using Tooltip that gives a
      table with the exact co-occurrence (contingency) values when
      hovering with the mouse over the rectangles. The option
      "Tooltips" takes the values True or False.

  (o) Visualizing non-existing contingencies ("ZeroProbability")

      The non-existing contingencies have to be represented in the
      mosaic plot. MosaicPlot uses very thin rectangles for them and
      the size of these rectangles is controlled with the option
      "ZeroProbability".

  (o) Coloring of the rectangles (ColorRules)

      The rectangles can be colored using the option ColorRules which
      specifies how the colors of the rectangles are determined from
      the indices of the data columns.

      More precisely, the values of the option ColorRules should be a
      list of rules, {i1->c1, i2->c2,...}, matching the form

      {(_Integer->(_RGBColor|_GrayLevel))..}.

      The column indices Subscript[i, k] can be negative (-1 meaning the last column).

      If coloring for only one column index is specified the value of
      ColorRules can be of the form

      {_Integer->{(_RGBColor|_GrayLevel)..}}.

      The colors are used with Blend in order to color the rectangles
      according to the order of the unique values of the specified
      data columns.

      The default value for ColorRules is Automatic. When Automatic is
      given to ColorRules, MosaicPlot finds the data column with the
      largest number of unique values and colors them according to
      their order using ColorData[7,"ColorList"].

      The grid of plots below shows mosaic plots of the same data with
      different values for the option ColorRules (given as plot
      labels).

      sData = Table[{RandomChoice[{1, 4, 5, 2} -> {"a", "b", "c", "d"}],
        RandomChoice[{4, 1, 5} -> {"A", "B", "C"}],
        RandomChoice[{1, 2} -> {"1", "2"}]}, {60}];
      t = MosaicPlot[sData, PlotLabel -> If[TrueQ[# === None], "None", #],
          ColorRules -> ReleaseHold[#], "Gap" -> 0.025, "GapFactor" -> 0.6,
          ImageSize -> 200] & /@ {{}, None,
            Automatic, {_ -> GrayLevel[0.7]},
            HoldForm[{1 -> Green, 2 -> Blue, 3 -> Red}],
            HoldForm[{-2 -> Blue, -1 -> Red}], HoldForm[{2 -> Blue}],
            HoldForm[{2 -> {Pink, Blue}}],
            HoldForm[{2 -> ColorData[11, "ColorList"]}]};
      Grid[ArrayReshape[t, {3, 3}, ""], Dividers -> All]


  TIPS: * When the number of unique values in a categorical column is
  large the gaps between the rectangles might "eat" the recntagles
  areas. Use smaller gap size for the option "Gap".


  TODO
  1. Pearson chi-squared correlation coloring. (After I
  implemented the option ColorRules this TODO item has low priority.)

  2. DONE Make MosaicPlot work with Dataset objects.

*)

(*
  2018-04-29: Updated the package to use the re-implementation of TriesWithFrequencies.m through Association.
*)

BeginPackage["MosaicPlot`"];

MosaicPlot::usage = "MosaicPlot[rarr] makes a mosaic plot that summarizes the conditional probabilities of categorical \
values co-occurrence in a list of records of the same length (a full array or dataset). MosaicPlot has options for \
adjusting the gap between the rectangles, the style of the labels, the rotation of the labels, and from which side to \
start the rectangle splitting. MosaicPlot also takes all the options of Graphics.";

MosaicPlotTooltipTable::usage = "MosaicPlotTriePathTable[triePath:{{catVal_?AtomQ,prob_?NumberQ}..}] makes a table \
of conditional probabilities from a trie path (suitable to be the second argument of Tooltip.)";

Begin["`Private`"];


If[Length[DownValues[TriesWithFrequencies`TrieMerge]] == 0,
  Import["https://raw.githubusercontent.com/antononcube/MathematicaForPrediction/master/TriesWithFrequencies.m"]
];

Clear[TrieUniqueRecords];
TrieUniqueRecords[data_?ArrayQ] :=
    Block[{uniqCVals, zeroRecs, res},
      uniqCVals = Table[Union[data[[All, i]]], {i, Dimensions[data][[2]]}];
      zeroRecs = Flatten[Outer[List, Sequence @@ uniqCVals], Length[uniqCVals] - 1];
      res = TriesWithFrequencies`TrieCreate[zeroRecs];
      Replace[res, x_Association :> Join[x, <|TriesWithFrequencies`$TrieValue -> 0|>], Infinity]
    ];

