heatmap help - how to display grouped data across 3 levels (LO, MED, HI) ?

How can I show 3 levels in a heatmap (LO MED, HI) and not just two (FALSE, TRUE for example) across n=4 GROUPS for a single outcome variable (continuous variable)? And is it possbile to (also) show % of total for the OUTCOME variable by GROUP? and not just mean or median.

Here is some working code and a dataset, but displays only two levels (LO, HI) for each GROUP:

load CHRX.mat % loads as CH = Chelsea; RX = Roxbury
P = prctile(CH.VLFN, 75); 
isVHI = CH.VLFN>=P==1;
VHI = isVHI; % this is the HIGH level
P1 = prctile(CH.VLFN, 25);
isVLO = CH.VLFN<=P1==1;
VLO = isVLO; % thius is the LOW level
isIQR = ~VHI&~VLO==1;
IQR = isIQR; % this is the MEDIUM level
CH1 = addvars(CH, VLO); CH2 = addvars(CH1 ,IQR); CH3 = addvars(CH2, VHI)

CH3 = 40914×26 table
          DateTime          LAeq      VLFN      LFN       MFN       HFN       Day     Season    DayNumber    Wkend    RushHR    Site     Temp     Hum     WD_Deg    ImpactWind      WS      Rain    mbBar     WS_HI     RF      UFPConc    lnUFP      VLO      IQR      VHI 
    ____________________    _____    ______    ______    ______    ______    _____    ______    _________    _____    ______    _____    ____    _____    ______    __________    ______    ____    ______    _____    _____    _______    ______    _____    _____    _____

    15-Apr-2016 12:34:00    54.15      82.3    68.426    57.616    53.474    true     false         6        false    false     false    8.9     34.06      80       railway      16.668     0      1031.3    false    false     22575     10.025    false    false    true 
    15-Apr-2016 12:36:00    53.25    79.223    66.039    55.502    53.072    true     false         6        false    false     false    8.9     34.06      80       railway      16.668     0      1031.3    false    false     14450     9.5784    false    false    true 
    15-Apr-2016 12:39:00     53.7    77.488    67.369    56.525    53.579    true     false         6        false    false     false    8.9     34.06      80       railway      16.668     0      1031.3    false    false     12500     9.4335    false    false    true 
    15-Apr-2016 12:41:00     54.8    70.488    69.756    56.901    54.207    true     false         6        false    false     false    8.9     34.06      80       railway      16.668     0      1031.3    false    false     18820     9.8427    false    true     false
    15-Apr-2016 12:43:00    63.55    72.117    70.894    58.103    63.814    true     false         6        false    false     false    8.9     34.06      80       railway      16.668     0      1031.3    false    false     19750     9.8909    false    false    true 
    15-Apr-2016 12:44:00    53.75    73.194    70.446    56.569    52.921    true     false         6        false    false     false    8.9     34.06      80       railway      16.668     0      1031.3    false    false     22250      10.01    false    false    true 
    15-Apr-2016 12:45:00    52.65    65.916    68.539    54.082     52.55    true     false         6        false    false     false    8.9     34.06      80       railway      16.668     0      1031.3    false    false     27800     10.233    false    true     false
    15-Apr-2016 12:46:00     53.3    71.771    69.697    54.478    52.908    true     false         6        false    false     false    8.9     34.06      80       railway      16.668     0      1031.3    false    false     20400     9.9233    false    false    true 
    15-Apr-2016 12:47:00    54.95     69.95    69.938    56.176    54.107    true     false         6        false    false     false    8.9     34.06      80       railway      16.668     0      1031.3    false    false     11700     9.3673    false    true     false
    15-Apr-2016 12:49:00     53.9    65.075    69.134    54.696    52.966    true     false         6        false    false     false    8.9     34.06      80       railway      16.668     0      1031.3    false    false     16700     9.7232    false    true     false
    15-Apr-2016 12:50:00     53.8    72.045    69.219    55.651    53.241    true     false         6        false    false     false    8.9     34.06      80       railway      16.668     0      1031.3    false    false     13150     9.4842    false    false    true 
    15-Apr-2016 12:51:00    59.85    72.058    70.259    59.153    59.137    true     false         6        false    false     false    8.9     34.06      80       railway      16.668     0      1031.3    false    false     14380     9.5736    false    false    true 
    15-Apr-2016 12:52:00     56.3    70.284    70.176    57.921    55.563    true     false         6        false    false     false    8.9     34.06      80       railway      16.668     0      1031.3    false    false     14100     9.5539    false    true     false
    15-Apr-2016 12:54:00     55.4    65.599    71.151    57.015    54.276    true     false         6        false    false     false    8.9     34.06      80       railway      16.668     0      1031.3    false    false     17300     9.7585    false    true     false
    15-Apr-2016 12:55:00     54.7    63.082    68.384    55.673    54.231    true     false         6        false    false     false    8.9     34.06      80       railway      16.668     0      1031.3    false    false     21600     9.9804    false    true     false
    15-Apr-2016 12:57:00    53.25    74.909    68.499    54.707    53.208    true     false         6        false    false     false    8.9     34.06      80       railway      16.668     0      1031.3    false    false     12350     9.4214    false    false    true 

