make a contour plot with dates on x axis, time on y axis and fog concentration as z values.

I need to make a contour plot based on the calculated total fog. I am unable to do so as I am new to MATLAB. Hope Someone can help me.For reference I am attaching a sample of my data. Thanks

[dpb]

Not sure how you envision this would work even conceptually...you need a 2D array of observations across dates and times and you've simply got a single time series.

_Maybe if you arrange by day by column and time by row similar as to meshgrid for an X,Y mesh you could fill in what data you do have in the array but I don't think it will make enough to draw a countour but just a trace (I think, I'm having difficulty envisioning what it would be like)

Look at the example at the doc for contour and make a much smaller grid than they do as far as size goes and look at the resulting X,Y, Z data for that to see the kind of data pattern contour expects/needs.

[Walter Roberson]

Your csv file needs some extra care in processing, as you have some lines that have

''

as their only content, looking like those are placeholders for missing data. textscan() can handle those with CommentStyle option.

[dpb]

Will need to use NaN to infill I think to get a regular grid; I imported the file as table to 'spearmint some.

contour may also require using datenum rather than datetime for the date/time axes; it is one of the specialty plotting routines not yet (R2017b) gotten to with aliased version for the new class.

[Toni]

For contour plots, x and y axis ticks labels can be modifed via xticklabels and yticklabels. Reshape in the answer below did not work due to missing data, but code below avoids that by inserting Nan where data is missing.

filename = '1st-10th_dec.csv'
data=readtable(filename)
data.Date = datetime(data.Date,'inputformat','dd-MM-yyyy');
data.Time = duration(data.Time,'inputformat','hh:mm');
date = unique(data.Date);
t = unique(data.Time);

for i=1:length(date)
    for j=1:length(t)
        index = find(data.Date == date(i) & data.Time == t(j));
        if ~isempty(index)
            v(j,i) = data.Fog___(index);
        else
            v(j,i) = NaN;
        end
    end
end

figure
contourf(datenum(date),datenum(t),v)

xticks(datenum(date))
xticklabels(string(date))

t_ticks = [min(t):hours(2):max(t)];
yticks(datenum(t_ticks))
yticklabels(string(t_ticks))

Answer 1

[jonas]

As dbp said, contour is not compatible with datetime format. Here's an example with surf, which is quite similar.

data=readtable('1st-10th_dec.csv')
date=datetime(data{:,1},'inputformat','dd-MM-yyyy');
t=duration(data{:,2},'inputformat','hh:mm');
v=data{:,3};
[~,udate]=findgroups(date);
[~,ut]=findgroups(t);
v=reshape(v,numel(ut),numel(udate))
surf(udate,ut,v)
view([0 90])

and here is the example with contour,

data=readtable('1st-10th_dec.csv')
v=data{:,3};
date=nan(size(data,1),1)
t=nan(size(data,1),1)
date(~isnan(v))=datenum(data{~isnan(v),1},'dd-mm-yyyy')
t(~isnan(v))=datenum(data{~isnan(v),2},'HH:MM')
[~,udate]=findgroups(date);
[~,ut]=findgroups(t);
v=reshape(v,numel(ut),numel(udate))
contour(udate,ut,v)
datetick('x','dd/mm','keeplimits','keepticks')
datetick('y','hh:mm','keeplimits','keepticks')

The resulting plot is quite ugly. This is however easy to fix by interpolation.