Clear[TrieAddMissingValues];
TrieAddMissingValues[trie_?TriesWithFrequencies`TrieQ, data_?ArrayQ] :=
    TriesWithFrequencies`TrieMerge[trie, TrieUniqueRecords[data]];

Clear[TrieSortNodes];
TrieSortNodes[trie_?TriesWithFrequencies`TrieQ] :=
    Replace[trie, x_Association :> Sort[x], Infinity];

TrieSortNodes[trie_] :=
    If[Length[trie] == 1, trie,
      Join[{trie[[1]]}, TrieSortNodes /@ SortBy[Rest[trie], #[[1, 1]] &]]
    ];

Clear[TriePruneNumericalLevel];

TriePruneNumericalLevel[trie_?TriesWithFrequencies`TrieQ, pruneLevel_Integer] :=
    Association@TriePruneNumericalLevel[First@Normal@trie, pruneLevel, 1];

TriePruneNumericalLevel[trie_?TriesWithFrequencies`TrieRuleQ, pruneLevel_Integer, level_Integer] :=
    Block[{t},
      With[{$TV = TriesWithFrequencies`$TrieValue},
        Which[
          Length[trie[[2]]] == 1 || pruneLevel < level,
          trie,

          pruneLevel == level && VectorQ[Keys@KeyDrop[trie[[2]], $TV], NumberQ],
          trie[[1]] -> <|$TV -> Total @ Keys @ KeyDrop[trie[[2]], $TV]|>,

          True,
          t =
              Association@
                  KeyValueMap[
                    TriePruneNumericalLevel[#1 -> #2, pruneLevel, level + 1] &,
                    KeyDrop[trie[[2]], $TV]
                  ];
          trie[[1]] -> Join[t, <|$TV -> Total @ Map[#[$TV] &, Values[t]]|>]
        ]
      ]
    ];

Clear[RectanglePartition];
Options[RectanglePartition] = {"Gap" -> 0.01, "ZeroWidth" -> 0.001, "SortNodes" -> False};
RectanglePartition[trie_,
  Rectangle[{x0_?NumberQ, y0_?NumberQ}, {x1_?NumberQ, y1_?NumberQ}],
  axis : ("x" | "y"), opts : OptionsPattern[]] :=

    Block[{ps, aps, xs, ys, gap = OptionValue["Gap"],
      zwidth = OptionValue["ZeroWidth"], sortQ = OptionValue["SortNodes"]},
      If[TrueQ[sortQ],
        ps = #[[1, 2]] & /@ SortBy[Rest[trie], #[[1]] &],
        ps = #[[1, 2]] & /@ Rest[trie]
      ];

      aps = FoldList[Plus, 0, ps /. (0 -> zwidth)];

      If[axis == "x",
        Map[Rectangle[{#[[1]], y0}, {#[[2]], y1}] &,
          MapIndexed[#1 + (#2[[1]] - 1) {gap, gap} &,
            Partition[Rescale[aps, {0, If[aps[[-1]] > 1, aps[[-1]], 1]}, {x0, x1 - gap (Length[ps] - 1)}], 2, 1]]],
        Map[Rectangle[{x0, #[[1]]}, {x1, #[[2]]}] &,
          MapIndexed[#1 + (#2[[1]] - 1) {gap, gap} &,
            Partition[Rescale[aps, {0, If[aps[[-1]] > 1, aps[[-1]], 1]}, {y0, y1 - gap (Length[ps] - 1)}], 2, 1]]]
      ]
    ];

(* The original version of the function TrieMosaicRec gives much better idea of what it does:

Clear[TrieMosaicRec]
TrieMosaicRec[trie_, r_Rectangle, axis : ("x" | "y"), gap_?NumberQ, zwidth_?NumberQ] :=
  Block[{rs},
   If[Length[trie] == 1 || r[[2, 1]] - r[[1, 1]] <= gap || r[[2, 2]] - r[[1, 2]] <= gap, r,
    rs = RectanglePartition[trie, r, axis, "Gap" -> gap, "ZeroWidth" -> zwidth];
    MapThread[TrieMosaicRec[#1, #2, axis /. {"x" -> "y", "y" -> "x"}, gap/2, zwidth] &, {Rest[trie], rs}, 1]
   ]
  ];

*)