%% HEATMAPS
% note: ImpactWind is the GROUP categorical variable (n=4)
% note: UFPConc is the OUTCOME, continuous variable or ColorVariable
format shortG
data = CH3; % Chelsea
figure()
h = heatmap(data,'ImpactWind','VHI','ColorVariable','UFPConc')

h = 
  HeatmapChart (Mean of UFPConc) with properties:

      SourceTable: [40914×26 table]
        XVariable: 'ImpactWind'
        YVariable: 'VHI'
    ColorVariable: 'UFPConc'
      ColorMethod: 'mean'

  Use GET to show all properties

h.YDisplayLabels = {'LO','HI'}

h = 
  HeatmapChart (Mean of UFPConc) with properties:

      SourceTable: [40914×26 table]
        XVariable: 'ImpactWind'
        YVariable: 'VHI'
    ColorVariable: 'UFPConc'
      ColorMethod: 'mean'

  Use GET to show all properties

Answer 1

[Voss]

You can make a variable that has three distinct values, say 0 for LO, 1 for MEDIUM, and 2 for HI, which can be constructed as follows:

level = 2*VHI + IQR;

and then add that variable to your table and create the heatmap based on that.

load CHRX.mat % loads as CH = Chelsea; RX = Roxbury
VHI = CH.VLFN >= prctile(CH.VLFN, 75); % this is the HIGH level
VLO = CH.VLFN <= prctile(CH.VLFN, 25); % this is the LOW level
IQR = ~VHI & ~VLO;                     % this is the MEDIUM level
level = VHI*2+IQR;
CH = addvars(CH,level);
%% HEATMAPS
% note: ImpactWind is the GROUP categorical variable (n=4)
% note: UFPConc is the OUTCOME, continuous variable or ColorVariable
format shortG
data = CH; % Chelsea
figure()
h = heatmap(data,'ImpactWind','level','ColorVariable','UFPConc');
h.YDisplayLabels = {'LO','MED','HI'};

[Douglas Leaffer]

That seems to work fine ! Thank you

[Voss]

You're welcome!

Answer 2

[William Rose]

@Douglas Leaffer,

Make a single variable Vstat with three categories: LOW, MED, HIGH, rather than three separate variables.

load CHRX.mat % loads as CH = Chelsea; RX = Roxbury
Phi = prctile(CH.VLFN, 75); 
Plo = prctile(CH.VLFN, 25);
[Vstat{1:height(CH),1}]=deal('MED');  % initialize Vstat
for i=1:height(CH)
    if CH.VLFN(i)<Plo, Vstat{i}='LOW';
    elseif CH.VLFN(i)>Phi, Vstat{i}='HIGH';
    end
end
data=addvars(CH,Vstat);  % Chelsea
%% HEATMAPS
% note: ImpactWind is the GROUP categorical variable (n=4)
% note: UFPConc is the OUTCOME, continuous variable or ColorVariable
format shortG
figure()
h = heatmap(data,'ImpactWind','Vstat','ColorVariable','UFPConc');
h.YDisplayLabels = {'HIGH','MED','LOW'};

There's probably a more elegant way to do it than the for loop I use below, but my way seems to work. Good luck.

[Douglas Leaffer]

Thanks I will check out your solution.