Bar plot doesn't accept width of 1.0

This is really puzzling me.

I am plotting two datetime time series with the bar function, with the width set to 1.0. But one of the time series shows gaps between the bars.

Here is a test script and the plot. The files are all attached.

load testBar.mat;

figure;
bar( datetimeSeis, dataSeis, 1.0, 'r' );
hold on;
bar( datetimeOLD, dataOLD, 1.0, 'k' );
xlim( [ datetime(1995,1,1) datetime(2000,1,1) ] );

What am I doing wrong?

Answer 1

[Rik]

As you can see, this behavior is caused by the x-values being dates. I suspect the underlying reason is that your data is not equally spaced, since your data is sampled once per month (and therefore the distance between each sample is not the same number of days (but 28, 30, or 31).

S=load('testBar.mat');
subplot(2,1,1)
bar(S.datetimeSeis,S.dataSeis, 1, 'r' )
subplot(2,1,2)
bar(1:numel(S.dataSeis),S.dataSeis,1,'r')

S.datetimeSeis(1:5)

ans = 1x5 datetime array
   16-Jan-1995   16-Feb-1995   19-Mar-1995   19-Apr-1995   20-May-1995

You can over-compensate:

figure
bar(S.datetimeSeis,S.dataSeis, 1.75, 'r' )

[dormant]

The distance between each sample is exactly 31 days.

>> datetimeSeis(1:10)

ans = 

  1×10 datetime array

   16-Jan-1995   16-Feb-1995   19-Mar-1995   19-Apr-1995   20-May-1995   20-Jun-1995   21-Jul-1995   21-Aug-1995   21-Sep-1995   22-Oct-1995

[dormant]

You are correct. The x values for datetimeOLD are all exactly 31 days apart. The last two x values for datetimeSeis are not 31 days apart.

My fault in my rebinning function.

Thank you very much.

Answer 2

[dpb]

load testBar.mat;

figure;
%bar( datetimeSeis, dataSeis, 1.0, 'r' );
%hold on;
bar( datetimeOLD, dataOLD, 1.0, 'k' );

%xlim( [ datetime(1995,1,1) datetime(2000,1,1) ] );
[datetimeSeis(1)  datetimeSeis(end)]

ans = 1x2 datetime array
   16-Jan-1995   18-Nov-2024

[datetimeOLD(1) datetimeOLD(end)]

ans = 1x2 datetime array
   16-Jan-1995   04-Oct-2020

figure;
%bar( datetimeSeis, dataSeis, 1.0, 'r' );
hold on;
bar( datetimeOLD, dataOLD, 1.0, 'k' );
xlim( [ datetime(1995,1,1) datetime(2000,1,1) ] );

figure;
%bar( datetimeSeis, dataSeis, 1.0, 'r' );
hold on;
hB1=bar( datetimeOLD, dataOLD, 1.0, 'k' );
xlim( [ datetime(1995,1,1) datetime(2000,1,1) ] );
hB2=bar( datetimeSeis, dataSeis, 1.0, 'r' );

It appears to be a fignewton from when the second is put onto the same axes is added with such a wide disparity of dates; the Seis dataset goes all the way to 2024 where OLD only covers 2020; each looks ok individually.

figure;
isOld=(datetimeSeis<=datetimeOLD(end));    % keep only overlapping data
bar(datetimeSeis(isOld), dataSeis(isOld), 1.0, 'r' );
hold on;
bar(datetimeOLD, dataOLD, 1.0, 'k' );
xlim( [ datetime(1995,1,1) datetime(2000,1,1) ] );

AHA! -- That does the trick, indeed!