Clear[MosaicPlotTooltipTable, MakeTooltipTable];
MosaicPlotTooltipTable[triePath_] := MakeTooltipTable[triePath];
MakeTooltipTable[triePath_] :=
    Block[{t},
      t =
          DeleteCases[
            Join @@ Table[{triePath[[1 ;; i, 1]], triePath[[i + 1 ;; j, 1]],
              Apply[Times, triePath[[i ;; j, 2]]]/triePath[[i, 2]]},
              {j, 2, Length[triePath]}, {i, 1, j - 1}],
            {}, 3];

      t = Map[{If[Length[#[[1]]] == 0, "",
        DisplayForm[
          FormBox[RowBox[
            Riffle[If[StringQ[#], "\"" <> # <> "\"", #] & /@ #[[1]],
              "\[Intersection]"]], TraditionalForm]]],
        DisplayForm[
          FormBox[RowBox[
            Riffle[If[StringQ[#], "\"" <> # <> "\"", #] & /@ #[[2]],
              "\[Intersection]"]], TraditionalForm]], #[[3]]} &, t];

      Grid[Prepend[t, Style[#, Blue, FontFamily -> "Times"] & /@ {"condition", "event", "probability"}], Alignment -> Left,
        Dividers -> {None, {False, True, False}}]
    ];

SIDEChangeRules = {Left -> Top, Top -> Right, Right -> Bottom, Bottom -> Left};
SIDEToCoordinateRules = {Left -> 0, Right -> 1, Top -> 1, Bottom -> 0};

Clear[TrieMosaicRec];
TrieMosaicRec[trie_, triePath_, posPath_, r_Rectangle,
  axis : ("x" | "y"), side : (Top | Bottom | Left | Right),
  gap_?NumberQ, gapFactor_?NumberQ,
  zwidth_?NumberQ, {xLabelRotation_, yLabelRotation_}, labelStyle_,
  addTooltipQ_, colors_, colorInds_] :=

    Block[{rs, t, c = side /. SIDEToCoordinateRules},
      If[Length[trie] == 1 || r[[2, 1]] - r[[1, 1]] <= gap || r[[2, 2]] - r[[1, 2]] <= gap,
        t = If[TrueQ[addTooltipQ], Tooltip[r, MakeTooltipTable[Append[triePath, trie[[1]]]]], r];
        If[Length[colorInds] == 0, t,
          {Which[
            Max[Abs[colorInds]] > Length[posPath], GrayLevel[0.7],
            Length[colorInds] == 1 && Length[posPath] == 1 && ! ListQ[colors[[1]]], colors[[1]],
            Length[colorInds] == 1 && ! ListQ[colors[[1]]], Blend[{White, colors[[1]]}, posPath[[colorInds[[1]]]]],
            Length[colorInds] == 1,
            Blend[colors[[1]], posPath[[colorInds[[1]]]]],
            True, Blend[colors, posPath[[colorInds]]]
          ], t}
        ],
      (*ELSE*)
        rs = RectanglePartition[trie, r, axis, "Gap" -> gap, "ZeroWidth" -> zwidth];
        If[axis == "x", t = Select[rs, #[[1, 2]] == c || #[[2, 2]] == c &];
        If[Length[t] == Length[rs],
          AppendTo[LABELS,
            MapThread[Text[Style[#1, labelStyle], {Mean[{#2[[1, 1]], #2[[2, 1]]}], c}, If[side === Top, -{0, 2}, {0, 2}],
              xLabelRotation] &, {Rest[trie][[All, 1, 1]], rs}]]],
        (*ELSE*)
          t = Select[rs, #[[1, 1]] == c || #[[2, 1]] == c &];
          If[Length[t] == Length[rs],
            AppendTo[LABELS,
              MapThread[Text[Style[#1, labelStyle], {c, Mean[{#2[[1, 2]], #2[[2, 2]]}]}, If[side === Left, -{0, 2}, {0, 2}],
                yLabelRotation] &, {Rest[trie][[All, 1, 1]], rs}]]]
        ];

        MapThread[
          TrieMosaicRec[#1, Append[triePath, trie[[1]]], Append[posPath, #3], #2,
            axis /. {"x" -> "y", "y" -> "x"},
            side /. SIDEChangeRules, gap*gapFactor, gapFactor,
            zwidth, {xLabelRotation, yLabelRotation}, labelStyle,
            addTooltipQ, colors, colorInds] &, {Rest[trie], rs, Range[Length[rs]]/Length[rs]}, 1]
      ]
    ];

