How to decide interpolation technique?

Hello, I plot the first column and fourth column of this txt file time series data and there are some empty data . how can I know what percentage of the empty data and interpolate the data?

Answer 1

[Walter Roberson]

You might want to choose a different aggregation method such as 'minutely'

format long g
filename = 'https://www.mathworks.com/matlabcentral/answers/uploaded_files/1083480/cbrn2020.txt';
data = readtable(filename);
data{1:5,1}

ans = 5×1

1.0e+00 * 2019.99863013699 2019.99863964992 2019.99864916286 2019.9986586758 2019.99866818874

tt = table2timetable(data, 'RowTimes', datetime(0,0,0) + years(data{:,1}));

ttt = retime(tt(:,4), 'daily', 'sum');
ttt(1:5,:)

ans = 5×1 timetable
       Time              Var4      
    ___________    ________________

    29-Nov-2019    504.559662800325
    30-Nov-2019    771.212892908936
    01-Dec-2019    772.532087984068
    02-Dec-2019    773.057727623874
    03-Dec-2019    763.898137033537

plot(ttt, 'Var4')

[Anak Agung Adhi Dermawan]

Yes, but I need it daily. So we can't interpolate the data?

[Walter Roberson]

The above code already summarizes daily.

You can interpolate if you want; provide the interpolation method instead of 'sum' .

But are you sure it is a good idea to interpolate February results all the way over to July ?

[Anak Agung Adhi Dermawan]

I think I don' need to interpolate the data sir. But, when we plot it, how can we know how many empty data in points/percentage?

[Walter Roberson]

numempty = sum(isnan(ttt.Var4));
percentempty = numepty / height(ttt) * 100; 

[Anak Agung Adhi Dermawan]

thank you very much sir

[Anak Agung Adhi Dermawan]

Sir, I can only plot the timetable data using stackedplot command, how can I plot the Var4 data with different aggregation and resample result with hold on command?

[Walter Roberson]

You can always extract data, such as

ttt.Var4

You can do things like,

aggregations = {'hourly', 'daily', 'monthly'};
for K = 1 : length(aggrevations)
    ttt = retime(tt(:,4), aggrevations{K}, 'sum');
    plot(ttt.Properties.RowTimes, ttt.Var4, 'DisplayName', aggrevations{K});
    hold on
end
hold off
legend show