MosaicPlot::nargs = "MosaicPlot takes as an argument a full array (that is list of records) or a dataset.";
MosaicPlot::ncno = "The value of the option \"ColumnNamesOffset\" should be a number.";
MosaicPlot::npnum = "The value of the option `1` should be a positive number.";
MosaicPlot::nfax = "The value of the option \"FirstAxis\" should be either \"x\" or \"y\".";
MosaicPlot::nlr = "The value of the option \"LabelRotation\" should be a pair of numbers or two pairs of numbers.";
MosaicPlot::ncr = "The value of the option ColorRules should be a list of rules of the form columnIndex->color.\
 If coloring for only one column index is specified its rule can be of the form colorIndex->{color1,color2,...} .";

Clear[MosaicPlot];
Options[MosaicPlot] =
    Join[{"ColumnNames" -> None, "ColumnNamesOffset"->0.05, "Gap" -> 0.02, "GapFactor" -> 0.5,
      "ZeroProbability" -> 0.001, "FirstAxis" -> "y",
      "LabelRotation" -> {{1, 0}, {0, 1}}, "LabelStyle" -> {},
      "ExpandLastColumn" -> False, "Tooltips"->True, ColorRules -> Automatic}, Options[Graphics]];

MosaicPlot[dataRecords_, opts : OptionsPattern[] ] :=
    Block[{dsetColNames, columnNames},

      columnNames = OptionValue[MosaicPlot, "ColumnNames"];

      dsetColNames = Normal[ Keys@First@dataRecords ];

      If[ TrueQ[columnNames === None || Length[Intersection[ columnNames, dsetColNames]] == 0 ],
        MosaicPlot[ Values@Normal@ dataRecords, Sequence @@ Prepend[ {opts},  "ColumnNames"->dsetColNames] ],
      (* ELSE *)
        dsetColNames = Intersection[ columnNames, dsetColNames];
        MosaicPlot[ Values@Normal@ dataRecords[All, columnNames], Sequence @@ Prepend[ {opts},  "ColumnNames"->dsetColNames] ]
      ]

    ]/;TrueQ[Head[dataRecords]===Dataset];

MosaicPlot[dataRecords_, opts : OptionsPattern[]] :=
    Block[{trie, rs,
      gap = OptionValue[MosaicPlot, "Gap"],
      gapFactor = OptionValue[MosaicPlot, "GapFactor"],
      zwidth = OptionValue[MosaicPlot, "ZeroProbability"],
      firstAxis = OptionValue[MosaicPlot, "FirstAxis"],
      labelRotation = OptionValue[MosaicPlot, "LabelRotation"],
      labelStyle = OptionValue[MosaicPlot, "LabelStyle"],
      columnNames = OptionValue[MosaicPlot, "ColumnNames"],
      frameLabelOffset = OptionValue[MosaicPlot, "ColumnNamesOffset"],
      expandLastColumnQ = TrueQ[OptionValue[MosaicPlot, "ExpandLastColumn"]],
      addTooltipQ = TrueQ[OptionValue[MosaicPlot, "Tooltips"]],
      colorRules = OptionValue[MosaicPlot, ColorRules],
      LABELS = {}, frameLabels, frameLabelCoords, frameLabelRotation,
      colors, colorInds, t, nvals},


      If[! (ArrayQ[dataRecords] && Length[Dimensions[dataRecords]]==2),
        Message[MosaicPlot::nargs];
        Return[{}]
      ];

      If[! TrueQ[ NumberQ[gap] && gap > 0 ],
        Message[MosaicPlot::npnum, "\"Gap\""];
        gap = 0.02;
      ];

      If[! TrueQ[ NumberQ[gapFactor] && gapFactor > 0 ],
        Message[MosaicPlot::npnum, "\"GapFactor\""];
        gapFactor = 0.5;
      ];

      If[! TrueQ[ NumberQ[zwidth] && zwidth > 0 ],
        Message[MosaicPlot::npnum, "\"ZeroProbability\""];
        zwidth = 0.001;
      ];

      If[! (TrueQ[colorRules === None] || TrueQ[colorRules === Automatic] ||
          MatchQ[colorRules, {_Integer -> {(_RGBColor | _GrayLevel) ..}} | {(_Integer -> (_RGBColor | _GrayLevel)) | (Rule[Blank[], (_RGBColor | _GrayLevel)]) ...}]),
        Message[MosaicPlot::"ncr"]
      ];

      Which[
        TrueQ[firstAxis == "x" || firstAxis == "X" || firstAxis == "Top"], firstAxis = "x",
        TrueQ[firstAxis == "y" || firstAxis == "Y" || firstAxis == "Left"], firstAxis = "y",
        True,
        Message[MosaicPlot::nfax];
        firstAxis = "y"
      ];

      If[VectorQ[labelRotation, NumberQ] && Length[labelRotation] == 2, labelRotation = {labelRotation, labelRotation}];
      If[TrueQ[labelRotation === None], labelRotation = {{1, 0}, {1, 0}}];
      If[! (MatrixQ[labelRotation, NumberQ] && Dimensions[labelRotation] == {2, 2}),
        Message[MosaicPlot::nlr];
        labelRotation = {{1, 0}, {0, 1}};
      ];

      If[TrueQ[labelStyle === None], labelStyle = {}];

      If[! NumberQ[frameLabelOffset],
        Message[MosaicPlot::ncno];
        frameLabelOffset = 0.05;
      ];

      If[Length[columnNames] == 0,
        frameLabels = {},
      (*ELSE*)
        frameLabelCoords = {{-frameLabelOffset, 0.5}, {0.5, 1 + frameLabelOffset}, {1 + frameLabelOffset, 0.5}, {0.5, -frameLabelOffset}};
        frameLabelRotation = {{0, 1}, {1, 0}, {0, -1}, {1, 0}};
        If[firstAxis == "x",
          frameLabelCoords = RotateLeft[frameLabelCoords, 1];
          frameLabelRotation = RotateLeft[frameLabelRotation, 1];
        ];
        frameLabels = MapThread[Text[#1, #2, {0, 0}, #3] &,
          {If[Length[columnNames] >= 4, columnNames[[1 ;; 4]], Join[columnNames, Table["", {4 - Length[columnNames]}]]], frameLabelCoords, frameLabelRotation}];
        frameLabels = Select[frameLabels, !TrueQ[#[[1]]==""]&];
      ];

      If[TrueQ[colorRules === None], colorRules = {}];
      If[! TrueQ[colorRules === Automatic],
        colorRules =
            Map[If[NumberQ[#[[1]]] && #[[1]] < 1, (Dimensions[dataRecords][[2]] + #[[1]] + 1) -> #[[2]], #] &, colorRules];
        colors = Map[{#, # /. colorRules} &, Range[1, Dimensions[dataRecords][[2]]]];
        colors = DeleteCases[colors, {_Integer, _Integer}];
        If[Length[colors] == 0, colorInds = {}, {colorInds, colors} = Transpose[colors]]
      ];

      trie = TriesWithFrequencies`TrieCreate[dataRecords];

      If[expandLastColumnQ,
        trie = TriePruneNumericalLevel[trie, Dimensions[dataRecords][[2]]];
        trie = TriesWithFrequencies`TrieNodeProbabilities[trie];
        trie = TrieAddMissingValues[trie, dataRecords[[All, 1 ;; Dimensions[dataRecords][[2]] - 1]]],
      (* ELSE *)
        trie = TriesWithFrequencies`TrieNodeProbabilities[trie];
        trie = TrieAddMissingValues[trie, dataRecords]
      ];

      trie = TriesWithFrequencies`TrieToListTrie[trie];

      (* If the color rules are Automatic we pick the column with the largest number of unique values *)
      If[TrueQ[colorRules === Automatic],
        nvals = {}; t = trie;
        While[Length[Rest[t]] > 0, AppendTo[nvals, Length[Rest[t]]]; t = t[[2]]];
        (*colors={{Lighter[Blue],Lighter[Red]}};*)
        colors = {ColorData[7, "ColorList"]};
        colorInds = Take[Flatten[Position[nvals, Max[nvals]]], 1]
      ];

      trie = TrieSortNodes[trie];
      rs = TrieMosaicRec[trie, {}, {}, Rectangle[{0, 0}, {1, 1}], firstAxis, firstAxis /. {"x" -> Top, "y" -> Left}, gap, gapFactor, zwidth, labelRotation, labelStyle, addTooltipQ, colors, colorInds];

      Graphics[{rs, Black, LABELS, frameLabels},
        DeleteCases[{opts}, ("Gap" | "GapFactor" | "ZeroProbability" | "FirstAxis" | "LabelRotation" | "ExpandLastColumn" | "ColumnNames" | "ColumnNamesOffset" | "Tooltips" | ColorRules) -> _]]

    ];

MosaicPlot[___] := Block[{}, Message[MosaicPlot::nargs]; {}];

End[];

EndPackage[]